WO2008000444A1 - Vorrichtung und verfahren zum erzeugen eines gefilterten aktivitätsmusters, quellentrenner, verfahren zum erzeugen eines bereinigten audiosignals und computerprogramm - Google Patents
Vorrichtung und verfahren zum erzeugen eines gefilterten aktivitätsmusters, quellentrenner, verfahren zum erzeugen eines bereinigten audiosignals und computerprogramm Download PDFInfo
- Publication number
- WO2008000444A1 WO2008000444A1 PCT/EP2007/005649 EP2007005649W WO2008000444A1 WO 2008000444 A1 WO2008000444 A1 WO 2008000444A1 EP 2007005649 W EP2007005649 W EP 2007005649W WO 2008000444 A1 WO2008000444 A1 WO 2008000444A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- activity pattern
- trajectory
- activity
- pattern
- trajectories
- Prior art date
Links
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Processing of the speech or voice signal to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
- G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
- G10L21/0272—Voice signal separating
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61N—ELECTROTHERAPY; MAGNETOTHERAPY; RADIATION THERAPY; ULTRASOUND THERAPY
- A61N1/00—Electrotherapy; Circuits therefor
- A61N1/18—Applying electric currents by contact electrodes
- A61N1/32—Applying electric currents by contact electrodes alternating or intermittent currents
- A61N1/36—Applying electric currents by contact electrodes alternating or intermittent currents for stimulation
- A61N1/36036—Applying electric currents by contact electrodes alternating or intermittent currents for stimulation of the outer, middle or inner ear
- A61N1/36038—Cochlear stimulation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04R—LOUDSPEAKERS, MICROPHONES, GRAMOPHONE PICK-UPS OR LIKE ACOUSTIC ELECTROMECHANICAL TRANSDUCERS; DEAF-AID SETS; PUBLIC ADDRESS SYSTEMS
- H04R2225/00—Details of deaf aids covered by H04R25/00, not provided for in any of its subgroups
- H04R2225/43—Signal processing in hearing aids to enhance the speech intelligibility
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04R—LOUDSPEAKERS, MICROPHONES, GRAMOPHONE PICK-UPS OR LIKE ACOUSTIC ELECTROMECHANICAL TRANSDUCERS; DEAF-AID SETS; PUBLIC ADDRESS SYSTEMS
- H04R2225/00—Details of deaf aids covered by H04R25/00, not provided for in any of its subgroups
- H04R2225/67—Implantable hearing aids or parts thereof not covered by H04R25/606
Definitions
- the present invention generally relates to an apparatus and method for generating a filtered activity pattern, to a source separator, to a method for generating a cleaned-up audio signal and to a computer program, in particular to a noise source filtering concept.
- German Patent Application No. 10 2005 030 327 describes use of a neurophysiological auditory model and generation of signals based thereon.
- German Patent Application No. 10 2005 030 327 describes use of a neurophysiological auditory model and generation of signals based thereon.
- US Pat. No. 6,442,510 Bl describes a concept for determining a transit time differential for signal waveforms for real-time pattern recognition, localization and monitoring of optical and acoustic signals.
- This concept comprises steps of segmental detection and coincidence of signal waveforms for conversion into monotone and continuous trajectories, for real-time pattern recognition, and for localization and monitoring of optical and acoustic signals.
- the method described in said document determines run-time differentials, wherein pre-programmed key signals are detected by signal sampling. Data is corrected from the sampled signals and pairs of signal combinations from given signal Time differentials are determined from the coincidence of the detected signals.
- the device comprises at least two receivers for generating sequences of digital values from incoming acoustic signals.
- the apparatus further comprises vector generators for shaping the digital values into input vectors, a signal detection unit arranged after each vector generator and having parallel, programmable signal flow chains and adder / comparator units.
- the adder / comparator units are arranged perpendicular to the signal flow chains at equidistant intervals.
- the device further comprises a multi-coincidence unit consisting of two anti-parallel shift registers forming flip-flop chains and AND gates.
- the object of the present invention to provide a hearing-adapted concept for producing a filtered representation of an audio signal, so that an influence of interfering sound sources in the filtered representation is reduced.
- a device for generating a filtered activity pattern according to claim 1 a source separator according to claim 50, a method for generating a filtered activity pattern according to claim 51, a method for generating a cleaned-up audio signal according to claim 52, and by a computer program according to claim 53 solved.
- the present invention provides a device for generating a filtered activity pattern according to the patent claim 1.
- a filtered activity pattern is particularly reliable Can be generated based on a first activity pattern on a hearing model of a first ear and a second activity pattern on a hearing model of a second ear by identifying in the first activity pattern and in the second activity pattern trajectories associated with a same sound event by determining whether the two identified trajectories are associated with a sound event of a useful sound source and the first activity pattern or the second activity pattern based on the result of determining whether a trajectory is a sound event of the useful sound source (or a sound event of a noise source ) is assigned.
- trajectories originating from a useful sound source are typically distorted or time-shifted differently in the two ears than trajectories belonging to the interfering sound event. Due to the distortion or displacement of two trajectories, which belong to the same sound event, in the two activity patterns a particularly simple assignment is thus possible, whether the Trajectories are associated with a Nutz-sound event or a Stör- sound event.
- a knowledge as to whether a trajectory is associated with a useful sound event or a disturbing sound event (or a useful sound source or a disturbing sound source) is exploited in the context of the filter according to the invention to the filtered activity pattern from the first activity pattern or from the second activity pattern such that in the filtered activity pattern activity events associated with a sound event of the useful sound source dominate over sound events associated with an interfering sound source or in the filtered activity pattern activity events that are not assigned to a sound event of the useful sound source are no longer present or removed.
- the inventive concept consists essentially in determining whether the trajectories are determined by a comparison or a determination of a distortion or temporal displacement of trajectories in the first activity pattern and in the second activity pattern, which are assigned to a same sound event belong to a useful sound event from a useful sound source or to an interfering sound event from an interfering sound source, and to filter the first activity pattern or the second activity pattern based on said information to obtain the filtered activity pattern.
- a significant advantage of the present invention is thus that when generating a filtered activity pattern of two ears are used.
- all available information from two hearing models for example, a left ear and a right ear
- For the identification of trajectories of useful Sound sources or interfering sound sources can thus relations between related (belonging to the same sound event) trajectories of the first activity pattern and the second activity pattern used. This achieves a particularly efficient and reliable differentiation of trajectories from useful sound sources and interfering sound sources, which moreover imitates a binaral processing of acoustic signals in the human brain.
- the identifier is configured to determine a time shift between the two identified trajectories associated with the same sound event.
- the determiner is designed to determine from the time shift whether two trajectories associated with the same sound event are associated with a sound event of the useful sound source or a sound event of the interfering sound source.
- the described concept according to an exemplary embodiment of the present invention is based on the recognition that a temporal shift between two trajectories is a particularly distinctive feature that allows trajectories originating from a useful sound event or a useful sound source, respectively Trajectories that originate from a noise source or a Stör-Schallereignis to separate.
- the temporal shift between the trajectories is in fact a measure of a spatial position of the sound source.
- a time shift is typically caused by a transit time difference between the sound source and the first ear and between the sound source and the second ear. This delay depends on a location of the sound source relative to the two (spaced apart) ears.
- the present invention further provides a corresponding method for generating a filtered activity pattern according to claim 51.
- the present invention further provides a source separator according to claim 50.
- the two channels of the audio signal are initially separated into activity patterns on an auditory model of a first ear and on a hearing model of a second ear. Based on the activity patterns, trajectories are then recognized in the activity patterns associated with a same sound event. Based on the identification or identification of two traits As has already been explained above, it is determined whether the two trajectories are associated with a sound event of a useful sound source or a sound event of a noise sound source. Thus can be done inventive manner on the basis of activity mu 'art source separation.
- the recognition of the trajectories that belong to useful sound events or interference sound events is achieved in a particularly reliable and efficient manner by the use of two patterns of hearing aid patterns by two ears.
- a subsequent inverse transformation of the filtered activity pattern into a time representation, a frequency representation or a subband representation of the cleaned-up audio signal, which is described by the filtered activity pattern another conventional post-processing of the filtered audio signal (ie the one obtained from the filtered activity pattern by inverse transformation Audio signal).
- the present invention provides a method for generating a cleaned-up audio signal according to claim 52.
- the said concept for generating a cleaned-up activity pattern corresponds in its operation to that of the source separator according to the invention.
- the present invention further provides a computer program according to claim 53.
- FIG. 1 shows a block diagram of an inventive device for generating an activity pattern, according to an embodiment of the present invention
- FIG. 2 shows a block diagram of an inventive device for generating a filtered activity pattern, according to an exemplary embodiment of the present invention
- 3a is a schematic representation of trajectories in an activity pattern describing a number of neurotransmitter vesicles
- FIG. 3b is a schematic representation of trajectories in an activity pattern on a plurality of nerve fibers
- FIG. Fig. 3c is a schematic representation of digitized signals representing the activity patterns of Figs. 3a and 3b;
- Fig. 3d is a schematic representation of a two-dimensional pattern describing the activity patterns of Figs. 3a or 3b and based on the digitized signal of Fig. 3c;
- 4a is a schematic representation of a first activity pattern and a second activity pattern comprising a plurality of trajectories
- FIG. 4b is a schematic representation of a filtered activity pattern generated based on the activity patterns of FIG. 4a;
- FIG. 5a shows a block diagram of a multi-coincidence device according to the invention according to an exemplary embodiment of the present invention
- FIG. 5b shows a detailed circuit diagram of a column of a multi-coincidence device according to the invention according to FIG. 5a.
- Fig. 5c is a circuit diagram of a coincidence cell for use in a multi-coincidence device according to Fig. 5a;
- FIG. 7 shows a block diagram of an identifier according to the invention for identifying trajectories in two activity patterns, according to an embodiment of the present invention
- FIG. 8 is a block diagram of an apparatus for performing pattern recognition based on an activity pattern according to the present invention.
- FIG. 10 shows a circuit diagram of a Hubel-Wiesen network for carrying out a pattern recognition according to the invention
- FIG. 11 shows a schematic representation of a sequence in a simulation of a human ear as well as the intermediate and final results occurring in the simulation;
- FIG. 12 shows a flow chart of a method according to the invention for generating a filtered activity pattern, according to an exemplary embodiment of the present invention
- 13 is a block diagram of a source separator according to the invention, according to an embodiment of the present invention.
- 14 shows a flowchart of a method according to the invention for generating a cleaned-up audio signal, according to an exemplary embodiment of the present invention;
- Fig. 16 is a schematic representation of a human cochlea and selected inner hair cells.
- FIG. 1 shows a block diagram of a device according to the invention for generating a filtered activity pattern based on a first activity pattern on a hearing model of a first ear, and a second activity pattern on a hearing model of a second ear.
- the device according to FIG. 1 is designated 100 in its entirety.
- the device 100 is designed to receive a first activity pattern 110 from a hearing model of a first ear.
- the device 100 is further configured to receive a second activity pattern 112 from a hearing model of a second ear.
- the device 100 includes an identifier 120 configured to receive the first activity pattern 110 and the second activity pattern 112.
- the identifier 120 is configured to detect a first trajectory in the first activity pattern and a second trajectory in a second activity pattern that corresponds to a same sound event (eg, beginning of a vowel, consonant, sound, crack, or other sound event) in a traveling wave on the basilar membrane results) are assigned.
- a sound event is understood to mean, for example, a sound event which leads to trajectories (at least approximately or within a tolerance range) of the same curvature and / or the same length. Preference is given to a same sound event an identical event in an acoustic signal emanating from a sound source understood.
- the identifier 120 is further configured to provide an information 126 describing the first trajectory 122 in the first activity pattern 110 and the second trajectory 124 in the second activity pattern 112, both associated with the same sound event.
- a determiner 130 receives the information 126, and is configured to determine based on the information 126 about the first trajectory 122 and the second trajectory 124 whether the two trajectories 122, 124 associated with the same sound event are a sound event a useful sound source or a sound event of a noise source are assigned.
- the determiner 130 therefore provides information 136 which indicates whether the trajectories 122, 124 are associated with a useful sound event of a useful sound source or an interfering sound event of an interfering sound source.
- the device 100 further comprises a filter 140 for filtering the first activity pattern 110 or the second activity pattern 112 based on a result 136 of determining whether a trajectory is associated with a sound event of the useful sound source or a sound event of the noise source.
- the filter 140 thus receives either the first activity pattern 110, or, alternatively, the second activity pattern 112. However, it is also possible for the filter 140 to receive both the first activity pattern 110 and the second activity pattern 112.
- the filter 140 is designed to generate a filtered activity pattern 146 such that activity events associated with a sound event of the useful sound source dominate in the filtered activity pattern, or that activity events that are not associated with a sound event of the useful sound source , in the filtered activity pattern is no longer present or removed.
- the identifier 120 receives a first activity pattern 110 and a second activity pattern 112.
- the activity patterns 110, 112 are respectively a plurality of parallel signals describing activity in or on auditory cells (preferably inner auditory cells or inner hair cells) of a hearing model
- the first activity pattern 110 includes a plurality of parallel signals describing activity in or on a plurality of auditory cells of a first ear (e.g., a left ear).
- the second activity pattern 112 also typically includes a plurality of parallel information or signals describing activity in or on auditory cells (preferably inner hair cells of the inner hair cells) of a second ear.
- the activity patterns may also describe activity on nerve fibers of auditory nerves.
- the first activity pattern 110 may be an activity (or a time course of an activity) on a plurality of nerve fibers leading to a first ear
- the second activity pattern 112 indicates activity on one
- a plurality of nerve fibers belonging to a second ear e.g., a right ear.
- the first activity pattern 110 and the second activity pattern 112 are obtained by using ear models for the first ear and the second ear.
- the hearing model for the first ear receives a first audio signal from, for example, a first microphone located on a left side of a (human, for example) Head is arranged, and provides the first activity pattern.
- the second activity pattern be detected by a hearing model of a second ear (e.g.
- the second audio signal may for example be supplied by a second microphone, which is arranged, for example, on the right side of a (for example human) head.
- the first activity pattern 110 and the second activity pattern 112 typically describe two audio signals from two different audio signal sources (e.g., from two differently located microphones).
- the two activity patterns can also describe audio signals from at least two channels of a multi-channel audio signal.
- first activity pattern 110 and the second activity pattern 112 are typically formed by a plurality of time signals describing a two-dimensional pattern (for example, in a two-dimensional representation thereof).
- first activity pattern N may contain information or parallel time signals describing an activity in or on N different auditory cells or on N different auditory nerves as a function of time.
- second activity pattern 112 may include a plurality of information or parallel (time) signals describing a time course of activity in or on a plurality of auditory cells or on a plurality of auditory nerves.
- an activity event that is to say, for example, an active state, which is indicated, for example, by a characteristic change in a concentration of a specific substance or an electrical potential
- An activity event may be, for example, a neurotransmitter vesicle occurrence in an auditory cell (for example at a synaptic cleft of the auditory cell) or an occurrence of an (active) action potential on a nerve fiber.
- the identifier 120 is configured to receive the first activity pattern 110 and the second activity pattern 112, and to identify a first trajectory 122 in the first activity pattern 110 into a second trajectory 124 in the second activity pattern 112 associated with a same sound event are. It should be noted in this regard that trajectories associated with an identical sound event typically have an at least similar shape and / or a similar length, and because of the features mentioned can be identified as belonging together or as belonging to the same sound event. Furthermore, it is known that trajectories associated with a same sound event or belonging to a same sound event typically occur within a predetermined maximum time interval.
- the identifier 120 is preferably designed to identify trajectories which are similar to one another, ie trajectories whose curvature deviates from one another by less than a predefined maximum permissible deviation, and which occur within a predefined maximum time interval, as trajectories belonging to a same sound event .
- the identifier 120 may be further configured to take into account a length of trajectories in determining whether two trajectories are associated with a same sound event.
- the identifier 120 may be designed, for example, to indicate that two trajectories belong to a same sound event if they are within a predetermined maximum time interval. tervalls occur, and if they also have, except for a predetermined maximum allowable deviation equal length.
- the identifier 120 Based on recognition that two trajectories 122, 124 are associated with a same sound event, the identifier 120 provides the information 126 that includes information about the two identified, associated trajectories 122, 124.
- the information 126 may, for example, comprise information that makes it possible to find at least one of the identified, associated trajectories 122, 124 (or associated activity events) in the first activity pattern 110 or in the second activity pattern 112.
- the information 126 may include information about a time when the associated trajectories 124 occur. Further, alternatively, the information 126 may include information about a length of the two trajectories 124, 126. Alternatively or additionally, the information 126 further includes a statement as to how strongly the two trajectories 122, 124 differ from one another. The named information 126 may, for example, give a statement as to how strongly the curvatures of the two associated trajectories 124, 126 assigned to the same sound event differ. Furthermore, the information 126 may alternatively or additionally carry information about how great a time shift between the two trajectories 124, 126 is.
- the determiner 130 is preferably configured to receive the information 126 from the identifier 120 and to determine based on whether the two identified, associated trajectories 122, 124 a useful sound event or a useful sound source or a sturgeon sound event or associated with a noise source.
- the The determiner 130 is adapted to derive the information as to whether the two associated identified trajectories are associated with a useful sound source or interfering sound source by comparing characteristics of the two associated identified trajectories 122, 124.
- the identifier 130 may be configured to determine the lengths of two identified as belonging together. Trajectories 122, 124 compare.
- this can be taken as an indication that the two identified trajectories 122, 124 are associated with an interfering sound source, for example assuming that the lengths of trajectories associated with a useful sound source are equal to each other within a given tolerance range.
- the determiner 130 may be configured to compare the curvatures of the two trajectories 122, 124 identified as belonging together to provide a statement based on whether the two trajectories 122, 124 are associated with a useful sound source or interfering sound source.
- the determiner is designed to determine a time shift between the two trajectories 122, 124 identified as belonging together, and to decide whether the trajectories identified as belonging together are associated with a sound result of a useful sound source or a noise source , based on a magnitude of the time shift between the trajectories 122, 124 identified as belonging together.
- the filter 140 is preferably designed to obtain information about which of the trajectories identified by the identifier 120 are associated with a sound event of a useful sound source and / or which of the trajectories identified by the determiner 130 Sound event associated with a noise source.
- the filter 140 is configured to receive the first activity pattern or the second activity pattern, and further to receive information about which of the trajectories included in the first activity pattern or in the second activity pattern are associated with a useful sound source.
- the filter 140 is designed to take the trajectories (or their activity events), which are assigned by the information 136, and which are assigned to a sound event of a useful sound source, into the filtered activity pattern 146, and for example the remaining trajectories (or their activity events) in the first activity pattern 110 and the second activity pattern 112 in the generation of the filtered activity pattern 146 to suppress or at least against the useful sound events of Nutz-Schallttle belonging trajectories and / or activity events to steam.
- the filter 140 is configured to obtain information, as the information 136, which of the trajectories 122, 124 identified by the identifier 120 are associated with a sound event of an interfering sound source.
- the filter 140 is preferably designed to remove or attenuate, from the first activity pattern 110 or the second activity pattern 112, those trajectories (or the activity events associated with the corresponding trajectories) that, according to the information 136, form a sound event Include noise source.
- the filter 140 may be designed, for example, to forward, based on the first activity pattern 110 or on the second activity pattern 112, only those trajectories or sound events which, according to the information 136, are associated with a sound event of the useful sound source.
- the filter 140 may be configured to be out of the first activity pattern 110 or to remove from the second activity pattern 112 those trajectories or activity events which according to the information 136 belong to a sound event of a noise source.
- a filtered activity pattern results overall in which trajectories or activity events that belong to sound events of the interfering sound source are suppressed or damped, and in the activity events or trajectories that belong to sound events of the useful sound source. unattenuated or reinforced are included.
- FIG. 2 shows a block diagram of a device according to the invention for generating a filtered activity pattern based on a first activity pattern on a hearing model of a first ear and a second activity pattern on a hearing model of a second ear.
- the device according to FIG. 2 is designated in its entirety by 200. Since the device 200 according to FIG. 2 is very similar to the device 100 according to FIG. 1, in the devices 100, 200 according to FIGS. 1 and 2, the same features and signals are designated by the same reference symbols and will not be explained again here.
- the identifier 120 in the device 200 is selected to identify a first trajectory 122 in the first activity pattern 110 and a second trajectory 124 in the second activity pattern 112 associated with the same sound event. This identification can, for example, as already described above, take place on the basis of a comparison of the curvatures of the two trajectories 122, 124 and / or the lengths of the two trajectories 122, 124. Otherwise, one of the pattern recognition devices described below can also be used to identify two trajectories 122, 124 which are assigned to a same sound event.
- the identifier 120 of the device 200 is configured to determine a time shift between the two associated trajectories 122, 124 (ie, associated with the same sound event).
- the identifier 120 therefore supplies the information about the time shift between the two identically identified trajectories 122, 124 as the information 126 and as part of the information 126 to the determiner 130.
- the determiner 130 is designed to use the time shift (resp At least based on the time shift) .DELTA.t determine whether the two trajectories 122, 124 are associated with a sound event of a useful sound source.
- the determiner 130 may check to see if the time shift ⁇ t between the two trajectories 122, 124 is within a predetermined allowable range. If the time shift ⁇ t between the two trajectories 122, 124 is within the predefined permissible range, then the determiner 130 can ascertain, for example, that the trajectories belong to a sound event of the useful sound source. If the time shift ⁇ t is outside the predetermined permissible range, the determiner 130 can determine that the trajectories 122, 124 are associated with a sound event of the interfering sound source.
- the permissible range which is defined by assuming that the two trajectories 122, 124 belong to the sound event of a useful sound source when there is a time shift ⁇ t within the permissible range, can for example be fixed. In a preferred embodiment, it is assumed that the allowable range for the time shift of
- Trajectories 122, 124 includes values for the time shift whose amount is smaller than a predetermined upper limit. In other words, the two trajectories become then identified as a sound event of a useful sound source when a time shift ⁇ t between the trajectories is less than a predetermined maximum value.
- both activity patterns 110, 112 indicate that the sound source is in front of the head in a straight line.
- a definition defining a maximum amount of the time difference .DELTA.t to identify two trajectories 122, 124 as belonging to the sound event of the useful sound source therefore corresponds to the definition of an angular range in front of the head of a person in the vicinity of their ears the audio signals are recorded on which the first activity pattern 110 and the second activity pattern 112 are based.
- the allowable range based on characteristics of the signals underlying the activity patterns 110, 112. For this purpose, for example, for a plurality of ranges of time shifts it may be determined when associated trajectories occur for each of the ranges of time shifts. In other words, pairs of all the associated trajectories (having the same curvature or associated with the same sound event) whose time difference with respect to one another lies in a range between ⁇ ti and At 2 are determined, for example.
- an occurrence pattern is then determined that indicates at which point in time trajectories occur with a time shift in the range between ⁇ ti and ⁇ t ⁇ . From the occurrence pattern or a statistic derived from the occurrence pattern, it is then determined, for example, whether the trajectories with time shifts in the range between ⁇ ti and At 2 belong to a speech signal.
- a statistic deviation for example, an average time interval between time instants at which trajectories occur, a standard between time points at which trajectories occur, a maximum time interval between the occurrence of two trajectories, a minimum time interval between the occurrence of two trajectories or another statistical variable (eg an associated standard deviation to one of the above-mentioned statistical values).
- a statistic deviation for example, an average time interval between time instants at which trajectories occur, a standard between time points at which trajectories occur, a maximum time interval between the occurrence of two trajectories, a minimum time interval between the occurrence of two trajectories or another statistical variable (eg an associated standard deviation to one of the above-mentioned statistical values).
- the temporal pattern, with the trajectories, the time shift in the range between ⁇ ti and AT 2 is, can be compared with at least one reference pattern to determine whether the trajectories with a time shift in the range between ⁇ ti and At 2 a useful signal o- describe an interference signal.
- the comparison patterns may comprise, for example, characteristic patterns which characterize the occurrence of trajectories in a typical useful signal (eg a speech signal) and / or in a typical interference signal.
- the determiner 130 in this case comprises an area setting means arranged to determine, for a plurality of intervals of delay times (between two associated trajectories), trajectories having delay times within the respective intervals of a useful sound source or an interfering sound source and to obtain the allowable range as a combination of those time intervals (of time shifts between related trajectories) associated with useful sound sources.
- the range selector is adapted to select the allowable range (ie, the range of time shifts At, such that associated trajectories 122, 124 with a time shift from the allowable range are identified as trajectories associated with sound events of a useful sound source) such that the allowable - Sige area includes one or more (continuous or non-contiguous) time intervals that describe time shifts that belong to trajectories based on sound events from one or more Nutzschall provoken.
- the allowable range ie, the range of time shifts At, such that associated trajectories 122, 124 with a time shift from the allowable range are identified as trajectories associated with sound events of a useful sound source
- the selection of the permissible range of time shifts ⁇ t can also be determined on the basis of other properties of the acoustic signals on which the first activity pattern 110 and / or the second activity pattern 112 are based. For example, it can be determined which trajectories in the first activity patterns belong to a loudest sound signal. The permissible range can then be set so that the trajectories, the to a loudest sound signal, have a time shift ⁇ t that is within the allowable range.
- an audio signal component eg correlation properties, volume, time profile of the intensity, bandwidth, occurrence times of trajectories
- whether the corresponding audio signal component is a useful signal is to be regarded by a useful signal source or as a noise signal from a noise source.
- a dynamic adaptation of the permissible range of time delays ⁇ t is possible, whereby trajectories with a time delay within the permissible range are identified by the determiner 130 as a trajectory belonging to a useful signal from a useful signal source.
- trajectories with a time delay within the permissible range are identified by the determiner 130 as a trajectory belonging to a useful signal from a useful signal source.
- no elaborate processing of the activity patterns 110, 112 (eg, in the form of a continuous determination of correlation properties) is required to detect trajectories of activity events based on the sound event of a payload source, trajectories of activity events, the based on a sound event of a noise source.
- the separation of useful signals and interference signals is maintained as long as a time shift between associated trajectories associated with a sound event from a useful sound source does not change significantly. Only when, for example, the position of the useful sound source changes with respect to the microphones which are used to record the audio signals on which the first activity pattern 110 and the second activity pattern 112 are based, a readjustment of the permissible range is required.
- the following describes how the presence of a first trajectory 122 in the first activity pattern 110 and the presence of a second trajectory 124 in the second activity pattern 112 associated with a same sound event can be detected, and further how the time shift ⁇ t between the two second trajectories 122, 124 can be determined in a technically advantageous manner.
- FIG. 3 a shows a graphic representation of two trajectories in a representation of a neurotransmitter vesicle occurrence in or on an inner auditory cell.
- a first representation 302 shows a chronological progression of a number of neurotransmitter vesicles that occur in or on a first inner auditory cell IHZ1.
- a time axis is denoted by 304, while a number axis 306 describes a number of neurotransmitter vesicles that are released within a time unit or that at a specific time point in or on the considered inner auditory cell (for example, in the released form). exist.
- the first plot 302 shows a first neurotransmitter vesicle occurrence 308 and a second neurotransmitter vesicle occurrence 309.
- the first neurotransmitter vesicle occurrence 308 may be considered as a first activity event
- the second neurotransmitter vesicle occurrence 309 For example, consider a second activity event.
- a second plot 310 shows a neurotransmitter vesicle occurrence at a second inner auditory cell IHZ2.
- a number axis 312 describes a number of neurotransmitter vesicles released per unit of time or at a given time in total (eg, in released form) in or on the inner auditory cell IHZ2.
- a first neurotransmitter vesicle occurrence at the inner auditory cell IHZ2 is indicated at 313, and a second neurotransmitter vesicle occurrence at the inner auditory cell IHZ2 is indicated at 314.
- a third graph 315 similarly shows a neurotransmitter vesicle occurrence at a third inner auditory cell IHZ3, with a first neuron rotransmitter-vesicle occurrence at the inner auditory cell IHZ3 is designated 318, wherein a second neurotransmitter vesicle occurrence is designated 319.
- a fourth graphical representation 320 finally shows a first neurotransmitter vesicle occurrence 323 at a fourth inner auditory cell IHZ4 and a second neurotransmitter vesicle occurrence 324 at the fourth inner auditory cell IHZ4.
- the first activity event 308 at the first inner auditory cell IHZ1, the first activity event 313 at the second inner auditory cell IHZ2, the first activity event 318 at the third inner auditory cell IHZ3 and the first Activity event 323 at the fourth inner auditory cell IHZ4 in the two-dimensional representation of the activity events over time by an approximately straight line 330 are interconnected.
- Line 330 is therefore a first trajectory.
- IHZ4 are connected in the graphical representation 300 of FIG. 3a by a second line 332, which forms a second trajectory.
- a trajectory is generally defined as a line-shaped course that connects two, but preferably more than two, related activity events that are based on the same sound event.
- a trajectory is typically either straight or curved in a single direction, so preferably changes Curvature direction not.
- a trajectory is typically a smooth curve, which thus has no kinks or non-differentiable thresholds.
- the term trajectory also includes the activity events assigned to the trajectory.
- the trajectory 330 includes the activity events 308, 313, 318, 323.
- the trajectory 332 also includes the activity events 309, 314, 319, 324.
- a trajectory preferably includes a plurality of activity events based on propagation of a traveling wave on a basilar membrane of a human inner ear.
- FIG. 3b shows a graphic representation of an activity on a plurality of nerve fibers which are coupled to inner auditory cells of a hearing model.
- the graphical representation of FIG. 3b is designated 340 in its entirety.
- a first plot 342 shows activity of a first nerve fiber coupled, for example, to the first inner auditory cell IHZ1.
- a time axis 344 describes the time while a potential axis 346 describes, for example, a potential on a first nerve fiber NF1.
- plot 342 shows a first activity event 348 that includes triggering an action potential on first nerve fiber NFl.
- a potential profile on the first nerve fiber NF1 has an activity (temporal change or impulse) that forms the activity event 348.
- the first plot 342 also shows a second activity event 350 on the first nerve fiber NFl.
- a second plot 360 depicts an activity of a second nerve fiber NF2 coupled, for example, to a second interior auditory cell IHZ2.
- a third plot 365 shows two activity events 368, 369 on a third nerve fiber NF3.
- a fourth graphical representation 370 shows two activity events 373, 374 on a fourth nerve fiber NF4.
- the activity events 348, 363, 368, 373 are associated with a first Trajektorie 380 are connected or all lie on the trajectory 308. Furthermore, the activity events 350, 364, 369, 374 are connected by the trajectory 382 or are all located on the trajectory 382.
- FIG. 3c shows a graphic representation of time-discretized and value-discretized signals, which are based, for example, on the activity events shown in FIGS. 3a or 3b.
- a plurality of time-discrete and discrete-value signals can be obtained, for example, which include the occurring activity events 308, 313, 318, 323, 309, 314, 319, 324 describe.
- a first signal 392 describes activity events occurring at or in the first inner auditory cell IHZ1.
- a second signal 394 describes activity events occurring at or in a second inner auditory cell IHZ2.
- a third signal 396 describes activity events occurring at or in the third inner auditory cell IHZ3, and a fourth signal 398, for example, describes activity events occurring at or in the fourth inner auditory cell IHZ4.
- the four signals 392, 394, 396, 398 may also describe activity events on the nerve fibers NF1, NF2, NF3, NF4.
- activity events or active states
- the parallel (eg, binary) signals 392, 394, 396, 398 represent only an electrical (eg, binary) representation of the activity events 308, 313, 318, 323, 309, 314, 319, 324, and activity events 348, respectively , 363, 368, 373, 350, 364, 369, 374.
- the fact that activity events lie on trajectories within a two-dimensional representation is not affected by the transition to a representation as parallel signals 392, 394, 396, 398 ,
- FIG. 3 d also shows a graphical representation of a pattern associated with the electrical signals 392, 394, 396, 398 shown in graph 390 of FIG. 3c.
- a time profile of the first signal 392 corresponds to a first column 402 of the pattern shown in the graphical illustration 400.
- a time profile of the second signal 394 corresponds to a second column 404 of the pattern shown in the graphical representation 400.
- a time profile of the third signal 396 is represented by a third column 406 of the pattern 400.
- a time profile of the fourth signal 398 is represented by a fourth column 408 of the pattern 400.
- the pattern 400 according to FIG. 3d represents an occurrence of activity events 308, 309, 313, 314, 318, 319, 323, 324; 348, 350, 363, 364, 368, 369, 373, 374 in or on a plurality of (inner) auditory cells of a hearing model or on a plurality of nerve fibers of the auditory model.
- a conversion of the activity pattern according to FIGS. 3 a, 3 b into a pattern according to FIG. 3 d can be effected, for example, by a temporal, parallel scanning of the signals 392, 394, 396, 398 and by a shifting of samples of said signals by a shifter , It should therefore be pointed out that the activity patterns according to FIGS. 3 a, 3 b are open to pattern recognition methods.
- FIG. 4a An example of a processing of two activity patterns will be described below with reference to FIGS. 4a and 4b.
- the graphical representation of FIG. 4a is designated 400 in its entirety.
- a first graphical representation 410 describes trajectories in an activity pattern on or in a plurality of inner auditory cells of a hearing model or on a plurality of nerve fibers of a hearing model.
- a first axis 412 describes the time, whereas a second axis 414 describes a spatial position of inner auditory cells (on whose activity the activity pattern or the trajectories are based) along a cochlea.
- the second axis 414 shows an index of nerve fiber, assuming that the nerve fibers are coupled to inner auditory cells, and that the index of the nerve fiber is a position of the inner auditory cell to which the nerve fiber is coupled, along the cochlea in a monotone Way describes.
- the graphical representation 400 no longer shows individual activity events for reasons of clarity, but that a plurality of associated activity events (belonging to a single sound event) are described by a trajectory connecting the activity events in the two-dimensional representation.
- the first graphical representation 410 describes an activity pattern on a hearing model of a first ear, wherein it is assumed that the ear model of the first ear is supplied with a first audio signal.
- the second graphical representation 420 shows a corresponding representation of an activity pattern in the form of associated trajectories on a hearing model of a second ear, it being assumed that the hearing model of the second ear is applied with a second audio signal.
- the second graphical representation 420 has an abscissa 422 analogous to the first graphical representation 410, at which time is plotted.
- An ordinate 424 describes a spatial position of inner auditory cells along the cochlea or an index of the nerve fibers, as explained above.
- the first plot 410 shows a first trajectory 430 that is rectilinear or has only a slight, first curvature.
- a second trajectory 432 has a greater curvature than the first trajectory 430, and occurs in time after the trajectory 430.
- a third trajectory 434 has, for example, a third curvature which differs from the first curvature of the first trajectory 430 and the second curvature of the second trajectory 432.
- the third trajectory 434 occurs after the first trajectory 430 and the second trajectory 432 in terms of time.
- the second activity pattern shown in graph 420 has a fourth trajectory 440.
- first trajectory 430 and the fourth trajectory 440 are based on the same sound event of a useful sound source, and thus the first trajectory 430 and the fourth trajectory 440 have equal curvatures or curvatures that are no more than a predetermined one distinguish permissible deviation.
- the fourth trajectory 440 occurs delayed in time by a time delay ⁇ ti with respect to the first trajectory 430.
- Graph 420 also shows a fifth trajectory 442 included in the second activity pattern. It is assumed here that the second trajectory 432 and the fifth trajectory 442 are based on the same sound event of the useful sound source. Therefore, the second trajectory 432 and the fifth trajectory 442 have the same curvature, or the curvatures of the second trajectory 432 and the fifth trajectory 442 differ by less than a predetermined maximum allowable deviation. It should also be noted that the fifth trajectory 442 occurs delayed by the delay time ⁇ t 2 compared to the second trajectory 432.
- the trajectories 430, 432, 440, 442 are all based on sound events of the same useful signal source, it can be assumed that the delay time ⁇ t 2 does not deviate by more than a predefined maximum permissible deviation from the delay time ⁇ ti.
- the second plot 420 further shows a sixth trajectory 444 included in the second activity pattern. It is assumed here that the third trajectory 434 and the sixth trajectory 444 are both based on a same sound event of a noise source. For this reason, it can be assumed that the third trajectory 434 and the sixth trajectory 444 have, for example, a same curvature, or that the curvatures of the third trajectory 434 and the sixth trajectory 444 differ by no more than a predefined maximum permissible deviation. It is further assumed that the interfering sound source is at a different location than the useful sound source. A graphical representation 450 shows this situation by way of example.
- the plot 450 shows a top view of a human head 452. Near a first ear 454 (eg, a left ear), a first microphone 456 is disposed. In the vicinity of a second ear 458 (for example, a right ear), a second microphone 460 is arranged.
- the graphical representation 450 further shows the useful sound source 462 and the noise source 464. It should also be noted that it is assumed that the first graphical representation 410 describes the first activity pattern generated from the audio signal received from the first microphone 456 using a hearing model of the human ear, and further that the second graphical representation 420 shows the second activity pattern based on that from the second microphone 460, by applying an auditory model (eg, human hearing) to the audio signal provided by the second microphone 460.
- an auditory model eg, human hearing
- the fourth trajectory 440 in the second activity pattern is generated later than the first trajectory 430 in the first activity pattern by the delay time ⁇ ti.
- the second delay time ⁇ t ⁇ is equal to the first delay time ⁇ ti.
- a distance between the interfering sound source 464 and the first microphone 456 is shorter by a path difference 472 than a distance between the interfering sound source 464 and the second microphone 460.
- the sixth trajectory 444, the a sound event of the interfering sound source 464 describes to the delay time .DELTA.t 3 later than the third trajectory 434, which describes the same sound event of the interfering sound source 464.
- the delay time ⁇ t 3 is determined by the path difference 472.
- the path difference 472 is greater than the path difference 470.
- the delay time ⁇ t 3 is greater than the delay time ⁇ ti or greater than the delay time ⁇ t 2 .
- the delay times ⁇ ti, ⁇ t 2 , ⁇ t 3 between trajectories 430, 440/432, 442; 434, 444 in the first activity pattern and in the second activity pattern, which belong to the same sound events is dependent on a position of the respective sound source relative to the two microphones 456, 460 (and depending on an angle, under which the corresponding sound sources a connecting line of the two microphones 456, 460 are located).
- the device according to the invention is designed to first identify a first trajectory in the first activity pattern and a second trajectory in the second activity pattern, which are based on the same sound event. This can be done, for example, by identifying trajectories that have the same curvature and / or the same length, and that optionally additionally occur within a predefined maximum time interval. By specifying a maximum time interval, it is taken into account that a time shift between two trajectories belonging to a same sound event in the first activity pattern and in the second activity pattern is limited by a transit time difference between the first microphone 456 and the second microphone 460 ,
- the identifier 120 recognizes that the first trajectory 430 in the first activity pattern and the fourth trajectory 440 in the second activity pattern are identical. have curvature, and therefore assigns the two trajectories 430, 440 to the same sound event.
- the identifier 120 further recognizes that the second trajectory 432 and the fifth trajectory 442 are based on the same sound event, since the two trajectories have an equal curvature and / or a same length. Further, the identifier 120 recognizes that the third trajectory 434 and the sixth trajectory 444 are based on the same sound event.
- the determiner 120 further determines the time delay ⁇ ti between the occurrence of the first trajectory 430 and the fourth trajectory 440, ie between the associated trajectories, which are based on the same sound event. In addition, the determiner also determines the time delay ⁇ t 2 between the occurrence of the second trajectory 432 and the occurrence of the fifth trajectory 442. In addition, the identifier 120 determines the time delay ⁇ t 3 between the occurrence of the third trajectory 434 and the sixth trajectory 444 ,
- the determiner 130 determines, based on the information about the trajectories 430, 432, 434, 440, 442, 444, which of the trajectories 430, 433, 434, 440, 442, 444 are assigned to a sound event of a useful sound source, and which Trajectories are associated with a sound event of the interfering sound source.
- the determiner 130 comprises a device for detecting whether a time shift between two associated trajectories 430, 440; 432, 442; 434, 444 within a predetermined allowable range. It is assumed here that the determiner 130 comprises a stored minimum deceleration value ⁇ t m i n and a stored maximum deceleration value ⁇ t max , the minimum permissible deceleration value ⁇ t m i n and the permissible maximum deceleration value ⁇ t max being added. casual range as an interval [ ⁇ t min; ⁇ t max ] It is assumed here that the following applies:
- the determiner 130 recognizes that the time delay ⁇ ti between the first trajectory 430 and the fourth trajectory 440 is in the allowable range, and further that the time delay ⁇ t 2 between the second trajec tor 432 and the fifth trajectory 442 is in the allowable range , Therefore, the determiner 130 signals through the signal 136 that the trajectories 430, 432, 440, 442 are trajectories associated with a sound event of the useful sound source. The determiner 130 further detects that the time delay .DELTA.t is 3 outside the allowable range, and hence signaled, for example via the signal 136, that the trajectories 434, 444 are associated with a sound event of the interfering sound source.
- the filter 140 generates a filtered activity pattern based on the information 136 provided by the determiner 130.
- An example of the filtered activity pattern is shown in Figure 4b.
- the graph of Fig. 4b in its entirety is designated 480.
- the graphical representation 480 shows in a first graphical representation 482 a first filtered activity pattern based on the first activity pattern according to the graphical representation 410.
- the first filtered activity pattern includes the first trajectory 430 and the second trajectory 432 corresponding to the first activity pattern. However, the first filtered activity pattern does not include the third trajectory 430 that is on a sound event the interfering signal source is based.
- the filter 140 when generating the first filtered activity pattern, for example, the filter 140 removes the trajectory 434 associated with the sound event of the interfering sound source. Alternatively, the filter 140 only captures the trajectories 430 and 432 based on useful sound events of the useful sound source , into the first filtered activity pattern.
- a second plot 484 shows a second filtered activity pattern.
- the second filtered activity pattern 484 contains, according to the second activity pattern, the fourth trajectory 440 and the fifth trajectory 442, but not the sixth trajectory 444.
- the second filtered activity pattern thus contains those trajectories of the first activity pattern, the useful sound events from the useful sound source are assigned, but not trajectories that are associated with noise interference events from the noise source.
- the second filtered activity pattern includes the sound events of the second activity pattern associated with useful sound events of the useful sound source, but not trajectories associated with disturbing sound events of the interfering sound source.
- the first filtered activity pattern includes a plurality of activity events that constitute the respective trajectories.
- each trajectory represents a plurality of individual activity events.
- the first filtered activity pattern describes the information content of the useful sound source perceivable at the location of the first microphone 456.
- the information content of the interfering sound source 464 is not included in the first filtered activity pattern or only in a weakened form.
- the second filtered activity tuschsmuster an information content of the Nutz-sound source 464, which is perceptible at the location of the second microphone 460, not or only in an attenuated form, however, an information content of the interfering sound source 464th
- nerve activity patterns are separately calculated for inner auditory cells with different responsivities.
- an extended identification device receives from a ear model of a first ear a first activity pattern describing activity events in or on inner auditory cells that have a first responsiveness, a second activity pattern, the activity events in or on inner auditory cells that provide a second responsiveness and a third activity pattern describing activity events in or on inner auditory cells having a third responsiveness.
- the first response sensitivity is preferably greater than the second response sensitivity
- the second responsiveness is preferably greater than the third response sensitivity.
- the first activity pattern, the second activity pattern, and the third activity pattern describe activity events on different types of inner auditory cells with different responsivities resulting from excitation of the different auditory cells based on the same audio signal.
- the three activity patterns are from internal auditory cells with a low spontaneous emission rate (LSR), with a mean spontaneous emission rate (MSR) or with a high spontaneous emission rate (HSR).
- LSR low spontaneous emission rate
- MSR mean spontaneous emission rate
- HSR high spontaneous emission rate
- the extended identifier further receives from a second ear auditory model a fourth activity pattern describing activity events in or on inner auditory cells having a fourth responsiveness, a fifth activity pattern describing activity events in or on inner auditory cells having a fifth responsiveness , and a sixth activity pattern describing activity events in or on inner auditory cells having a sixth responsiveness.
- the fourth response is preferably greater than the fifth response
- the fifth response is preferably greater than the sixth response.
- the fourth activity pattern, the fifth activity pattern, and the sixth activity pattern describe activity events on different types of inner auditory cells with different responsivities resulting from excitation of the different auditory cells based on the same audio signal.
- the hearing models of the first ear and of the second ear can optionally be part of the device according to the invention.
- the first activity pattern describes, for example, activity events which result in high sensitivity to internal auditory cells when the auditory cells of the high sensitivity are excited with a first audio signal.
- the second activity pattern describes activity events that occur in or on auditory cells of medium sensitivity when the auditory cells are excited by the first audio signal.
- the third activity pattern also describes an activity pattern or activity events that arise in or on inner auditory cells with a low sensitivity when the inner auditory cell (or an ear model, which speaking inner auditory cells) is excited by the first audio signal.
- the fourth activity pattern describes activity events that result in high sensitivity to inner auditory cells when the corresponding inner auditory cells (or the underlying auditory model) are stimulated by a second audio signal.
- the fifth activity pattern describes activity events in or on inner auditory cells of average sensitivity when the inner auditory cells are excited by the second audio signal, and the sixth activity pattern describes activity events in or on inner auditory cells of low sensitivity when the corresponding inner auditory cells through the second Audio signal to be eligible.
- the first, second and third activity patterns describe activity events on inner auditory cells of different sensitivities of a hearing model of a first ear, which is excited by the first audio signal.
- the fourth, fifth and sixth activity patterns describe activity events on inner auditory cells of different sensitivities of a hearing model of a second ear, which is excited by the second audio signal.
- the identifier in this case is designed to process activity patterns on the ear model of the ear and activity patterns on the ear model of the second ear that are associated with inner auditory cells of a same sensitivity level (high sensitivity, medium sensitivity, low sensitivity).
- the identifier is designed to identify a first trajectory in the first activity pattern and a fourth trajectory in the fourth activity pattern, which are assigned to a same sound event.
- the first activity pattern and the fourth activity pattern are fed to a first detection device which is designed to identify in the first activity pattern and the fourth activity pattern trajectories of the same curvature and / or the same length as associated trajectories.
- the first identification device may be, for example, a multi-coincidence unit (for example according to FIG. 5 a) or a recognition device according to FIG. 7.
- the corresponding first recognition device thus provides information about at what times and with what timely displacement trajectories associated with a same sound event occur in the first activity pattern and the fourth activity pattern.
- a second recognizer receives, for example, the second activity pattern and the fifth activity pattern and determines therein trajectories associated with a same sound event.
- the second recognition device thus provides information about when and / or with which temporal shift relative to one another in the second activity pattern and the fifth activity pattern trajectories are contained, which are assigned to a same sound event.
- a third recognition device also receives the third activity pattern and the sixth activity pattern and identifies therein trajectories that are assigned to the same sound events.
- the third detection means therefore provides information about trajectories in the third activity pattern and the sixth activity pattern associated with the same sound events, as well as information about a time shift between trajectories associated with the same sound events.
- volume levels are distinguished. Thus it is assumed that there are sound events of high volume, sound events of medium volume and low volume sound events. It is believed that a high volume sound event results in a trajectory in both the first activity pattern, the second activity pattern, the third activity pattern, the fourth activity pattern, the fifth activity pattern, and the sixth activity pattern. It is further believed that a medium volume sound event results in a trajectory in the first activity pattern, the second activity pattern, the fourth activity pattern, and the fifth activity pattern, but not trajectories in the third activity pattern and the sixth activity pattern.
- the third activity pattern and the sixth activity pattern are formed by low-sensitivity auditory cells, because it is assumed that a middle-volume sound event is not strong enough to produce the low-sensitivity auditory cells (their response through the third activity pattern) the sixth activity pattern is described).
- a low-volume sound event only generates trajectories in the first activity pattern and in the fourth activity pattern. It is believed that a low volume sound event is insufficient to excite medium sensitivity and low sensitivity auditory cells such that a low volume sound event does not result in a trajectory in the second activity pattern and in the fifth activity pattern (formed by medium sensitivity auditory cells) as well neither does it result in the third activity pattern and the sixth activity pattern (formed by low sensitivity auditory cells).
- it is possible to obtain information about a loudness of a sound event by comparing the trajectories in the first activity pattern, in the second activity pattern and in the third activity pattern.
- a sound event volume recognizer that is part of the identifier or part of the determiner, for example, may be configured to generate information and volume of a sound event underlying a trajectory for a plurality of trajectories.
- the volume recognizer may be configured to detect whether or not in the first activity pattern and in the second activity pattern (or within a predetermined maximum time interval) trajectories of a similar curvature (or trajectories of different curvature whose curvatures are less than a given maximum allowable curvature difference different) present.
- the volume determiner can recognize, for example, that those present in the first activity pattern
- the volume recognizer may alternatively or additionally be designed to detect whether trajectories of at least approximately the same curvature occur at least approximately simultaneously in the second activity pattern and in the third activity pattern the second activity pattern no corresponding approximately simultaneously occurring associated trajectory approximately the same curvature
- the volume recognizer can determine, for example, that the sound event, eg u the curvature in question in the second activity pattern heard, has a medium volume.
- the volume recognizer may, for example, determine that a sound event has a high volume if trajectories of approximately the same curvature occur approximately simultaneously in both the first activity pattern and in the second and third activity patterns.
- such processing can also be carried out for the fourth activity pattern, the fifth activity pattern and the sixth activity pattern. If a trajectory occurs in the fourth activity pattern for which there is no corresponding approximately simultaneous trajectory of approximately the same curvature in the fifth activity pattern, the volume recognizer can determine that the curvature occurring in the fourth activity pattern is associated with a low-volume sound event is. If the volume recognizer further identifies a trajectory in the fifth activity pattern for which there is no associated approximately simultaneous trajectory of approximately equal curvature in the sixth activity pattern, the volume recognizer may determine that the trajectory identified in the fifth activity pattern is a sound event assigned to medium volume.
- the volume recognizer may determine that corresponding trajectories are associated with a high volume sound event.
- the volume recognizer is designed to match that found in the activity patterns being studied Trajectories (or at least part of the trajectories) to assign a volume information.
- the volume recognizer generally analyzes whether trajectories of the same curvature occur approximately simultaneously (i.e., within a predetermined maximum time interval) in the various activity patterns under consideration and, if so, in which of the activity patterns approximately simultaneously trajectories of approximately equal curvature occur. Based on said information or based on a comparison between trajectories occurring in different activity patterns, the inner auditory cells are assigned different sensitivity, the volume recognizer thus determines the volume information associated with the trajectories.
- the filter 140 is configured to account for the volume information in the filtering.
- the filter 140 may be configured to receive the volume information from the volume recognizer.
- the filter 140 may be designed to filter out or remove or dampen all trajectories whose associated volume information is less than a predetermined minimum volume.
- the filter may also be configured to generate the filtered activity pattern such that only trajectories are associated with the filtered activity pattern associated with sound events having a volume within a predetermined level according to the information provided by the volume detector Have area.
- the filter 140 may be configured to generate the filtered activity pattern such that the activity pattern emphasizes trajectories associated with low volume sound events, while, for example, trajectories corresponding to sound events having a larger sound level Volume may be relatively damped or even completely removed.
- the filtered activity pattern only or at least essentially trajectories are contained which belong to sound events of low volume.
- the filtered activity pattern just low-signal components in the original activity patterns are described, whereby a comprehensibility of the quiet portions in the original activity patterns is improved or only possible.
- the device according to the invention in accordance with FIG. 1 or FIG. 2 can be expanded by the device receiving two sets of patterns of activity of hearing models of two ears, each of which has activity events on auditory cells with a high spontaneous amplitude Emission rate, a medium spontaneous emission rate and a low spontaneous emission rate.
- An identification of trajectories belonging to the same sound events then takes place separately for activity patterns of auditory cells of different emission rates.
- an activity pattern of the ear model of the first ear for high emission rate auditory cells is processed together with an activity pattern on the ear model of the second ear for high frequency auditory cells to provide information about corresponding trajectories (based on a same sound event) receive.
- co-processing is performed for two activity patterns belonging to middle-rate auditory cells.
- joint processing is performed for two activity patterns of low emission rate auditory cells.
- both the volume information and the information about the time shift are taken into account.
- the filter in conjunction with the determiner may be configured to obtain the filtered activity pattern such that in the filtered activity pattern trajectories dominate whose associated volume information is within a predetermined range, for example, and their associated time shift information within a predetermined range.
- the filter in conjunction with the determiner may be configured to, for example, generate the filtered activity pattern so that the filtered activity pattern essentially comprises an audio signal originating in a spatially limited area.
- An angular limitation of the range consists in limiting the permissible temporal shift between two associated trajectories.
- a distance limitation arises by limiting the volume to a predetermined range.
- circuitry of Figures 1 and 2 (or at least the identifier 120 thereof) is replicated in triplicate in one embodiment. This results in three parallel branches for three loudness spaces HSR, MSR, LSR.
- the structure is structurally the same, but once activity events or activity patterns of auditory cells (or spiral ganglion cells) with a high spontaneous emission rate, once activity patterns of auditory cells with a mean spontaneous emission rate and once acute activity patterns of auditory cells with a low spontaneous emission rate.
- the circuit e.g., the identifier, the determiner, the multi-coincidence unit 500, and / or the recognizer 700
- the circuit is tri-built for an HSR space, an MSR space, and an LSR space.
- the present invention thus comprises the following aspects:
- Binaural vesicle filtering in which delay or delay trajectories are matched by a time shift
- the present invention provides a method and apparatus according to aspects a), b), and c) in all three HSR, MSR, LSR spaces (wherein a plurality of high-level auditory cells are located below an HSR space Spontaneous emission rate is understood, wherein a MSR space is a plurality of auditory cells with a mean spontaneous emission rate is understood, and wherein a LSR space a plurality of auditory cells with low spontaneous emission rate is understood).
- the system according to the invention is threefold in one embodiment, wherein separate signals or activity patterns from spiral ganglion cells with high spontaneous emission rate, average spontaneous emission rate and low spontaneous emission rate (HSR, MSR, LSR ) will be used. Since the spiral ganglion cells with different spontaneous emission rates have different response level ranges, a vesicle output (or general: see activity pattern) looks different for loud, medium, and quieter tones. Thus, different loud signal sources can be separated.
- FIG. 5a shows a circuit diagram of a multi-coincidence unit according to the invention for use in a device according to the invention, according to an exemplary embodiment of the present invention.
- the circuit arrangement according to FIG. 5a is designated in its entirety by 500.
- the circuit arrangement 500 is designed to receive a first plurality 510 of parallel signals and a second plurality 512 of parallel signals. It is assumed that the first plurality 510 of parallel signals, for example, describes an activity pattern. In other words, time courses of the signals of the first plurality 510 of parallel signals describe a two-dimensional pattern (compare FIGS. 3a to 3d). The same applies to the signals of the second plurality 512 of parallel signals.
- Circuitry 500 comprises a field of coincidence cells.
- An example of a single coincidence cell is shown in Fig. 5c.
- the Fig. 5c shows a circuit diagram of a coincidence cell.
- a coincidence cell 580 has a first data input 582 and a second data input 584.
- the coincidence cell 580 also has a first data output 586 and a second data output 588.
- the coincidence cell 580 includes a first memory device 590 and a second memory device 592.
- the first memory device 590 and the second memory device 592 are preferably configured to receive a common (or separate) clock signal 594 and to be responsive to take on the clock signal 594 information from the input and store.
- the first memory cell 590 is configured to accept, store, and output to the first output 586 information (eg, a binary value) from the first input 582 in response to the clock signal 594.
- the second memory cell 592 is preferably configured to accept, store, and output at the second output 588 information (eg, a binary value) from the second input 584 in response to the clock signal 594.
- the coincidence cell 580 further includes an AND gate configured, for example, to detect when both outputs 586, 588 of the coincidence cell 580 are in an active state.
- the AND gate is designated 596 and provides a coincidence signal 598.
- the multi-coincidence unit 500 consists of a plurality of coincidence cells 580, whose interconnection is described below. It should be noted that the multi-coincidence unit 500 is constructed in the form of a matrix of coincidence cells 580.
- the coincidence cells are referred to below as two indices i, j, where the index i denotes a row, and wherein the index j denotes a column.
- the multi-coincidence unit comprises at least I rows and at least J columns, where I> 3 and J ⁇ 3.
- individual coincidence cells are denoted by Z (i, j).
- a first output of the coincidence cell Z (i, j) is coupled to a first input of an adjacent coincidence cell Z (i, j + 1), where 1 ⁇ j ⁇ J-I.
- the second input of the coincidence cell Z (i, j) is coupled to the second output of the coincidence cell Z (i, j + 1), where 1 ⁇ j ⁇ J-1.
- the coincidence cells of the i-th row are arranged to forward the i-th input signal of the first plurality 510 of parallel input signals stepwise (in response to the timing signal 594) in a first direction (eg, from left to right), and step-by-step forwarding the ith input signal of the second plurality 512 of parallel input signals in a second direction (eg, from right to left).
- the forwarded signals are present.
- the coincidence cells are designed to detect when the forwarded signal of the first plurality 510 of parallel signals at the first output 586 of the respective coincidence cell and the forwarded signal of the second plurality 512 of parallel signals at the second output 588 of the coincidence cell simultaneously are active.
- the coincidence cell in question outputs a coincidence signal at the output 598.
- the multi-coincidence unit 500 further comprises summers 530.
- a j-th summer 530 is configured to receive and sum the coincidence signals 598 of the coincidence cells 580 of the jth column.
- the jth summer 530 is designed to determine coincidence signals 598 occur at the outputs, such as many coincidence cells 580 of the jth column, simultaneously or within a given time interval.
- summer 530 may be configured to receive a reset signal synchronously or asynchronously.
- time intervals may be defined within which a number of coincidences in the j-th column of the multicoincidence unit 500 are summed up in order, for example, to detect a trajectory.
- the multi-coincidence unit 500 further includes threshold decision gates 540 configured to receive an associated sum signal from an associated summer 530.
- the threshold decision 540 are designed to detect when within the time interval during which a summation occurs within a single column j at least a predetermined number of coincidences have occurred in the coincidence cells 580 of the j-th column. In this case, the threshold decision 540 provide an associated threshold signal 542.
- the coincidence unit 500 is configured to determine whether at least a predetermined number of coincidences have occurred within a predetermined time interval in a j-th column while two activity patterns have been pushed in an opposite direction by the multi-coincidence unit 500.
- coincidence is, as defined above, the simultaneous presence of two active signals in a coincidence stage 580 to understand.
- the coincidence unit 500 shifts two activity patterns in opposite directions containing trajectories. So one occurs Coincidence when both an activity event of the first activity pattern and an activity event of the second activity pattern are present in a same coincidence cell. Simplified, coincidence occurs when a first trajectory in the first activity pattern (input to the multi-coincidence unit 500, for example, via the first plurality 510 of parallel inputs) associates with a second trajectory in the second activity pattern (eg, the second plurality 512 from parallel inputs to the multi-coincidence unit 500).
- the multi-coincidence unit is overall designed to determine at which locations trajectories of the first activity pattern and trajectories of the second activity pattern intersect in the course of a displacement of the two activity patterns.
- FIG. 5 b shows a circuit diagram of a column, as may be used in a multi-coincidence unit 500, for example. A detailed description will be omitted here, as the column 570 shown in Fig. 5b is substantially identical in construction to a column j of coincidence cells 580 (see Figs. 5a and 5c).
- FIG. 6 a shows a schematic representation of the processes in the multi-coincidence unit 500 when two trajectories of a first activity pattern and a second activity pattern are pushed through the multi-coincidence unit 500. It should be noted here that pushing through a trajectory by the multi-coincidence unit 500 in a pictorial description corresponds to a displacement of the trajectory (for example, by one column). It should be noted, moreover, that the trajectories shown in FIGS. 6 a, 6 b and 6 c in the context of an evaluation by the multi-coincidence unit 500 by a plurality of parallel time signals are represented.
- Figure 6a shows in a first plot 610 a first trajectory 612 included, for example, in the first activity pattern and assumed to be input to the multi-coincidence unit by a plurality of parallel signals of the first plurality 510 of parallel signals becomes.
- Graph 610 also shows a second trajectory 614 in the second activity pattern. It is assumed that the second trajectory 614 in the second activity pattern is input to the multi-coincidence unit 500 through a plurality of parallel signals of the second plurality 512 of parallel binary signals. It is further assumed that the two trajectories 612, 614 have the same curvature and also occur simultaneously.
- the graphs 620, 624, 628, 632 show a stepwise displacement of the trajectories 612, 614 by the stages of the multi-coincidence unit 500. Since it is assumed that the trajectories 612, 614 have the same curvature and also occur simultaneously, it comes in the course The displacement of the trajectories 612, 614 to a coincidence in each case in a middle column of the multi-coincidence unit 500. However, with a progression of a displacement of the trajectories, coincidences occur in different lines of the multi-coincidence unit 500.
- the summer 530 associated with the center column of the multi-coincidence unit 500 therefore increases its count as the trajectories 612, 614 shift, as shown by thick-drawn dots 638 in the graphs 624, 628, 632, 636.
- a time-delayed trajectory 640 is input to the multicoinzide unit. Since the trajectory 640 is delayed in time relative to the trajectory 612, the trajectory 612 is already further displaced by the multi-coincidence unit than the trajectory 640, when there is a coincidence between the trajectory 612 and the trajectory 640.
- the corresponding situation is illustrated by graphs 644, 648, 652, 656 and 660.
- a middle column of the polygonal unit 500 is denoted by 664.
- a first trajectory supplied to the multi-coincidence unit 500 via the first plurality 510 of parallel signals and a second trajectory 614, 640 are supplied to the multi-coincidence unit 500 via a second plurality 512 of parallel signals is, have an equal curvature, so decides a time shift between the first trajectory 612 and the second trajectory 614, 640 on which column j of the multi-coincidence unit 500, a coincidence occurs.
- information about a time lag between a first trajectory in the first activity pattern (supplied to the multi-coincidence unit via the first plurality 510 of parallel signals) and a second trajectory in the second activity pattern (that of the multi-coincidence unit 500 via the second plurality 512 from parallel signals) can be deduced from which column of the multi-coincidence unit the coincidence occurs.
- FIG. 6 b shows a schematic representation of the resulting relationships in the multicoinzide unit 500 when the multicoinzoning unit 500 includes a first trajectory 670 included in the first activity pattern and a second trajectory 674 included in the second activity. It is assumed that the first trajectory 670 and the second trajectory 674 have a different curvature. In the illustrated extreme example according to FIG. 6b, it is assumed for the purpose of illustration that the first trajectory 670 is unconstrained or rectilinear. If the first trajectory 670 and the second trajectory 674 are displaced in the opposite direction by the multi-coincidence unit 500 as described, a first coincidence occurs in a column of the multi-coincidence unit 500, which is denoted 680 here.
- trajectory 670 is further shifted in the first direction, whereas the trajectory 674 is shifted further in the opposite direction, a next coincidence does not occur in the column 680 but typically in a column 682 of the multi-coincidence unit 500 adjacent thereto (see FIG Illustration 690).
- a coincidence finally occurs in a column 684, which is typically adjacent to the column 682 (see graph 692). Further continuation of the shift is shown in plot 694, where a coincidence results in yet another column 686.
- the multi-coincidence unit 500 is designed so that coincidences occur in the same column, respectively, when trajectories of the same curvature are shifted in the opposite direction by the multi-coincidence unit 500 stepwise. If, on the other hand, trajectories of different curvature are shifted stepwise through the multicoincidence unit 500, coincidences occur in different columns.
- the threshold value of the threshold decision 540 is set so that the threshold decision 540 only respond (or output an active output signal) when trajectories of a same curvature (or with different curvatures, which differ from each other by a predetermined maximum deviation) are pushed in the opposite direction by the Mui- tikoinzidenzappel 500.
- the multi-coincidence unit 500 is designed so that the occurrence of an active output signal at any of the threshold decision 540 indicates that in the first activity pattern and the second activity pattern two trajectories are at least approximately (within a predetermined tolerance range) of the same curvature. From the fact of which the threshold value decision 540 outputs an active signal at its output, there is also information about a time-Liehe shift between the trajectories of the same curvature.
- the multi-coincidence unit 500 can be replaced by any other device capable of recognizing in a first activity pattern and in a second activity pattern trajectories (or linear geometric shapes) which at least approximately ( ie, for example, within a tolerance range) have the same curvatures in order to determine a time shift between the trajectories of the same or approximately the same curvature.
- a length of the trajectories in the first activity pattern and in the second activity pattern in order to decide whether a first trajectory in the first activity pattern and in the second activity pattern a second trajectory in the second activity pattern is associated with a same sound event.
- Fig. 7 shows a block diagram of an identifier according to the invention according to an embodiment of the present invention.
- the identifier according to FIG. 7 is denoted by 700 in its entirety.
- the identifier 700 is configured to receive a first activity pattern in the form of a first plurality 710 of parallel signals or information.
- the identifier 700 is designed to receive a second activity pattern in the form of a second plurality 720 of parallel signals.
- the description of the activity patterns by the parallel signals 710, 720 can be carried out, for example, by binary signals, wherein active states characterize the occurrence of an activity event.
- the identifier 700 includes a first trajectory recognizer 730 and a second trajectory recognizer 732.
- the first trajectory detection 730 is configured to receive the first activity pattern via the first plurality 710 of parallel signals and to detect trajectories in the first activity pattern ,
- the first trajectory recognizer 730 is designed to detect in the first activity pattern line-shaped structures having a certain predetermined curvature.
- the trajectory recognizer may be configured to detect line-shaped structures having a plurality of different predetermined curvatures.
- the second trajectory recognizer 732 is configured to receive the second activity pattern via the second plurality 720 of parallel signals.
- the second trajectory recognizer 732 is designed, for example, to detect trajectories of different curvature.
- the first trajectory recognizer 730 is configured to activate at least one output line based on the first activity pattern so that the fact of which of the plurality of output lines is activated includes information about a curvature of a trajectory identified in the activity pattern.
- the trajectory recognizer 730 detects a trajectory In the first activity pattern, the trajectory recognizer preferably activates one (but possibly more than one) of its output lines 740, 742, 744, 746 as a function of the curvature of the trajectory.
- the second trajectory recognizer 732 is designed to activate at least one (preferably exactly one, but possibly also several) of its output lines 750, 752, 754, 756 in dependence on a curvature of a trajectory detected in the second activity pattern.
- the first trajectory recognizer 730 and the second trajectory recognizer 732 are designed such that corresponding output lines of the two trajectory recognizers 730, 732 indicate identical curvatures of trajectories (within a tolerance range).
- an ith output signal of the first trajectory recognizer 730 indicates presence of a trajectory having a certain ith curvature in the first activity pattern
- an ith output of the second trajectory recognizer 732 indicates presence a trajectory with the same ith curvature in the second activity pattern.
- the identifier 700 further includes a coincidence recognizer 770.
- the coincidence recognizer 770 is structurally similar to the multi-coincidence unit 500, for example.
- the coincidence recognizer 700 receives as input signals to the first plurality 510 of inputs, the output signals 740, 742, 744, for example 746 of the first trajectory recognizer 730.
- the coincidence recognizer 770 further preferably receives at the inputs of the second plurality 512 of inputs the output signals 750, 752, 754, 756 of the second trajectory recognizer 732. It is assumed here that the i-th row of the multi-coincidence unit 500 is supplied with the i-th output signal of the first trajectory recognizer 730 and the i-th output signal of the second trajectory recognizer 732.
- the first trajectory recognizer 730 and the second trajectory recognizer 732 detect trajectories of the same curvature, so ac-
- the first trajectory recognizer 730 and the second trajectory recognizer 732 respectively output corresponding output signals (for example, the respective ith output signal).
- coincidence occurs in the i-th row of the multi-coincidence unit 500 (which forms the coincidence recognizer 770).
- the coincidence recognizer 770 thus provides a plurality of preferably parallel output signals which indicate with which timely displacement trajectories of the same curvature occur. Incidentally, the output signals of the coincidence recognizer 770 are denoted by 780.
- the circuit arrangement 700 represents a further alternative in order to determine, based on two activity patterns, whether trajectories of the same curvature (ie trajectories associated with the same sound events) are present in the two activity patterns, and if so, which temporal shift these trajectories have ,
- the summers 530 may be replaced by an OR operation such that the outputs of the ORs are the outputs 780 of the coincidence Recognizer 770 and the identifier 700 form.
- various devices are described with reference to FIGS. 8, 9 and 10, which can be used, for example, as a trainee recognizer, and thus can take the place of the trajectory recognizers 730, 732 in certain embodiments of the present invention.
- FIG. 8 shows a block diagram of a device for processing the nerve activity pattern according to the invention.
- the device shown in Figure 8 is designated in its entirety by 15000.
- the illustrated device 15000 has a plurality of stages 15100, 15120, 15140, with the first stage 15100 receiving in parallel signals 15200, 15220, 15240 from nerve cells.
- the signals 15200, 15220, 15240 preferably describe an activity pattern (e.g., action potentials on nerve fibers coupled to the respective nerve cells or inner hair cells, or neurotransmitter vesicle occurrence on a plurality of inner auditory cells).
- the signals 15200, 15220, 15240 thus describe, for example, the nerve activity pattern.
- the signals can also describe a different activity pattern, for example a neurotransmitter vesicle occurrence in a plurality of inner hair cells.
- first stage 15100 for example, the first signal or nerve signal 15200 is then subjected to a delay in a first delay device 15300 and then forwarded as a delayed signal or nerve signal 15320 to a second stage 15120.
- the second signal or nerve signal 15220 is also delayed in the first stage 15100 and forwarded as a delayed signal nerve signal to the second stage 15120.
- the other signals or nerve signals in the first stage 15100 are processed (that is, for example, the nth signal or nerve signal 15240).
- the second stage 15120 is designed parallel to the first stage 15100, thus in turn allows the delayed transmission of the delayed signals or nerve signals 15320, 15340, 1536, whereby twice delayed signals or nerve signals arise.
- An apparatus for processing the activity pattern according to the invention comprises a plurality of stages connected in series, which are constructed the same as the first stage 15100 and the second stage 15120, respectively.
- the signals or nerve signals 15200, 15220, 15240 are thus forwarded in parallel through the plurality of stages 15100, 15120, 15140, each stage adding an adjustable delay to the signals or nerve signals.
- each of the stages 15100, 15120, 15140 is designed to form a sum of the signals or nerve signals (or m times delayed nerve signals) entering and exiting from it.
- stages 15100, 15120, 15140 are preferably designed to compare this sum with an adjustable threshold value in order to determine whether, at a given time, at least a predetermined number of signals or delayed signals or nerve signals (ie incoming signals or nerve signals or outgoing signals or nerve signals) are active (or have an action potential).
- the delays of the delay devices present in the stages 15100, 15120, 15140 are set differently, so that, for example, a first signal or nerve signal 15200 is subject to a different delay when passing through the stages 15100, 15120, 15140 Delays can be set, for example, in such a way that different total delays result for the signals or nerve signals 15200, 15220, 15240 when passing through the stages 15100, 15120, 15140 (although it is admittedly permissible that for example two signals or nerve signals are delayed in the same way).
- the device 15000 is preferably designed so that the same delays do not result for all signals or nerve signals.
- an activity pattern entering a device 15000 according to the invention over time as it passes through the device described is distorted in time such that individual signals or nerve signals are shifted in time compared to other signals or nerve signals. Due to the distortion, curved line-like patterns, ie trajectories, can be bent in a temporal representation in the activity pattern.
- the summation map within a stage can be used to detect when an originally curved trajectory in the activity pattern has been bent to a straight line (a straight line being described or recognized as having a predetermined number of delayed ones Signals or nerve signals have an active state or an action potential almost simultaneously or temporally overlapping).
- FIG. 9 shows an exemplary graphical representation of the signals in a device 15000 for processing the activity pattern according to the invention.
- the graph of Figure 9 is designated in its entirety 16000.
- a first graphical representation 16100 describes an exemplary activity pattern at inputs of the device 15000. Shown are, by way of example, the signals of four nerve cells (or four nerve fibers) in one time course. For the rest, it is pointed out that the action potentials 16120 form a trajectory 16140. As shown, the trajectory 16140 has a large curvature in the temporal representation, since the action potentials 16120 of the different nerve fibers at the inputs of the first stage 15100 have a marked temporal offset. Thus, only one action potential is present in the first stage 15100 at a fixed time, so that a threshold value for a sum of the action potentials applied to the first stage, which is set to two, for example, is not exceeded. Consequently, the first stage does not provide an output at a threshold output.
- a second graph 16200 describes the conditions at an output of the first stage 15100. It is assumed here that in the first stage 15100 the nerve signal delivered by the first nerve cell NZ 1 is delayed more than the nerve signals supplied by the other stages. Incidentally, in the given example, it is considered that the nerve signal delivered from the fourth nerve cell NZ4 is least delayed, while the nerve signal from the third nerve cell NZ3 is somewhat more delayed, and the delay for nerve signals from the nerve cells NZ2 and NZ1 always becomes increases more. Generally speaking, signals associated with nerve cells that respond to a lower frequency are less delayed than nerve signals from nerve cells that detect higher frequencies.
- the second graph thus again shows action potentials 16240 as a function of time, with the action potentials 16220 forming a trajectory 16240.
- the curvature of the trajectory 16240 at the first stage outputs is less than a (temporal-spatial) curvature of the trajectory 16160 at the first stage inputs. This results from the un- Different delay of the nerve cells belonging to different nerve signals in the delay devices (eg 15300) of the first stage. As a result, a curved trajectory is straightened as it were.
- the second trajectory 16240 still has a residual curvature, so that the action potentials 16220 originating from different nerve cells or nerve fibers do not all rest simultaneously at the outputs of the first stage 15100 or inputs of the second stage 15120 ,
- a third graph 1630 shows the second stage delayed nerve signals at second stage 15120. It can be seen from the third plot 16300 that, in the present example, the nerve signals at the second stage outputs are each delayed such that action potentials 16320 from several nerve cells are present at the outputs of the second stage simultaneously. In other words, a trajectory 16340, which is described by the action potential 16320, is at least approximately straight.
- the action potentials 16320 thus occur simultaneously or approximately simultaneously (but at least overlapping in time) so that the simultaneous occurrence by a summation of the signals present at the outputs of the second stage (or inputs of the third stage) has a clear peak which is large enough is to exceed a predetermined threshold (eg two or three).
- a predetermined threshold eg two or three
- a suitable summing means or other suitable means
- the corresponding information enables a return Conclude both the beginning time of the trajectory and the shape of the trajectory. Namely, it can be determined after passing through how many steps a trajectory has just been bent. As a result, knowing the delays for the individual nerve signals in the stages of the device 15000, it is also possible to deduce an original form of the trajectory. Furthermore, the passage time for the steps is preferably known, so that the time at which a trajectory has entered the device 15000 can also be determined. Thus, both characteristic time information of the trajectories and information about the shape or curvature of the trajectories can be determined in order to determine which activity events belong to a trajectory and / or which activity events do not belong to a trajectory.
- a fourth graphical representation 16400 still shows output signals at outputs of a third stage to improve the understanding.
- Action potentials 16420 describe a trajectory 16440 which, however, is curved again by further bending the trajectory.
- the delays in stages 15100, 15120, 15160 can be achieved in various ways.
- the delay devices eg 15300
- one or more delay devices may be deactivated in a predetermined stage for one or more nerve signals, so that some nerve signals are passed through a stage with the least possible delay.
- the device 15000 as a whole may be implemented as an analog or digital circuit.
- an evaluation of a nerve activity pattern has been described above. However, the device 15000 can be used for locating trajectories in any activity patterns.
- the signals 15200, 15220, 15240 correspond to the signals 710 and 720, for example.
- the threshold-weighted sum signals ⁇ x , ⁇ 2 , ⁇ i otherwise correspond to the signals 740, 742, 744, 746 or 750, 752, 754, 756.
- FIG. 10 shows a circuit diagram of an exemplary Hubel-Wiesel network for the calculation according to the invention of an analysis representation of an audio signal according to the second exemplary embodiment of the present invention.
- the circuit diagram of FIG. 17 is designated 17000 in its entirety.
- a first circuit block 17100 receives input signals 17200, 17220, 17240, which may represent, for example, a nerve activity pattern, a basilar membrane excitation pattern, or a neurotransmitter vesicle occurrence.
- the input signals 17200, 17220, 17240 are then passed through a plurality of stages 17300, 17320, 17340.
- An input signal 17200 thus passes through a plurality of stages 17300, 17320, 17340, wherein an input signal 17200 in a stage 17300, 17320, 17340 either undergoes a delay or is forwarded directly to a subsequent stage.
- the delay devices can also be bridged.
- each stage comprises a switchable delay device for each signal, wherein the delay device can be switched on or bypassed in a signal path through which an input signal passes.
- Signals at the inputs of each stage are tapped and summers 17400, 17420, 17440 are added. leads, wherein in each case the signals present at the inputs of a stage are added up.
- the first circuit block 17100 thus forms a grid of delay elements and adders connected in the manner shown.
- the Hubel-Wiesel network 17000 also has a threshold value .17500, wherein in each case one value from a threshold value register 17600, 17620, 17640 as well as an output of a summer 17400, 17430, 17440 is fed to a comparator 17700, 17720, 17720.
- Output signals 17800, 17820, 17840 of the comparators 17700, 17720, 17740 provide a statement as to whether at the inputs of a predetermined stage 17300, 17320, 17340 a number of signals are simultaneously active, with a minimum number at which an active output signal 17800, 17820, 17840, is defined by the threshold registers 17600, 17620, 17640.
- comparators 17700, 17720, 17740 may be used in conjunction with summers 17400, 17420, 17440 and threshold registers 17600, 17620, 17640 if (or after passing through many of stages 17300, 17320, 17340) a Trajek- torie, which was read in via the inputs 17200, 17220, 17240 of the first block 17100, is bent straight.
- the delays of the individual stages 17300, 17320, 17340 can be suitably specified in this case, in order to enable recognition of the greatest possible number of trajectories (or trajectory forms).
- the input signals 17200, 17220, 17240 correspond, for example, to the signals 710, 720 in accordance with FIG. 7, while the output signals 17800, 17820, 17840, for example, correspond to the signals 740, 724, 744, 746 or the signals 750, 752, 754, 756 according to FIG 7 correspond.
- an activity pattern for use in the context of the present invention is described.
- the invention can be calculated based on a hearing model of a human ear.
- an activity pattern is a description of an activity in or on a plurality of auditory cells of a hearing model or on a plurality of auditory nerves of a hearing model.
- intermediate variables can be calculated, for example, in the evaluation of a hearing model of an ear. Each of these intermediate variables is suitable, for example, for a determination of an activity pattern, wherein an activity event occurs when the intermediate variable in question deviates from a quiescent value which results without the presence of an audio signal by more than one assigned value.
- an activity event occurs when one of the intermediate variables or end variables mentioned below exceeds or falls below a predefined threshold value. It has also been shown that the transdermal release rate of 2680, the neurotransmitter vesicle occurrence 2760 and the nerve activity pattern 2840 are particularly well suited for the formation of the activity pattern.
- FIG. 11 shows a schematic representation of the sequence in a simulation of a human hearing as well as the intermediate and final results occurring in the simulation.
- the schematic representation of FIG. 1 is designated in its entirety by 2000.
- Fig. 11 describes a hearing model of a human ear used in conjunction with the present invention.
- the schematic representation 2000 of FIG. 11 thus describes a simulation model of a human ear.
- An audio signal 2100 serves as input signal for the simulation model 2000.
- a mechanical sound conversion in an outer ear is evaluated, whereby an excitation 2180 of an eardrum is determined.
- sound transmission via the auditory ossicles is calculated or simulated, whereby an excitation 2260 of an oval window between a middle ear and a cochlea is determined from the excitation 2180 of the eardrum.
- a hydromechanical vibration excitation 2340 of the cochlea is calculated or simulated.
- a basilar membrane movement 2420 is determined from the vibrational excitation 2340 of the cochlea.
- the basilar membrane movement 2420 is used to deduce a deflection 2520 of a stereocilium.
- a calcium concentration 2600 in a hair cell is then calculated in a sixth step 2560.
- the calcium concentration 2600 is then used to calculate a transmitter release rate 2680 of transmitter substances in a seventh step 2640.
- a neurotransmitter vesicle occurrence 2760 is derived, which describes an occurrence of neurotransmitter vesicles.
- a nerve activity pattern 2840 is derived from the neurotransmitter vesicle occurrence 2760.
- the nerve activity pattern 2840 describes approximately an activity occurring in healthy hearing on nerve cells of a (human or animal) auditory nerve.
- the nerve activity pattern 2840 is thus well suited to provide evidence of stimulation of auditory nerves by a cochlear implant. It should be noted that in the context of the simulation model 2000, several of the steps 2140, 2220, 2300, 2380, 2460, 2560, 2640, 2720, 2800 can be combined without calculating a corresponding intermediate result. In other words, several steps may be processed in a simplified step without calculating the intermediate steps shown in the graphical representation 2000.
- Both the stereodirection 2520, the calcium concentration 2600 in an inner hair cell, the median release rate 2680 in an inner hair cell, a neurotransmitter vesicle occurrence 2760 in an inner hair cell, or a nerve activity pattern applied to the inner hair cell 2840 (or the associated time course) can each represent an activity pattern or a part of an activity pattern on an inner hair cell.
- the concentrations 2600 of calcium ions in a plurality of inner hair cells may form an activity pattern.
- An activity event when looking at the calcium concentration is, for example, a significant increase in the calcium concentration above a certain threshold or a decrease in the calcium concentration below a certain threshold.
- the activity pattern may, for example, also describe a deviation of the current calcium concentrations from equilibrium calcium concentrations.
- An activity event in this case is described by a significant deviation of the respective calcium concentrations upwards or downwards.
- timing of a transmitter release rate 2680 or a transmitter release probability may also be in a plurality of inner hair cells be used as an activity pattern over time.
- the activity pattern is formed by the transmitter release rate or transmitter release probability in the ith inner hair cell.
- transmitter release rate 2680 or transmitter release probability for example, there is the presence of a transmitter release rate greater than zero, or a transmitter release probability greater than zero, as an activity event can be understood.
- neurotransmitter vesicle appearance 2760 may also form an activity pattern in a plurality of inner hair cells.
- an activity pattern in the mentioned case describes how many neurotransmitter vesicles are released, for example, during a certain time interval.
- neurotransmitter vesicle occurrence 2760 may also describe how many neurotransmitter vesicles are re-released during a period of time.
- the corresponding time course can also describe, for example, how many neurotransmitter vesicles are released in the presynaptic inner hair cell in a unit of time or are present in free form at a time interval or at a time altogether.
- neurotransmitter vesicle occurrence 2760 may also describe how many neurotransmitter vesicles diffuse, for example, per unit time from the presynaptic inner hair cell into the synaptic cleft, which couples the presynaptic inner hair cell to a nerve fiber.
- a neurotransmitter vesicle occurrence is understood, for example, to mean a number of neurotransmitter vesicles actually present or released per unit time, it not being relevant to the present invention exactly where within one Hair cell or synapse the appropriate number of neurotransmitter vesicles or the corresponding release rate of neurotransmitter vesicles is determined. It can be assumed that neurotransmitter vesicles which are released in the presynaptic part of the inner hair cell diffuse into the synaptic gap with a certain time constant.
- a neurotransmitter-vesicle occurrence can be described both qualitatively and quantitatively in order to describe an activity pattern in the sense of the present invention.
- a time course associated with a neurotransmitter vesicle occurrence 2760 may, for example, describe how many neurotransmitter vesicles are released or are present in free form. The course can also only provide qualitative information as to whether neurotransmitter vesicles are released in a time interval or are present in free form.
- the activity events may be the release of a number of neurotransmitter vesicles in a time unit greater than a certain threshold.
- a certain threshold For example, it can be assumed that an activity event always occurs if a neurotransmitter vesicle is ever released within a time interval. Further, it may be assumed that there is an activity event when at least one (or more generally, more than a predetermined minimum number) free neurotransmitter vesicle is present in an inner hair cell.
- an activity event is an event that occurs in a single inner hair cell.
- An activity event typically turns into a local one Minimum or maximum of the activity pattern or in an overshoot or undershoot of a threshold noticeable.
- a pattern of activity activity on a plurality of inner hair cells of a hearing model can also be used as the activity pattern.
- the nerve activity pattern describes the activity or the time course of the activity on several different nerve fibers, which are coupled with several different inner hair cells.
- An ith temporal course of the activity pattern is associated, for example, with a time profile of a potential or a voltage at a nerve fiber coupled to an ith inner hair cell.
- Action potentials occur on the individual nerve fibers and their occurrence is described by the activity pattern.
- An activity event in this case is to be understood as the occurrence of an action potential on one of the nerve fibers considered by n.
- FIG. 12 shows a flow diagram of a method according to the invention for generating a filtered activity pattern based on a first activity pattern on a hearing model of a first ear and a second activity pattern on a hearing model of a second ear.
- the method according to FIG. 12 is designated in its entirety by 1200.
- the method 1200 comprises, in a first step 1210, identifying a first trajectory in the first activity pattern and a second trajectory in the second Activity patterns associated with a same sound event.
- the method 1200 includes, in a second step 1220, determining whether the two trajectories are associated with a sound event of a useful sound source.
- the method 1200 further includes, in a third step 1230, filtering the first activity pattern or the second activity pattern based on a result of determining whether a trajectory is associated with a sonic event of the useful sound source, such that in a filtered activity pattern, activity events corresponding to a sound event of the one Are attributed to the useful sound source, or that activity events that are not associated with a sound event of the useful sound source are no longer contained in the filtered activity pattern.
- FIG. 13 shows a block diagram of a source separator according to an embodiment of the present invention.
- the source separator of Fig. 13 is designated in its entirety by 1300.
- Source divider 1300 is configured to receive a first channel 1310 of an at least two-channel audio signal.
- the source separator 1300 is further configured to receive a second channel 1320 of the at least two-channel audio signal.
- Source divider 1300 is further configured to provide a clean audio signal 1330 based on the audio signal having at least two channels.
- the source separator 1300 includes a first activity pattern calculator 1340 configured to generate a first activity pattern 110 based on the first channel 1310 of FIG To calculate audio signal.
- the activity pattern calculator 1340 includes or uses an ear model of an ear.
- the source separator 1300 further includes a second activity pattern calculator 1342 configured to calculate a second activity pattern 112 based on the second channel 1320 of the audio signal.
- the activity pattern calculator 1342 is designed, for example, to apply an ear model of an ear to the second channel 1320 of the audio signal in order to obtain the second activity pattern 112.
- the source separator 1300 includes an identifier 120, a determiner 130, and a filter 140 as already described with reference to FIG. 1.
- the device of the source separator 1300 which correspond to devices of the device 100, are designated by the same reference numerals as in FIGS. 1 and 2 and will not be explained again here. Rather, reference is made to the description with reference to FIGS. 1 and 2.
- the source separator 1300 includes, in addition to the devices of the device 100, 200, a synthesizer 1350 configured to obtain the filtered activity pattern 146 and to generate the adjusted audio signal 1330 based on the filtered activity pattern 146.
- the synthesizer is designed to transform the cleaned activity pattern 146 into a time representation, a frequency representation or a subband representation.
- the synthesizer 1350 is preferably designed to at least partially reverse the calculations made in determining the activity pattern based on the auditory model.
- the synthesizer 1350 is designed, for example, to reconstruct the adjusted audio signal as a time-domain audio signal based on an activity pattern representing a neurotransmitter vesicle occurrence.
- the synthesizer 1350 may also be configured to, for example, interpret a neural activity pattern in a time-based manner. transform rich audio signal.
- a frequency domain representation that is to say a representation of energies or complex amplitude values in a plurality of spectral regions, or as a representation in the form of a plurality of complex pointers in a plurality of frequency bands may also be used instead of the time domain representation of the audio signal , be used.
- Source divider 1300 thus allows source separation using a multi-channel (at least two-channel) audio signal, source separation based on recognition of related trajectories in two activity patterns representing the two channels of the audio signal.
- FIG. 14 shows a flow chart of a method according to the invention for generating a cleaned-up audio signal based on an audio signal with at least two channels.
- the method according to FIG. 14 is designated in its entirety by 1400.
- the method 1400 includes, in a first step 1410, generating a first activity pattern on a hearing model of a first ear based on a first channel of the audio signal.
- the method 1400 further comprises, in a second step 1420, generating a second activity pattern on a hearing model of a second ear based on a second channel of the audio signal.
- the method 1400 further comprises, in a third step 1430, generating a filtered activity pattern based on the first activity pattern and the second activity pattern, as already described above with reference to FIGS. 1 and 2.
- the method 1400 further includes, in a fourth step 1440, converting the filtered activity pattern into a time representation, frequency representation, or subband representation to obtain the adjusted audio signal.
- method 1400 may be extended to include all steps performed by the inventive devices described in the present application.
- the generation of the filtered activity pattern based on the first activity pattern and the second activity pattern may be performed by using the method 1200 of FIG. 12, for example.
- FIG. 15 shows an excerpt from a block diagram of a device according to the invention for calculating a filtered activity pattern based on two audio signals.
- the device according to FIG. 15 is denoted by 1500 in its entirety.
- the device 1500 is configured to receive a first audio signal 1510 from, for example, a first microphone located, for example, in an environment of a first ear (eg, a human).
- the device 1500 is further configured to receive a second audio signal 1512 from, for example, a second microphone located, for example, in an environment of a second (eg, human ear).
- the device 1500 is configured to receive the first audio signal 1510 from a microphone disposed at a left side of a human head, and to receive the second audio signal 1512 from a microphone disposed at a right side of a human head.
- the device 1500 further comprises a first activity pattern calculator 1520, which is designed, for example, to calculate a first activity pattern 1522 based on the first audio signal 1510 using a hearing model.
- the device 1500 further comprises an A second activity pattern calculator 1530 configured to calculate a second activity pattern 1532 based on the second audio signal 1512 using an ear model.
- the apparatus 1500 further includes a first Hubel-Wiesel network 1540 configured to receive the first activity pattern 1522 and, based thereon, to generate a plurality of parallel signals 1542 configured to detect the presence of trajectories different curvature in the activity pattern 1522 display.
- the plurality of parallel signals 1542 is configured to indicate when a trajectory having a curvature associated with a signal is comprised of a plurality of trajectories having different curvatures.
- the parallel lines 1542 correspond in their function to the signals 740, 742, 744, 746 of the device 700, and the first Hubel-Wiesel network 1540 corresponds to the first trajectory recognizer 730.
- the device 1500 further comprises a second hub.
- Wiesel network 1550 configured to generate a plurality of parallel signals 1552 based on the second activity pattern 1532.
- the parallel signals of the plurality 1542 of parallel signals in this case signal the presence of a trajectory with a certain curvature in the second activity pattern 1532.
- the parallel lines of the plurality 1552 of parallel lines that applies with regard to the plurality of parallel lines 1542 the said.
- the parallel lines 1552 correspond to the lines 750, 752, 754, 756 of the device 700
- the second Hubel-Wiesel network 1550 corresponds to the second trajectory recognizer 732 of the device 700.
- the apparatus 1500 further includes a multi-coincidence unit 1560 configured to receive first input signals via the first plurality 1542 of parallel lines and to receive second input signals via the second plurality 1552 of parallel lines. It should also be noted that the multi-coincidence unit 1560 substantially corresponds to the coincidence recognizer 770 of the apparatus 700. It should be noted that the multi-coincidence unit 1560 can be, for example, a multi-coincidence unit, as already described with reference to FIG. 5.
- the multi-coincidence unit 1560 is thus designed to determine, in conjunction with the Hubel-Wiesel networks 1540, 1550, whether trajectories of the same curvature are included in the activity patterns 1522, 1532, and further a time shift between the trajectories with the same To determine curvature.
- the corresponding information may then be used to filter the first activity pattern 1522 and / or the second activity pattern 1532 to obtain a filtered activity pattern, as discussed above.
- FIG. 16 shows how the inner hair cells or inner hair cells IHZ1, IHZ2, IHZ3, IHZ4 can be arranged along a basilar membrane.
- Fig. 16 shows for this purpose a graphic representation of a geometry of the basilar membrane and a response of the basilar membrane to an excitation.
- a first plot 2610 shows that a width of basilar membrane 2620 increases from the base of the cochlea to the end (apex) of the cochlea by about a factor of ten. Further, graph 2610 shows various characteristic frequencies (in hertz) with respect to which maximum sensitivity exists at different locations of the cochlea. Near the base of the cochlea, frequencies of the order of 20,000 hertz are perceived most strongly. From the base of the cochlea, the frequency for which maximum sensitivity results decreases continuously.
- Plot 610 further shows four exemplary inner hair cells IHZ1, IHZ2, IHZ3, IHZn, along the cochlea are arranged, and which are coupled with associated nerve fibers NFl, NF2, NF3, NFn.
- the first inner hair cell IHZ1 responds most strongly to an excitation with a frequency of approximately 6,000 Hz.
- the second inner hair cell IHZ2 has a maximum sensitivity when excited at a frequency of about 3,300 Hz.
- the remaining inner hair cells IHZ3, IHZn have other frequencies of maximum sensitivity.
- a second plot 2650 further describes coupling an acoustic wave into the cochlea via an oval window 2660.
- the coupling through the oval window 2660 creates a traveling wave 2670 in the cochlea that extends from a base 2680 of the cochlea to an apex 2682 of the cochlea Cochlea runs while deflecting the Basilarmembran 2620.
- the nerve cells located closer to the base of the cochlea are excited earlier than nerve cells farther from the base of the cochlea.
- the location of the traveling wave 2670 as a function of time can be considered as the trajectory of the traveling wave 2670.
- the trajectory can also be imaged on discrete nerve cells, so that a trajectory also describes in which temporal sequence several spatially separated nerve cells are excited by a traveling wave.
- the inner hair cell IHZ1 is earlier excited by the traveling wave 2670 than the other inner hair cells IHZ2, IHZ3, IHZn.
- the first inner hair cell IHZ1 is closer to the oval window 2660, with the traveling wave 2670 propagating from the oval window (ie, the base 2680 of the cochlea) to the apex 2682 of the cochlea.
- the first inner hair cell IHZ1, the second inner hair cell IHZ2, the third inner hair cell IHZ3 and the n-th inner hair cell IHZn are successively excited.
- a zige traveling wave generated at the inner hair cells shown activity events in a time sequence, the time intervals by the propagation velocity of the traveling wave 2670 and the location of the corresponding inner hair cells IHZL, IHZ2, IHZ3, IHZn be determined.
- the activity events on the inner hair cells (as viewed in a two-dimensional representation as a function of time and the index i of the inner hair cell IHZi form a trajectory, for example a time trajectory of a location of maximum excursion of the traveling wave 2670 equivalent.
- the methods according to the invention can be implemented in hardware or in software.
- the implementation may be on a digital storage medium, such as a floppy disk, CD, DVD, ROM, PROM, EPROM, EEPROM, or flash memory, with electronically readable control signals that may interact with a programmable computer system to perform the appropriate method .
- the invention also consists in a computer program product with program code stored on a machine-readable carrier for carrying out the method according to the invention, when the computer program product runs on a computer.
- the invention can thus be realized as a computer program with a program code for carrying out the method when the computer program runs on a computer.
- delay trajectories or delay trajectories (resulting, for example, from a propagation of a traveling wave on a cochlea of an inner ear when the cochlea is excited by a sound event and which reflect in activity patterns due to of the auditory model) in a multi- Tikoinzidenzü or multi-coincidence unit (eg in a multi-coincidence 500 according to FIG. 5a) run against each other.
- the multi-coincidence unit comprises n antiparallel delay lines of length m, where n describes a number of inner hair cells (or nerve fibers) which are taken into account in a calculation of the activity pattern, and where m describes a number of basins or bins and thus defines a maximum delay time and / or a maximum time resolution.
- activity patterns e.g., vesicles
- multi-coincidence unit e.g., multi-coincidence unit 500 of Figure 5a
- a summation e.g., of multi-coincidence signals from coincidence outputs of coincidence cells 580
- integrator cells or summers 530, respectively.
- Each delay trajectory has or includes, for example, n activity events or vesicles. Consequently, it is advantageous that each integrator (or summer 530) is provided with a threshold that is slightly below m (e.g., at least 50% of n). The threshold is otherwise represented in the block diagram according to FIG.
- n I.
- the integrators or the adders 530 are also reset within an integration time of about 10 to 100 ms (by one allow reliable detection of individual trajectories, and avoid mixing coincidence events of different trajectories).
- the delay trajectories in the activity patterns
- the delay trajectories are then matched in the multi-coincidence unit (eg, in the coincidence recognizer 700).
- the coincidence recognizer 700 it is recognized when trajectories are present in the activity patterns, and further, a time lag between the deceleration trajectories having the same curvature is determined.
- the present invention thus provides a device for binaural filtering of activity patterns or for binaural vesicle filtering.
- Delay trajectories in the activity patterns are matched against each other by a time shift. Details regarding this can be found, for example, in US Pat. No. 6,442,510 B1, the teaching of which is hereby incorporated by reference.
- the present invention enables a timely timing and synchronization of a left and a right cochlear implant so that sound sources can be located correctly in the room.
- the present invention thus enables binaural noise source filtering, which makes it possible to mitigate a cocktail party effect.
- the present invention makes it possible to carry out an angle determination of sound sources, for example by using a temporal displacement of trajectories belonging to the same sound event is determined. Furthermore, it is possible to select or select activity events or vesicles (as part of the filtering by the filter) depending on the sound source. In other words, depending on which sound source originates from activity events or vesicles, the activity events or vesicles are taken over into the filtered activity pattern or the filtered activity patterns are damped or suppressed.
- the method according to the invention brings about particular advantages when a cochlear implant is activated by the filtered activity pattern (or even better, by two filtered activity patterns).
- binaural sound source location can not be performed by the brain. This is the case, for example, because the electrodes of the implants lie in different places along the basilar membrane in both ears. Thus, no equal assignment to inner auditory cells or inner hair cells is possible.
- an advantageous use of the inventive concept can also be carried out when afferent outer hair cells are driven by an opposite ear (or by an output of the multi-coincidence unit).
- a cochlear implant can be built that excites intact outer hair cells in the same way, creating a new generation of cochlear implants.
- the output of the multi-coincidence unit is an input to an ef- fective feedback loop to control the outer motor hair cells to selectively adjust a dynamic level range.
- an ear model is paired. Signals from inner auditory cells or inner hair cells of the left and the right side are brought in pairs in antiparallel delay lines to coincidence (circuit variant 1).
- the system is tri-built with the separate signals from HSR, MSR and LSR spiral ganglion cells. Since these have a different response level range, the vesicle output varies for loud, medium, and quiet sounds. Thus, different loud signal sources can be separated.
- the concept according to the invention serves to reduce a cocktail party effect by locating the sound sources through the device according to the invention or the method according to the invention. Thereafter, vesicles that belong to a noise source (also referred to as a noise source) (or that belong to a noise source) are selectively filtered (or removed) (e.g., when the filtered activity pattern is generated).
- a noise source also referred to as a noise source
- vesicles that belong to a noise source are selectively filtered (or removed) (e.g., when the filtered activity pattern is generated).
- the multi-coincidence unit is preceded by a Hubel-Wiesel network.
- outputs of the multi-coincidence unit are applied to a feedback loop that innervates outer hair cells through efferent pathways.
- a selective activation signal is calculated from the input parameters, an action potential is triggered at the calculated efferent cell (or at the calculated effective cells) and thus the outer hair cell is induced to contract.
- the present invention thus provides a concept that can be used to generate a filtered activity pattern, wherein the filtered activity pattern is again advantageously usable for driving a cochlear implant, so that a conventionally occurring cocktail party can be applied to a wearer of the cochlear implant. Effect is reduced.
Abstract
Description
Claims
Priority Applications (6)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
AT07764867T ATE490802T1 (de) | 2006-06-30 | 2007-06-26 | Vorrichtung und verfahren zum erzeugen eines gefilterten aktivitätsmusters, quellentrenner, verfahren zum erzeugen eines bereinigten audiosignals und computerprogramm |
DE502007005899T DE502007005899D1 (de) | 2006-06-30 | 2007-06-26 | Vorrichtung und verfahren zum erzeugen eines gefilterten aktivitätsmusters, quellentrenner, verfahren zum erzeugen eines bereinigten audiosignals und computerprogramm |
US12/305,661 US8189834B2 (en) | 2006-06-30 | 2007-06-26 | Device and method for generating a filtered activity pattern, source divider, method for generating a debugged audio signal and computer program |
EP07764867A EP2024022B1 (de) | 2006-06-30 | 2007-06-26 | Vorrichtung und verfahren zum erzeugen eines gefilterten aktivitätsmusters, quellentrenner, verfahren zum erzeugen eines bereinigten audiosignals und computerprogramm |
AU2007264056A AU2007264056B2 (en) | 2006-06-30 | 2007-06-26 | Device and method for producing a filtered activity pattern, source separator, method for producing a corrected audio signal and computer program |
JP2009516978A JP4927166B2 (ja) | 2006-06-30 | 2007-06-26 | フィルタ処理された活性パターンを生成するためのデバイスおよび方法、音源分割器、デバッグされた音声信号を生成するための方法およびコンピュータ・プログラム |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
DE102006030276.1 | 2006-06-30 | ||
DE102006030276A DE102006030276A1 (de) | 2006-06-30 | 2006-06-30 | Vorrichtung und Verfahren zum Erzeugen eines gefilterten Aktivitätsmusters, Quellentrenner, Verfahren zum Erzeugen eines bereinigten Audiosignals und Computerprogramm |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2008000444A1 true WO2008000444A1 (de) | 2008-01-03 |
Family
ID=38512664
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/EP2007/005649 WO2008000444A1 (de) | 2006-06-30 | 2007-06-26 | Vorrichtung und verfahren zum erzeugen eines gefilterten aktivitätsmusters, quellentrenner, verfahren zum erzeugen eines bereinigten audiosignals und computerprogramm |
Country Status (8)
Country | Link |
---|---|
US (1) | US8189834B2 (de) |
EP (1) | EP2024022B1 (de) |
JP (1) | JP4927166B2 (de) |
AT (1) | ATE490802T1 (de) |
AU (1) | AU2007264056B2 (de) |
DE (2) | DE102006030276A1 (de) |
ES (1) | ES2357715T3 (de) |
WO (1) | WO2008000444A1 (de) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
DE102017221611B4 (de) * | 2017-11-30 | 2019-06-19 | Sivantos Pte. Ltd. | Verfahren zum Betreiben einer Vorrichtung zur Tinnitus-Charakterisierung sowie entsprechende Vorrichtung |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5434924A (en) * | 1987-05-11 | 1995-07-18 | Jay Management Trust | Hearing aid employing adjustment of the intensity and the arrival time of sound by electronic or acoustic, passive devices to improve interaural perceptual balance and binaural processing |
US20020012438A1 (en) * | 2000-06-30 | 2002-01-31 | Hans Leysieffer | System for rehabilitation of a hearing disorder |
US20030115054A1 (en) * | 2001-12-14 | 2003-06-19 | Nokia Corporation | Data-driven filtering of cepstral time trajectories for robust speech recognition |
US20030171786A1 (en) * | 2000-06-19 | 2003-09-11 | Blamey Peter John | Sound processor for a cochlear implant |
US20050069162A1 (en) * | 2003-09-23 | 2005-03-31 | Simon Haykin | Binaural adaptive hearing aid |
US20050177205A1 (en) * | 2004-01-09 | 2005-08-11 | Bomjun Kwon | Stimulation mode for cochlear implant speech coding |
US20050192646A1 (en) * | 2002-05-27 | 2005-09-01 | Grayden David B. | Generation of electrical stimuli for application to a cochlea |
US6987856B1 (en) * | 1996-06-19 | 2006-01-17 | Board Of Trustees Of The University Of Illinois | Binaural signal processing techniques |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
DE19750835C2 (de) | 1997-11-17 | 2002-06-27 | Frank Klefenz | Verfahren und Einrichtung zur Laufzeitdifferenzenbestimmung von akustischen Signalen |
FR2835124B1 (fr) | 2002-01-24 | 2004-03-19 | Telediffusion De France Tdf | Procede de synchronisation de deux flux de donnees numeriques de meme contenu |
US8843378B2 (en) | 2004-06-30 | 2014-09-23 | Fraunhofer-Gesellschaft Zur Foerderung Der Angewandten Forschung E.V. | Multi-channel synthesizer and method for generating a multi-channel output signal |
DE102005030327A1 (de) | 2005-06-29 | 2007-01-04 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Vorrichtung, Verfahren und Computerprogramm zur Analyse eine Audiosignals |
DE102005030326B4 (de) | 2005-06-29 | 2016-02-25 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Vorrichtung, Verfahren und Computerprogramm zur Analyse eines Audiosignals |
-
2006
- 2006-06-30 DE DE102006030276A patent/DE102006030276A1/de not_active Ceased
-
2007
- 2007-06-26 WO PCT/EP2007/005649 patent/WO2008000444A1/de active Application Filing
- 2007-06-26 AT AT07764867T patent/ATE490802T1/de active
- 2007-06-26 EP EP07764867A patent/EP2024022B1/de not_active Not-in-force
- 2007-06-26 JP JP2009516978A patent/JP4927166B2/ja not_active Expired - Fee Related
- 2007-06-26 ES ES07764867T patent/ES2357715T3/es active Active
- 2007-06-26 DE DE502007005899T patent/DE502007005899D1/de active Active
- 2007-06-26 AU AU2007264056A patent/AU2007264056B2/en not_active Ceased
- 2007-06-26 US US12/305,661 patent/US8189834B2/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5434924A (en) * | 1987-05-11 | 1995-07-18 | Jay Management Trust | Hearing aid employing adjustment of the intensity and the arrival time of sound by electronic or acoustic, passive devices to improve interaural perceptual balance and binaural processing |
US6987856B1 (en) * | 1996-06-19 | 2006-01-17 | Board Of Trustees Of The University Of Illinois | Binaural signal processing techniques |
US20030171786A1 (en) * | 2000-06-19 | 2003-09-11 | Blamey Peter John | Sound processor for a cochlear implant |
US20020012438A1 (en) * | 2000-06-30 | 2002-01-31 | Hans Leysieffer | System for rehabilitation of a hearing disorder |
US20030115054A1 (en) * | 2001-12-14 | 2003-06-19 | Nokia Corporation | Data-driven filtering of cepstral time trajectories for robust speech recognition |
US20050192646A1 (en) * | 2002-05-27 | 2005-09-01 | Grayden David B. | Generation of electrical stimuli for application to a cochlea |
US20050069162A1 (en) * | 2003-09-23 | 2005-03-31 | Simon Haykin | Binaural adaptive hearing aid |
US20050177205A1 (en) * | 2004-01-09 | 2005-08-11 | Bomjun Kwon | Stimulation mode for cochlear implant speech coding |
Also Published As
Publication number | Publication date |
---|---|
ES2357715T3 (es) | 2011-04-29 |
DE102006030276A1 (de) | 2008-01-03 |
US8189834B2 (en) | 2012-05-29 |
ATE490802T1 (de) | 2010-12-15 |
EP2024022B1 (de) | 2010-12-08 |
US20100232631A1 (en) | 2010-09-16 |
AU2007264056A1 (en) | 2008-01-03 |
JP4927166B2 (ja) | 2012-05-09 |
EP2024022A1 (de) | 2009-02-18 |
AU2007264056B2 (en) | 2010-09-30 |
DE502007005899D1 (de) | 2011-01-20 |
JP2009540978A (ja) | 2009-11-26 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
AT500645B1 (de) | Hörprothese, bilaterales hörprothesengerät und arbeitsverfahren hierfür | |
EP1977626B1 (de) | Verfahren zur aufnahme einer tonquelle mit zeitlich variabler richtcharakteristik und zur wiedergabe | |
DE102005030327A1 (de) | Vorrichtung, Verfahren und Computerprogramm zur Analyse eine Audiosignals | |
DE102017214164B3 (de) | Verfahren zum Betrieb eines Hörgeräts und Hörgerät | |
EP1207718A2 (de) | Verfahren zur Anpassung eines Hörgerätes, Vorrichtung hierzu und Hörgerät | |
EP1473967B1 (de) | Verfahren zur Unterdrückung mindestens eines akustischen Störsignals und Vorrichtung zur Durchführung des Verfahrens | |
EP2405673B1 (de) | Verfahren zum Lokalisieren einer Audioquelle und mehrkanaliges Hörsystem | |
EP1896123B1 (de) | Vorrichtung, verfahren und computerprogramm zur analyse eines audiosignals | |
DE102010039589A1 (de) | Hörhilfe- und/oder Tinnitus-Therapie-Gerät | |
WO2017157519A1 (de) | Verfahren zur automatischen bestimmung einer individuellen funktion einer dpoae-pegelkarte eines menschlichen oder tierischen gehörs | |
EP1981582B1 (de) | Vorrichtung und computerprogramm zum erzeugen eines ansteuersignals für ein cochlea-implantat basierend auf einem audiosignal | |
EP4017031A1 (de) | Hörsystem mit anwenderspezifischer programmierung | |
EP1463378A2 (de) | Verfahren zur Bestimmung einer Einfallsrichtung eines Signals einer akustischen Signalquelle und Vorrichtung zur Durchführung des Verfahrens | |
WO2001047335A2 (de) | Verfahren zur elimination von störsignalanteilen in einem eingangssignal eines auditorischen systems, anwendung des verfahrens und ein hörgerät | |
DE102014118674A1 (de) | Vorrichtung und Verfahren zur Ermittlung der Frequenz von mindestens einem dominanten Ton eines Ohrgeräuschs | |
EP2024022B1 (de) | Vorrichtung und verfahren zum erzeugen eines gefilterten aktivitätsmusters, quellentrenner, verfahren zum erzeugen eines bereinigten audiosignals und computerprogramm | |
DE102016214745B4 (de) | Verfahren zur Stimulation einer implantierten Elektrodenanordnung einer Hörprothese | |
EP3535019B1 (de) | Vorrichtung zur transkraniellen hirnstimulation | |
DE2739609A1 (de) | Verfahren und vorrichtung zur schulung und umschulung von weitgehend tauben | |
EP3834723A1 (de) | Verfahren zur ermittlung der hörschwelle einer testperson | |
DE3630518C2 (de) | Einrichtung zum lautweisen Identifizieren eines Sprechmusters | |
EP1380028A2 (de) | Verfahren zur elimination von störsignalanteilen in einem eingangssignal eines auditorischen systems, anwendung des verfahrens und ein hörgerät | |
DE102005004766A1 (de) | Verfahren zum Aufbereiten von Schallsignalen, insbesondere für das Zurverfügungstellen von aufzubereitenden Schallsignalen für schwerhörige Personen, und Vorrichtung zum Aufbereiten von Schallsignalen |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 07764867 Country of ref document: EP Kind code of ref document: A1 |
|
DPE1 | Request for preliminary examination filed after expiration of 19th month from priority date (pct application filed from 20040101) | ||
WWE | Wipo information: entry into national phase |
Ref document number: 2007764867 Country of ref document: EP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 12305661 Country of ref document: US |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2009516978 Country of ref document: JP |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
NENP | Non-entry into the national phase |
Ref country code: RU |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2007264056 Country of ref document: AU |
|
ENP | Entry into the national phase |
Ref document number: 2007264056 Country of ref document: AU Date of ref document: 20070626 Kind code of ref document: A |