US20060093048A9 - Partial Spectral Loss Concealment In Transform Codecs - Google Patents
Partial Spectral Loss Concealment In Transform Codecs Download PDFInfo
- Publication number
- US20060093048A9 US20060093048A9 US11/011,780 US1178004A US2006093048A9 US 20060093048 A9 US20060093048 A9 US 20060093048A9 US 1178004 A US1178004 A US 1178004A US 2006093048 A9 US2006093048 A9 US 2006093048A9
- Authority
- US
- United States
- Prior art keywords
- frequency
- coding coefficient
- frame
- spectral
- error concealment
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B1/00—Details of transmission systems, not covered by a single one of groups H04B3/00 - H04B13/00; Details of transmission systems not characterised by the medium used for transmission
- H04B1/66—Details of transmission systems, not covered by a single one of groups H04B3/00 - H04B13/00; Details of transmission systems not characterised by the medium used for transmission for reducing bandwidth of signals; for improving efficiency of transmission
- H04B1/667—Details of transmission systems, not covered by a single one of groups H04B3/00 - H04B13/00; Details of transmission systems not characterised by the medium used for transmission for reducing bandwidth of signals; for improving efficiency of transmission using a division in frequency subbands
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L63/00—Network architectures or network communication protocols for network security
- H04L63/14—Network architectures or network communication protocols for network security for detecting or protecting against malicious traffic
- H04L63/1441—Countermeasures against malicious traffic
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/50—Monitoring users, programs or devices to maintain the integrity of platforms, e.g. of processors, firmware or operating systems
- G06F21/55—Detecting local intrusion or implementing counter-measures
- G06F21/56—Computer malware detection or handling, e.g. anti-virus arrangements
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L51/00—User-to-user messaging in packet-switching networks, transmitted according to store-and-forward or real-time protocols, e.g. e-mail
- H04L51/21—Monitoring or handling of messages
- H04L51/212—Monitoring or handling of messages using filtering or selective blocking
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L63/00—Network architectures or network communication protocols for network security
- H04L63/14—Network architectures or network communication protocols for network security for detecting or protecting against malicious traffic
- H04L63/1441—Countermeasures against malicious traffic
- H04L63/1483—Countermeasures against malicious traffic service impersonation, e.g. phishing, pharming or web spoofing
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L63/00—Network architectures or network communication protocols for network security
- H04L63/20—Network architectures or network communication protocols for network security for managing network security; network security policies in general
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/48—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using compressed domain processing techniques other than decoding, e.g. modification of transform coefficients, variable length coding [VLC] data or run-length data
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/85—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using pre-processing or post-processing specially adapted for video compression
- H04N19/89—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using pre-processing or post-processing specially adapted for video compression involving methods or arrangements for detection of transmission errors at the decoder
- H04N19/895—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using pre-processing or post-processing specially adapted for video compression involving methods or arrangements for detection of transmission errors at the decoder in combination with error concealment
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2221/00—Indexing scheme relating to security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F2221/03—Indexing scheme relating to G06F21/50, monitoring users, programs or devices to maintain the integrity of platforms
Definitions
- the present invention generally relates to error concealment, and more particularly to a frequency-domain error concealment technique for use on the decoding side of a codec such as a sub-band codec or transform codec.
- a media encoder is a device, circuitry or computer program that is capable of analyzing an information stream such as an audio, video or image data stream and outputting an information stream representing the media in an encoded form. The resulting information is often used for transmission, storage and/or encryption purposes.
- a decoder is a device, circuitry or computer program that is capable of inverting the encoder operation, in that it receives the encoded information stream and outputs a decoded media stream.
- each frame of the input signal is analyzed in the frequency domain.
- the result of this analysis is quantized and encoded and then transmitted or stored depending on the application.
- a decoding procedure followed by a synthesis procedure allow to restore the signal in the time domain.
- Codecs are often employed for compression/decompression of information such as audio and video data for efficient transmission over bandwidth-limited communication channels.
- the most common audio and video codecs are sub-band codecs and transform codecs.
- a sub-band codec is based around a filter bank, and a transform codec is normally based around a time-to-frequency domain transform as for example the DCT (Discrete Cosine Transform).
- DCT Discrete Cosine Transform
- these two types of codecs can be regarded as mathematically equivalent. In a sense they are based on the same principle, where a transform codec can be seen as a sub-band codec with a large number of sub-bands.
- a common characteristic of these codecs is that they operate on blocks of samples: frames.
- the coding coefficients resulting from a transform analysis or a sub-band analysis of each frame are quantized according to a dynamic bit allocation, and may vary from frame to frame.
- the decoder upon reception of the bit-stream, computes the bit allocations and decodes the encoded coefficients.
- the quantized coding coefficients and/or parameters may be grouped in packets.
- a packet may contain data relevant to several frames, one frame or contain only partial frame data.
- Error concealment schemes typically rely on producing a replacement for the quantized coding coefficient(s) of a lost or more generally speaking erroneous packet that is similar to the original. This is possible since information such as audio, and in particular speech, exhibits large amounts of short-term self-similarity. As such, these techniques work optimally for relatively small loss rates (10%) and for small packets (4-40 ms).
- a technique known in the field of information transmission over unreliable channels is multiple description coding.
- the coder generates several different descriptions of the same audio signal and the decoder is able to produce a useful reconstruction of the original audio signal with any subset of the encoded descriptions.
- This technique assumes the occurrence of an error or a loss independently on each description. This would mean that each description would be transmitted on its own channel or that the descriptions share the same channel but are displaced, in time, with respect to each other. In this case the probability that the decoder receives valid data at each moment is high. The loss of one description can therefore be bridged by the availability of another description of the same signal.
- the method obviously increases the overall delay between the transmitter and the receiver. Furthermore, either the data rate has to be increased or some quality has to be sacrificed in order to allow the increase in redundancy.
- the estimation of missing signal intervals can be done in the time domain, i.e. at the output of the decoder, or in the frequency domain, i.e. internally to the decoder.
- the method of repetition is very similar to the muting technique, but instead of replacing the data by a zero signal when erroneous data occur, it repeats a part of the data that was last received.
- This method performs better than muting at the expense of an increase of memory consumption.
- the performance of this method is however limited and some quite annoying artifacts occur. For instance, if the last received frame was a drumbeat, then the latter is repeated which may lead to a double drumbeat where only one drumbeat was expected. Other artifacts may occur if, for instance, the frequency of repetition is short, which introduces a buzzy sound due to a comb filtering effect.
- a data substitution approach is employed with a hearing adjusted choice of the spectral energy. More particularly, a pattern is found in the intact audio data prior to the occurrence of erroneous data. When this pattern is found, replacement data is determined based on this pattern.
- a frequency-domain error concealment technique is described.
- the described technique is quite general and applies to transform coders. It uses prediction in order to restore the lost or erroneous coefficients.
- the prediction of an erroneous bin/frequency channel coefficient is based on the past coefficients of the same bin/channel, and may thus consider how the phase in a bin/frequency channel evolves over time in an attempt to preserve the so-called horizontal phase coherence. In some cases, this technique may provide quite satisfactory results.
- Wiese et al describe a technique for error concealment that is based on switching between several masking strategies, which include at least muting a sub-band and repeating or estimating the sub-band.
- the present invention overcomes these and other drawbacks of the prior art arrangements.
- Yet another object of the invention is to provide a general and efficient frequency-domain error concealment technique that can be applied to both sub-band and transform codecs.
- the invention concerns a frequency-domain error concealment technique for information that is represented, on a frame-by-frame basis, by coding coefficients.
- the basic idea is to conceal an erroneous coding coefficient by exploiting coding coefficient correlation in both time and frequency.
- the technique is applicable to any information, such as audio, video and image data, that is compressed into coding coefficients and transmitted under adverse channel conditions.
- the error concealment technique proposed by the invention has the clear advantage of to exploiting the redundancy of the original information signal in time as well as frequency. For example, this offers the possibility to exploit redundancy between frames as well as within frames.
- coding coefficients from the same frame as the erroneous coding coefficient is sometimes referred to as intra-frame coefficient correlation and it is a special case of the more general frequency correlation.
- using coefficients from one or more previous and/or subsequent frames is sometimes referred to as inter-frame correlation, or simply time correlation.
- the error concealment according to the invention is preferably performed by estimating a new coding coefficient based on at least one other coding coefficient within the same frame as the erroneous coefficient and at least one coding coefficient of one or more other frames, and replacing the erroneous (typically lost) coding coefficient by the new coding coefficient.
- the information may be represented by coding coefficients of a number of frequency bins, either frequency bands or transform frequency components.
- the new coding coefficient of this frequency bin may be estimated at least partly based on at least one coding coefficient of at least one other frequency bin in the same frame, and preferably also based on at least one coding coefficient of the same frequency bin in one or more other frames. It may be advantageous to consider also at least one coding coefficient of at least one other frequency bin in one or more other frames.
- a particularly beneficial implementation which does not introduce any extra delay, is based on estimating an erroneous coefficient not only from previous data of the erroneous or missing bin, but also on current and/or previous data of other bins. This means that both tune and frequency redundancy are exploited. This is especially true for the case of an audio signal that consists of the sum of harmonics whose frequency varies slowly over time. For this very common audio case, the locations of the peaks of the spectrum vary over time. For instance, a peak that is located at frame m ⁇ 1 would be located elsewhere at frame m. The use of an estimator or predictor exploiting this type of double redundancy is therefore very desirable.
- the present invention also suggests a special technique for estimating a new coding coefficient by predicting a spectral phase component based on approximate group delay matching between frames, using a predetermined approximation criterion. This is preferably performed by first estimating group delay from at least one other frame, and then calculating the spectral phase by at least approximately matching the group delay of the erroneous spectral component to the estimated group delay.
- a spectral amplitude component can be predicted based on matching the energy of spectral coefficients of the considered frame with the energy of corresponding spectral coefficients of at least one other frame.
- a new complex spectral coding coefficient of a certain frequency bin is preferably estimated by predicting spectral amplitude and phase separately and subsequently combining the predicted spectral amplitude and phase into a new complex spectral coding coefficient.
- the spectral energy matching and group delay matching can then be used for individually predicting the spectral amplitude component and spectral phase component, respectively, of the complex coding coefficient.
- an erroneous coding coefficient may be a partially erroneous coefficient or a totally lost coding coefficient.
- it may be possible to distinguish errors in the least significant bits from errors in the most significant bits of a coding coefficient, and in this way re-use at least parts of the information.
- the invention offers the following advantages
- FIG. 1 is a schematic overview of a conventional source coding application
- FIGS. 2 A-H are schematic diagrams illustrating various exemplary cases of exploiting both time and frequency correlation of coding coefficients
- FIG. 3 is a schematic diagram of a possibly overlapping frame division of the time domain input samples
- FIG. 4 is a schematic block diagram of an example of a basic transform-based coder
- FIG. 5 is a schematic block diagram of an example of a basic transform-based decoder with error concealment
- FIG. 6 is a schematic block diagram of an error concealment unit according to a preferred embodiment of the invention.
- FIG. 7 is a schematic block diagram of an example of a basic sub-band coder
- FIG. 8 is a schematic block diagram of an example of a basic sub-band decoder with error concealment
- FIGS. 9 A-B are schematic diagrams illustrating phase extrapolation based on group delay matching.
- FIG. 10 is a schematic block diagram of an estimator for complex coefficients according to a preferred embodiment of the invention.
- a codec is a composite device, circuitry or computer program capable of processing an information stream, and it generally comprises an encoding part and a decoding part. Codecs are often employed for compression/decompression of information such as audio and video data for efficient transmission over bandwidth-limited communication channels.
- each frame of the input signal is analyzed in the frequency domain.
- the result of this analysis is encoded and then transmitted.
- a synthesis procedure restores the signal in the time domain.
- frequency domain coding The basic concept in frequency domain coding is to divide the spectrum into frequency bands or components, commonly denoted frequency bins, using either a filter bank or a block transform analysis. After encoding and decoding, these frequency bins can be used to re-synthesize a replica of the input signal by either filter bank summation or an inverse transform.
- codecs Two well-known types of codecs that belong to the class of frequency domain codecs are sub-band codecs and transform codecs.
- the basic principle in both types of codecs is the division of the spectrum into frequency bins.
- sub-band coding a filter bank is employed to split the input signal into a number of relatively broad frequency bands.
- transform coding a block transformation method is employed to provide a much finer frequency resolution.
- codecs operate on blocks of samples: frames.
- the coding coefficients resulting from a transform analysis or a sub-band analysis of each frame are quantized, encoded and transmitted.
- the encoded and quantized coding coefficients are decoded to restore the original information.
- the coder 10 executes an encoding process for transforming the information stream into encoded form, typically as quantized and encoded coding coefficients.
- the encoded information is then forwarded to a channel-processing block 20 to put the encoded information in suitable form for transmission over the communication channel.
- the incoming bit stream is normally processed by the channel-processing block 30 , which may perform demultiplexation and error detection.
- the packets may be checked for bit errors by performing CRC (Cyclic Redundancy Check) checks or equivalent error detection. Often, packets with incorrect checksums are simply discarded.
- CRC Cyclic Redundancy Check
- an error concealment block is often employed in the decoding process of block 40 for concealing erroneous or missing coding coefficients by estimating new replacement coefficients.
- the decoding block 40 then executes a synthesis process on the non-erroneous coefficients and the estimated replacement coefficients to restore the original information.
- the invention concerns a specially designed technique for frequency-domain error concealment that is based on the idea of concealing an erroneous coding coefficient by exploiting coding coefficient correlation in both time and frequency.
- the technique is applicable to any information, such as audio, video and image data, that is compressed into coding coefficients and transmitted under adverse channel conditions.
- the error concealment technique proposed by the invention exploits the redundancy of the information signal in time and frequency, and offers the possibility to exploit redundancy between frames as well as within frames.
- a new coding coefficient based on coding coefficients from the same frame as the erroneous coding coefficient together with coefficients from one or more previous and/or subsequent frames.
- Another possibility is to use multiple coefficients from each of a number of previous and/or subsequent frames. Diagonal patterns of coefficient dependency in time and frequency can also be exploited.
- FIGS. 2A-2H are schematic diagrams illustrating various exemplary cases of exploiting both time and frequency correlation of coding coefficients. It should be understood that many other variations are possible, depending on design choice, desired computational complexity and so forth.
- FIG. 2A illustrates a basic example in which the considered erroneous coefficient is replaced based on the previous coefficient of the same frequency bin together with the coefficients of two adjacent bins within the same frame as the considered erroneous coefficient.
- This is a basic example of exploiting coefficient dependency in both time and frequency.
- the use of coding coefficients from the same frame as the erroneous coding coefficient is sometimes referred to as intra-frame coefficient correlation and it is a special case of the more general frequency correlation.
- using coefficients from one or more previous and/or subsequent frames is referred to as inter-frame correlation or time correlation.
- the principle of concealing an erroneous coding coefficient based on inter-frame as well as intra-frame coefficient correlation is particularly useful.
- FIG. 2B illustrates an example of successive erroneous coefficients in the same bin. It is here assumed that the erroneous coefficient of frame m has been replaced by a new estimated replacement coefficient, for example as illustrated in FIG. 2A . In the next frame m+1, the erroneous coefficient is replaced based on the replacement coefficient (indicated by an encircled dashed cross) of the same frequency bin in the previous frame m together with for example the coefficients of two adjacent bins within the same frame as the considered erroneous coefficient. It may be desirable to be able to adjust the influence of estimated replacement coefficients compared to non-erroneous coefficients.
- weighting factors may vary depending on whether the coefficients are non-erroneously transmitted coefficients or estimated replacement coefficients, and also depending on the “distance” in time (i.e. the number of frames) and/or frequency (i.e. the number of bins) from the considered erroneous coefficient
- FIG. 2C illustrates an example when several of the coding coefficients in the current frame are erroneous.
- the non-erroneous coding coefficient in the current frame is used together with the previous coefficient of the same frequency bin as well as coefficients of other frequency bins in the previous frame. This process is normally repeated for each of the erroneous coefficients of the current frame until they are all replaced by new coefficients
- FIG. 2D illustrates an example where several coding coefficients of more than one previous frame are considered together with coefficients within the current frame.
- FIG. 2E illustrates yet another example where the coefficients of the same frequency bin from several previous frames are used together with the coefficients of several bins within the current frame.
- FIG. 2F illustrates an example with a diagonal correlation pattern.
- FIG. 2G illustrates a basic example where a coefficient of the same bin in a subsequent frame is used together with the coefficients of two adjacent bins within the same frame as the considered erroneous coefficient. This means that when an erroneous coefficient is detected within a given frame, the error concealment algorithm has to wait until the next frame in order to access the coefficient(s) of the subsequent frame. Hence, this introduces a one-frame delay, and also assumes that the coefficient of the same bin in the subsequent frame is a non-erroneous/recovered coefficient.
- FIG. 2H illustrates another example with a delay of two frames, where a number of coefficients within the same frame as the considered erroneous coefficient are used together with as many non-erroneous/recovered coefficients as possible in the two directly following frames.
- each analysis frame m may be composed of possibly overlapping blocks of the time domain input samples x(n).
- FIG. 4 is a schematic block diagram of an example of a simple transform codec. It can be seen that each block x(m,n) of the input signal is multiplied by a weighting function h(n) and then transformed in the frequency domain by the use of an FFT (Fast Fourier Transform) unit 12 .
- FFT Fast Fourier Transform
- an FFT based encoder is just an example and that other types of transforms may be used, for example MDCT (Modified Discrete Cosine Transform).
- the obtained frequency-domain complex coefficients y(m,k), indexed by the bin number k, are quantized by the quantizer 14 into quantized complex coefficients y q (m,k).
- the quantized coefficients are then encoded and multiplexed by block 16 into a multiplexed information stream.
- the resulting framed bit stream is then packetized by block 18 and finally transmitted to the decoder on the receiving side.
- the incoming bit stream is de-packetized by the block 32 , which produces a framed bit stream as well as a bad frame indicator bfi(m) for each frame m.
- the bad frame indicator may be the result of a CRC check or detection of a lost packet.
- the framed bit stream and the corresponding bad frame indicator are forwarded to block 42 , which performs demultiplexation and decoding to extract quantized complex transform coefficients.
- the quantized coefficients are simply inverse transformed in an IFFT (Inverse FFT) unit 46 to obtain a time domain signal, which is multiplied by a window function w(n) and overlap-added in the overlap-add unit 48 to restore a time domain decoded signal x q (n).
- IFFT Inverse FFT
- the error concealment unit (ECU) 44 therefore receives an indication bci(m,k) of which spectral coefficients that are erroneous or missing, in addition to the extracted non-erroneous spectral coefficients y q (m,k). Based on the bad coefficient indicator bci(m,k), the error concealment unit 44 replaces those spectral coefficients that are indicated as erroneous or missing by new spectral coefficients.
- FIG. 6 is a schematic block diagram of an error concealment unit 44 according to a preferred embodiment of the invention.
- the logical units 52 and 54 operate to distinguish erroneous coefficients from non-erroneous coefficients.
- the bad coefficient indicator bci(m,k) is Boolean.
- the indicator is set to TRUE, which means that the coefficient is replaced by the output of the estimator 56 .
- the estimator needs to be always running in order to keep its internal memory state up-to-date, so it is only its output that is bridged as a replacement.
- the bci(m,k) therefore serves to select which spectral coefficients that need to be replaced by the spectral coefficient estimated by the estimator 56 .
- a recombination unit 58 receives and arranges the estimated replacement coefficients and non-erroneous coefficient of frame m for output.
- sub-band codec case will also be briefly illustrated with reference to FIGS. 7 and 8 .
- FIG. 7 is a schematic block diagram of an example of a basic sub-band coder.
- a bank of filters 12 - 1 to 12 -N is employed to split the input signal into a number N of frequency bands, each of which is normally low-pass translated to zero frequency to generate a corresponding coding coefficient y q (m,k).
- the obtained coefficients y(m,k), indexed by the bin number k, are then separately quantized by a set of quantizers 14 - 1 to 14 -N into quantized complex coefficients y q (m,k).
- the quantized coefficients are then encoded and multiplexed by block 16 , and then packetized by block 18 before transmission to the decoder on the receiving side.
- the incoming bit stream is de-packetized by the block 32 , which produces a framed bit stream as well as a bad frame indicator bfi(m) for each frame m.
- the framed bit stream and the bad frame indicator are forwarded to block 42 , which performs demultiplexation and decoding to extract quantized complex transform coefficients and a bad coefficient indicator bci(m,k). If no errors are detected, the quantized coefficients are simply translated back to their original frequency positions by a bank of filters 46 - 1 to 46 -N and summed together to give an approximation x q (n) of the original signal.
- the error concealment unit 44 receives an indication bci(m,k) of which spectral coefficients that are erroneous, in addition to the extracted non-erroneous coefficients y q (m,k). Based on the bad coefficient indicator, the error concealment unit 44 replaces those coefficients that are indicated as bad or missing by new spectral coefficients, similarly to what has been described above.
- amplitude and phase are preferably predicted separately and then combined.
- the gain G(m) can be estimated on several spectral bands by grouping the spectral coefficients into sub-bands and estimating the gain in each sub-band.
- the sub-band grouping can be on a uniform scale or a bark scale, which is motivated by psychoacoustics.
- ⁇ q ( m,k ) G ( m,l ) ⁇ Y q ( m ⁇ 1 ,k ), k ⁇ subband(l).
- the estimated gain on each spectral band greatly benefit from smoothing both in the time domain (smoothing in m) as well as in the frequency domain (smoothing in l) by the use of, for example, low pass filtering in the time and the frequency domain or a polynomial fit in the frequency domain and low pass filtering in the time domain.
- the sub-band embodiment is especially useful if the spectral missing coefficients are uniformly spread over the frequency axis. In certain situations, the spectral coefficients of a previously assigned sub-band grouping may all be lost. In this case, it is possible to merge the neighboring sub-band groupings or determine the gain associated with the sub-band as the average of the gain estimated in the neighboring sub-bands.
- Other strategies may of course be used without departing from the basic idea.
- a linear adaptive gain predictor may be used.
- phase prediction is more critical, since if the phase of the predicted spectral coefficients is far from the true one, a phase mismatch on the overlap sections leads to severe audible artifacts.
- phase vocoders in the context of time-stretching phase vocoders one of the main reasons for artifacts is the lack of phase coherence.
- the phase prediction technique proposed by the present invention uses redundancies of the information signal in both time and frequency.
- weighting coefficients are set proportional to the magnitude of the spectrum of the previous frame, or the predicted magnitude of the current frame, or a smoothed spectral envelope. This allows to emphasize the importance of the spectral peaks and to filter out the bad estimates of the phase derivative introduced by noise in the spectral valleys.
- the phase prediction is preferably based on estimating the group delay from at least one other (previous) frame, and determining the spectral phase of the erroneous coefficient such that the group delay associated with the erroneous coefficient gets as close as possible to the estimated group delay, according to some approximation criterion.
- phase prediction without departing from the basic idea of group delay matching/conservation. For instance, by additional exploitation of time-domain redundancies with the group delay conservation.
- FIG. 10 is a schematic block diagram of an estimator for complex coefficients according to a preferred embodiment of the invention.
- the estimator 56 basically comprises a storage unit 60 for storing coding coefficients belonging to a selectable number of frames, and a unit 70 for performing the necessary calculations to estimate new replacement coefficients.
- the storage unit 60 receives the extracted coefficients of the current frame and stores these together with non-erroneous/recovered coding coefficients belonging to one one previous frames.
- the calculation unit 70 receives information S on which coefficients to estimate, and calculates corresponding replacement coefficients based on the stored coefficients accessed from the storage unit 60 .
- the calculation unit 70 comprises an amplitude estimation unit 72 that operates based on the previously described energy matching principles, a phase estimation unit 74 that operates based on the previously described group delay matching principles, as well as a combination unit 76 for combination of the estimated phase and amplitude components into complex coefficients.
- phase and amplitude estimation techniques proposed by the invention can be used independently.
- phase could be estimated based on group delay matching as indicated above, with a simpler estimate of amplitude.
- amplitude could be estimated based on spectral energy matching as indicated above, with a simpler estimate of phase.
- complex spectral coefficients are predicted directly.
- the predictor output ⁇ q (m,k) is preferably dependent on at least the previous spectral coefficient(s) of the same bin as well as previous and/or current spectral coefficients of other bins.
- k, k 2 . . . , k p denote the indices of the non-erroneous spectral coefficients.
- the predictor function can, for instance, take the form of a linear predictor.
Landscapes
- Engineering & Computer Science (AREA)
- Computer Security & Cryptography (AREA)
- Signal Processing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Computer Hardware Design (AREA)
- General Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Virology (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Compression, Expansion, Code Conversion, And Decoders (AREA)
Abstract
Description
- The present invention generally relates to error concealment, and more particularly to a frequency-domain error concealment technique for use on the decoding side of a codec such as a sub-band codec or transform codec.
- A media encoder is a device, circuitry or computer program that is capable of analyzing an information stream such as an audio, video or image data stream and outputting an information stream representing the media in an encoded form. The resulting information is often used for transmission, storage and/or encryption purposes. On the other-hand a decoder is a device, circuitry or computer program that is capable of inverting the encoder operation, in that it receives the encoded information stream and outputs a decoded media stream.
- In most state-of the art audio and video encoders, each frame of the input signal is analyzed in the frequency domain. The result of this analysis is quantized and encoded and then transmitted or stored depending on the application. At the receiving side, or when using the stored encoded signal, a decoding procedure followed by a synthesis procedure allow to restore the signal in the time domain.
- Codecs are often employed for compression/decompression of information such as audio and video data for efficient transmission over bandwidth-limited communication channels.
- The most common audio and video codecs are sub-band codecs and transform codecs. A sub-band codec is based around a filter bank, and a transform codec is normally based around a time-to-frequency domain transform as for example the DCT (Discrete Cosine Transform). However, these two types of codecs can be regarded as mathematically equivalent. In a sense they are based on the same principle, where a transform codec can be seen as a sub-band codec with a large number of sub-bands.
- A common characteristic of these codecs is that they operate on blocks of samples: frames. The coding coefficients resulting from a transform analysis or a sub-band analysis of each frame are quantized according to a dynamic bit allocation, and may vary from frame to frame. The decoder, upon reception of the bit-stream, computes the bit allocations and decodes the encoded coefficients.
- In packet-based communications, the quantized coding coefficients and/or parameters may be grouped in packets. A packet may contain data relevant to several frames, one frame or contain only partial frame data.
- Under adverse channel conditions, the encoded/compressed information from the coder may get lost or arrive at the decoding side with errors. In general, transmission of audio, video and other relevant data under adverse channel conditions has become one of the most challenging problems today. In order to alleviate the effect of errors introduced by packet losses or corrupted data during transmission, so-called error concealment is often employed to reduce the degradation of the quality of audio, video or other data represented by the coding coefficients.
- Error concealment schemes typically rely on producing a replacement for the quantized coding coefficient(s) of a lost or more generally speaking erroneous packet that is similar to the original. This is possible since information such as audio, and in particular speech, exhibits large amounts of short-term self-similarity. As such, these techniques work optimally for relatively small loss rates (10%) and for small packets (4-40 ms).
- A technique known in the field of information transmission over unreliable channels is multiple description coding. The coder generates several different descriptions of the same audio signal and the decoder is able to produce a useful reconstruction of the original audio signal with any subset of the encoded descriptions. This technique assumes the occurrence of an error or a loss independently on each description. This would mean that each description would be transmitted on its own channel or that the descriptions share the same channel but are displaced, in time, with respect to each other. In this case the probability that the decoder receives valid data at each moment is high. The loss of one description can therefore be bridged by the availability of another description of the same signal. The method obviously increases the overall delay between the transmitter and the receiver. Furthermore, either the data rate has to be increased or some quality has to be sacrificed in order to allow the increase in redundancy.
- In the case of block or frame oriented transform codecs, the estimation of missing signal intervals can be done in the time domain, i.e. at the output of the decoder, or in the frequency domain, i.e. internally to the decoder.
- In the time domain, several error concealment techniques are already known in the prior art. Rudimentary technology as the muting-based methods repair their losses by muting the output signal for as long as the data is erroneous. The erroneous data is replaced by a zero signal. Although very simple, this method leads to very unpleasant effects due to the perceived discontinuities it introduces with sudden falls of the signal energy.
- The method of repetition is very similar to the muting technique, but instead of replacing the data by a zero signal when erroneous data occur, it repeats a part of the data that was last received. This method performs better than muting at the expense of an increase of memory consumption. The performance of this method is however limited and some quite annoying artifacts occur. For instance, if the last received frame was a drumbeat, then the latter is repeated which may lead to a double drumbeat where only one drumbeat was expected. Other artifacts may occur if, for instance, the frequency of repetition is short, which introduces a buzzy sound due to a comb filtering effect.
- Other more sophisticated techniques aim at interpolating the audio signal by, for example, either waveform substitution, pitch based waveform replication or time scale modification. These techniques perform much better than the previously described rudimentary techniques. However, they require much more complexity. Moreover, the amount of delay that is required to perform the interpolation is, in many cases, unacceptable.
- Techniques well known in the literature of audio restoration, e.g. [1], [2], [3], offer useful insights, and in fact deal with similar problems.
- Error concealment in the frequency-domain has been considered in [4], [5]. In the case of the DCT (Discrete Cosine Transform) transform, it is found that a simple concealment technique is to clip large DCT coefficients.
- In [6], a data substitution approach is employed with a hearing adjusted choice of the spectral energy. More particularly, a pattern is found in the intact audio data prior to the occurrence of erroneous data. When this pattern is found, replacement data is determined based on this pattern.
- In [7], a frequency-domain error concealment technique is described. The described technique is quite general and applies to transform coders. It uses prediction in order to restore the lost or erroneous coefficients. The prediction of an erroneous bin/frequency channel coefficient is based on the past coefficients of the same bin/channel, and may thus consider how the phase in a bin/frequency channel evolves over time in an attempt to preserve the so-called horizontal phase coherence. In some cases, this technique may provide quite satisfactory results.
- However, the error concealment technique proposed in [7] generally results in a loss of so-called vertical phase coherence, which may lead to frame discontinuities and perceived artifacts.
- In [8], Wiese et al describe a technique for error concealment that is based on switching between several masking strategies, which include at least muting a sub-band and repeating or estimating the sub-band.
- The present invention overcomes these and other drawbacks of the prior art arrangements.
- It is a general object of the present invention to provide an improved error concealment technique.
- It is another object of the invention to provide a frequency-domain error concealment technique that more optimally exploits the redundancy of the original information signal.
- Yet another object of the invention is to provide a general and efficient frequency-domain error concealment technique that can be applied to both sub-band and transform codecs.
- It is also an object to provide an improved frequency-domain error concealment arrangement, as well as a decoder and a receiver comprising such an error concealment arrangement.
- These and other objects are met by the invention as defined by the accompanying patent claims.
- The invention concerns a frequency-domain error concealment technique for information that is represented, on a frame-by-frame basis, by coding coefficients. The basic idea is to conceal an erroneous coding coefficient by exploiting coding coefficient correlation in both time and frequency. The technique is applicable to any information, such as audio, video and image data, that is compressed into coding coefficients and transmitted under adverse channel conditions. The error concealment technique proposed by the invention has the clear advantage of to exploiting the redundancy of the original information signal in time as well as frequency. For example, this offers the possibility to exploit redundancy between frames as well as within frames.
- There are many possibilities of exploiting time and frequency correlation, including using coding coefficients from the same frame in which the erroneous coding coefficient resided together with coefficients from one or more previous and/or subsequent frames, using several different coefficients from each of a number of previous and/or subsequent frames, or even using diagonal patterns of coding coefficients. It should though be understood that using coding coefficients from one or more subsequent frames generally introduces a delay, which may or may not be acceptable depending on the application.
- The use of coding coefficients from the same frame as the erroneous coding coefficient is sometimes referred to as intra-frame coefficient correlation and it is a special case of the more general frequency correlation. Similarly, using coefficients from one or more previous and/or subsequent frames is sometimes referred to as inter-frame correlation, or simply time correlation.
- The error concealment according to the invention is preferably performed by estimating a new coding coefficient based on at least one other coding coefficient within the same frame as the erroneous coefficient and at least one coding coefficient of one or more other frames, and replacing the erroneous (typically lost) coding coefficient by the new coding coefficient.
- As normal in sub-band and transform codecs, the information may be represented by coding coefficients of a number of frequency bins, either frequency bands or transform frequency components. In a particularly beneficial and practical implementation, when an erroneous coding coefficient is detected for a certain frequency bin in a certain frame, the new coding coefficient of this frequency bin may be estimated at least partly based on at least one coding coefficient of at least one other frequency bin in the same frame, and preferably also based on at least one coding coefficient of the same frequency bin in one or more other frames. It may be advantageous to consider also at least one coding coefficient of at least one other frequency bin in one or more other frames.
- A particularly beneficial implementation, which does not introduce any extra delay, is based on estimating an erroneous coefficient not only from previous data of the erroneous or missing bin, but also on current and/or previous data of other bins. This means that both tune and frequency redundancy are exploited. This is especially true for the case of an audio signal that consists of the sum of harmonics whose frequency varies slowly over time. For this very common audio case, the locations of the peaks of the spectrum vary over time. For instance, a peak that is located at frame m−1 would be located elsewhere at frame m. The use of an estimator or predictor exploiting this type of double redundancy is therefore very desirable.
- In particular, the present invention also suggests a special technique for estimating a new coding coefficient by predicting a spectral phase component based on approximate group delay matching between frames, using a predetermined approximation criterion. This is preferably performed by first estimating group delay from at least one other frame, and then calculating the spectral phase by at least approximately matching the group delay of the erroneous spectral component to the estimated group delay.
- A spectral amplitude component can be predicted based on matching the energy of spectral coefficients of the considered frame with the energy of corresponding spectral coefficients of at least one other frame.
- In the case of transform coding, when the coding coefficients are complex spectral transform coefficients, a new complex spectral coding coefficient of a certain frequency bin is preferably estimated by predicting spectral amplitude and phase separately and subsequently combining the predicted spectral amplitude and phase into a new complex spectral coding coefficient. The spectral energy matching and group delay matching can then be used for individually predicting the spectral amplitude component and spectral phase component, respectively, of the complex coding coefficient.
- It should be understood that an erroneous coding coefficient may be a partially erroneous coefficient or a totally lost coding coefficient. In more advanced error detection protocols, it may be possible to distinguish errors in the least significant bits from errors in the most significant bits of a coding coefficient, and in this way re-use at least parts of the information.
- The invention offers the following advantages;
-
- Improved error concealment;
- Optimal exploitation of the redundancy of the original information signal;
- Generally applicable to any sub-band or transform codec application.
- Other advantages offered by the present invention will be appreciated upon reading of the below description of the embodiments of the invention.
- The invention, together with further objects and advantages thereof, will be best understood by reference to the following description taken together with the accompanying drawings, in which;
-
FIG. 1 is a schematic overview of a conventional source coding application; - FIGS. 2A-H are schematic diagrams illustrating various exemplary cases of exploiting both time and frequency correlation of coding coefficients;
-
FIG. 3 is a schematic diagram of a possibly overlapping frame division of the time domain input samples, -
FIG. 4 is a schematic block diagram of an example of a basic transform-based coder; -
FIG. 5 is a schematic block diagram of an example of a basic transform-based decoder with error concealment; -
FIG. 6 is a schematic block diagram of an error concealment unit according to a preferred embodiment of the invention, -
FIG. 7 is a schematic block diagram of an example of a basic sub-band coder; -
FIG. 8 is a schematic block diagram of an example of a basic sub-band decoder with error concealment; - FIGS. 9A-B are schematic diagrams illustrating phase extrapolation based on group delay matching; and
-
FIG. 10 is a schematic block diagram of an estimator for complex coefficients according to a preferred embodiment of the invention. - Throughout the drawings, the same reference characters will be used for corresponding or similar elements.
- For a better understanding of the invention, it may be useful to begin with a brief overview of a common source coding application that involves transmission of encoded information over a communication channel. As mentioned earlier, a codec is a composite device, circuitry or computer program capable of processing an information stream, and it generally comprises an encoding part and a decoding part. Codecs are often employed for compression/decompression of information such as audio and video data for efficient transmission over bandwidth-limited communication channels.
- In most state-of the art audio and video codecs, each frame of the input signal is analyzed in the frequency domain. The result of this analysis is encoded and then transmitted. At the receiving side, a synthesis procedure restores the signal in the time domain.
- The basic concept in frequency domain coding is to divide the spectrum into frequency bands or components, commonly denoted frequency bins, using either a filter bank or a block transform analysis. After encoding and decoding, these frequency bins can be used to re-synthesize a replica of the input signal by either filter bank summation or an inverse transform.
- Two well-known types of codecs that belong to the class of frequency domain codecs are sub-band codecs and transform codecs. The basic principle in both types of codecs is the division of the spectrum into frequency bins. In sub-band coding, a filter bank is employed to split the input signal into a number of relatively broad frequency bands. In transform coding on the other hand, a block transformation method is employed to provide a much finer frequency resolution.
- A common characteristic of these codecs is that they operate on blocks of samples: frames. The coding coefficients resulting from a transform analysis or a sub-band analysis of each frame are quantized, encoded and transmitted. On the receiving side, the encoded and quantized coding coefficients are decoded to restore the original information.
- With reference to
FIG. 1 , thecoder 10 executes an encoding process for transforming the information stream into encoded form, typically as quantized and encoded coding coefficients. The encoded information is then forwarded to a channel-processing block 20 to put the encoded information in suitable form for transmission over the communication channel. On the receiver side, the incoming bit stream is normally processed by the channel-processing block 30, which may perform demultiplexation and error detection. In packet-based communication for example, the packets may be checked for bit errors by performing CRC (Cyclic Redundancy Check) checks or equivalent error detection. Often, packets with incorrect checksums are simply discarded. In order to alleviate the effect of errors introduced into the packets during transmission, an error concealment block is often employed in the decoding process ofblock 40 for concealing erroneous or missing coding coefficients by estimating new replacement coefficients. Thedecoding block 40 then executes a synthesis process on the non-erroneous coefficients and the estimated replacement coefficients to restore the original information. - The invention concerns a specially designed technique for frequency-domain error concealment that is based on the idea of concealing an erroneous coding coefficient by exploiting coding coefficient correlation in both time and frequency. The technique is applicable to any information, such as audio, video and image data, that is compressed into coding coefficients and transmitted under adverse channel conditions. The error concealment technique proposed by the invention exploits the redundancy of the information signal in time and frequency, and offers the possibility to exploit redundancy between frames as well as within frames.
- There are many possibilities of exploiting the time and frequency correlation/dependency of the coding coefficients. In order to estimate a new coding coefficient to be used instead of an erroneous or lost coefficient, it is desirable to analyze and determine how phase and/or amplitude evolves over time (between frames) and also how the phase and/or amplitude evolves with respect to frequency. This is sometimes also referred to as horizontal correlation/dependency and vertical correlation/dependency, respectively.
- For example, for a given erroneous coefficient it is possible to estimate a new coding coefficient based on coding coefficients from the same frame as the erroneous coding coefficient together with coefficients from one or more previous and/or subsequent frames. Another possibility is to use multiple coefficients from each of a number of previous and/or subsequent frames. Diagonal patterns of coefficient dependency in time and frequency can also be exploited.
- It should though be understood that using coding coefficients from one or more subsequent frames generally introduces a delay, which may or may not be acceptable depending on the application. In general, it is of course possible to use not only non-erroneous coding coefficients, but also previously estimated replacement coefficients.
-
FIGS. 2A-2H are schematic diagrams illustrating various exemplary cases of exploiting both time and frequency correlation of coding coefficients. It should be understood that many other variations are possible, depending on design choice, desired computational complexity and so forth. - In the simplified schematic of
FIG. 2A it is assumed that an erroneous coding coefficient (indicated by a cross) has been detected for a given frequency bin k in a given block or frame m.FIG. 2A illustrates a basic example in which the considered erroneous coefficient is replaced based on the previous coefficient of the same frequency bin together with the coefficients of two adjacent bins within the same frame as the considered erroneous coefficient. This is a basic example of exploiting coefficient dependency in both time and frequency. The use of coding coefficients from the same frame as the erroneous coding coefficient is sometimes referred to as intra-frame coefficient correlation and it is a special case of the more general frequency correlation. Similarly, using coefficients from one or more previous and/or subsequent frames is referred to as inter-frame correlation or time correlation. The principle of concealing an erroneous coding coefficient based on inter-frame as well as intra-frame coefficient correlation is particularly useful. -
FIG. 2B illustrates an example of successive erroneous coefficients in the same bin. It is here assumed that the erroneous coefficient of frame m has been replaced by a new estimated replacement coefficient, for example as illustrated inFIG. 2A . In the next frame m+1, the erroneous coefficient is replaced based on the replacement coefficient (indicated by an encircled dashed cross) of the same frequency bin in the previous frame m together with for example the coefficients of two adjacent bins within the same frame as the considered erroneous coefficient. It may be desirable to be able to adjust the influence of estimated replacement coefficients compared to non-erroneous coefficients. This may be accomplished by providing weighting factors that may vary depending on whether the coefficients are non-erroneously transmitted coefficients or estimated replacement coefficients, and also depending on the “distance” in time (i.e. the number of frames) and/or frequency (i.e. the number of bins) from the considered erroneous coefficient -
FIG. 2C illustrates an example when several of the coding coefficients in the current frame are erroneous. In this case, the non-erroneous coding coefficient in the current frame is used together with the previous coefficient of the same frequency bin as well as coefficients of other frequency bins in the previous frame. This process is normally repeated for each of the erroneous coefficients of the current frame until they are all replaced by new coefficients -
FIG. 2D illustrates an example where several coding coefficients of more than one previous frame are considered together with coefficients within the current frame. -
FIG. 2E illustrates yet another example where the coefficients of the same frequency bin from several previous frames are used together with the coefficients of several bins within the current frame. -
FIG. 2F illustrates an example with a diagonal correlation pattern. -
FIG. 2G illustrates a basic example where a coefficient of the same bin in a subsequent frame is used together with the coefficients of two adjacent bins within the same frame as the considered erroneous coefficient. This means that when an erroneous coefficient is detected within a given frame, the error concealment algorithm has to wait until the next frame in order to access the coefficient(s) of the subsequent frame. Apparently, this introduces a one-frame delay, and also assumes that the coefficient of the same bin in the subsequent frame is a non-erroneous/recovered coefficient. -
FIG. 2H illustrates another example with a delay of two frames, where a number of coefficients within the same frame as the considered erroneous coefficient are used together with as many non-erroneous/recovered coefficients as possible in the two directly following frames. - The invention will now be described in more detail, mainly with reference to transform and sub-band codecs. For more detailed information on sub-band and transform codecs including information on bit allocation, step sizes and decimation, reference is made to [9].
- As shown in
FIG. 3 , each analysis frame m may be composed of possibly overlapping blocks of the time domain input samples x(n).FIG. 4 is a schematic block diagram of an example of a simple transform codec. It can be seen that each block x(m,n) of the input signal is multiplied by a weighting function h(n) and then transformed in the frequency domain by the use of an FFT (Fast Fourier Transform)unit 12. Obviously, it should be understood that an FFT based encoder is just an example and that other types of transforms may be used, for example MDCT (Modified Discrete Cosine Transform). The obtained frequency-domain complex coefficients y(m,k), indexed by the bin number k, are quantized by thequantizer 14 into quantized complex coefficients yq(m,k). The quantized coefficients are then encoded and multiplexed byblock 16 into a multiplexed information stream. The resulting framed bit stream is then packetized byblock 18 and finally transmitted to the decoder on the receiving side. - As illustrated in
FIG. 5 , on the receiving side, the incoming bit stream is de-packetized by theblock 32, which produces a framed bit stream as well as a bad frame indicator bfi(m) for each frame m. The bad frame indicator may be the result of a CRC check or detection of a lost packet. The framed bit stream and the corresponding bad frame indicator are forwarded to block 42, which performs demultiplexation and decoding to extract quantized complex transform coefficients. If no errors are detected, the quantized coefficients are simply inverse transformed in an IFFT (Inverse FFT)unit 46 to obtain a time domain signal, which is multiplied by a window function w(n) and overlap-added in the overlap-add unit 48 to restore a time domain decoded signal xq(n). - Depending on how the encoded data is multiplexed and packetized, data relative to one frame can be partially or entirely lost. This has the effect that at least parts of the spectral coefficients may be erroneous. Demultiplexation of the bad frame indicator bfi(m) will determine which coding coefficients that are erroneous, thus producing a bad coefficient indicator bci(m,k). In a preferred embodiment of the invention, the error concealment unit (ECU) 44 therefore receives an indication bci(m,k) of which spectral coefficients that are erroneous or missing, in addition to the extracted non-erroneous spectral coefficients yq(m,k). Based on the bad coefficient indicator bci(m,k), the
error concealment unit 44 replaces those spectral coefficients that are indicated as erroneous or missing by new spectral coefficients. -
FIG. 6 is a schematic block diagram of anerror concealment unit 44 according to a preferred embodiment of the invention. Based on the bad coefficient indicator of all frequency bins k in frame m, thelogical units error concealment unit 44 simply outputs its input, i.e. ŷq(m,k)=yq(m,k). On the other hand, when a bad or missing coefficient is detected, the indicator is set to TRUE, which means that the coefficient is replaced by the output of theestimator 56. Sometimes the estimator needs to be always running in order to keep its internal memory state up-to-date, so it is only its output that is bridged as a replacement. The bci(m,k) therefore serves to select which spectral coefficients that need to be replaced by the spectral coefficient estimated by theestimator 56. In the following, the set of indices k of the erroneous spectral coefficients in frame m is denoted S={k, such that bci(m,k) TRUE}. Arecombination unit 58 receives and arranges the estimated replacement coefficients and non-erroneous coefficient of frame m for output. - For generality, the sub-band codec case will also be briefly illustrated with reference to
FIGS. 7 and 8 . -
FIG. 7 is a schematic block diagram of an example of a basic sub-band coder. In a sub-band coder, a bank of filters 12-1 to 12-N is employed to split the input signal into a number N of frequency bands, each of which is normally low-pass translated to zero frequency to generate a corresponding coding coefficient yq(m,k). The obtained coefficients y(m,k), indexed by the bin number k, are then separately quantized by a set of quantizers 14-1 to 14-N into quantized complex coefficients yq(m,k). The quantized coefficients are then encoded and multiplexed byblock 16, and then packetized byblock 18 before transmission to the decoder on the receiving side. - As illustrated in
FIG. 8 , on the receiving side, the incoming bit stream is de-packetized by theblock 32, which produces a framed bit stream as well as a bad frame indicator bfi(m) for each frame m. The framed bit stream and the bad frame indicator are forwarded to block 42, which performs demultiplexation and decoding to extract quantized complex transform coefficients and a bad coefficient indicator bci(m,k). If no errors are detected, the quantized coefficients are simply translated back to their original frequency positions by a bank of filters 46-1 to 46-N and summed together to give an approximation xq(n) of the original signal. Under adverse channel conditions, when errors occur during the transmission, theerror concealment unit 44 receives an indication bci(m,k) of which spectral coefficients that are erroneous, in addition to the extracted non-erroneous coefficients yq(m,k). Based on the bad coefficient indicator, theerror concealment unit 44 replaces those coefficients that are indicated as bad or missing by new spectral coefficients, similarly to what has been described above. - Without loss of generality, some examples of combined time and frequency correlation utilization will now be described for the case of complex coding coefficients. It should though be understood that some of the basic underlying principles for exploiting coefficient correlation in time as well as frequency described below may also be applied to single-valued coding coefficients. In addition, we mainly focus on implementations for real-time applications that require no or very small delays. Hence, only previous frames are considered for estimation of new coding coefficients in the following examples.
- Amplitude and Phase Prediction
- In this embodiment, amplitude and phase are preferably predicted separately and then combined. The amplitude and the phase of the spectrum are related to the spectral coefficients by the following relations:
Y q(m,k)=√{square root over (Re(y q(m,k))2 +Im(y q(m,k))2)}
φq(m,k)=arctan(Im(y q(m,k))/Re(y q(m,k))) - The predictor then predicts the amplitude Ŷq(m,k) and the phase {circumflex over (φ)}(m,k) and then combines them to obtain the predicted spectral coefficient;
{circumflex over (y)}q(m,k)=Ŷ q(m,k)cos({circumflex over (φ)}q(m,k))+jŶ q(m,k)sin({circumflex over (φ)}q(m,k)).
Amplitude Prediction - Conventionally, the amplitude prediction is often based on simply repeating the previous bin amplitude:
Ŷ q(m,k)=Y q(m−1,k). - This has the drawback that if for example an audio signal has a decreasing magnitude, the prediction leads to over-estimation which can be badly perceived.
- A more elaborate scheme, proposed by the present invention, exploits redundancy in both time and frequency, which allows a better prediction of the spectral magnitude. For example, the predicted spectral magnitude can be written as:
Ŷ q(m,k)=γG(m)·Y q(m−1,k),
where G(m) is an adaptive gain obtained by matching the energy of non-erroneous/recovered spectral coefficients of the current frame with the corresponding spectral coefficients of the previous frame, the factor γ is an attenuation factor, 0<γ≦1, e.g. γ=0.9. An example of energy matching can be to compute the adaptive gain as: - Other types of spectral energy matching measures may be used without departing from the basic idea of the invention.
- In another embodiment, the gain G(m) can be estimated on several spectral bands by grouping the spectral coefficients into sub-bands and estimating the gain in each sub-band. The sub-band grouping can be on a uniform scale or a bark scale, which is motivated by psychoacoustics. The adaptive gain in sub-band l is therefore estimated by:
- The predicted amplitude of the spectral coefficients in frequency sub-band l is given by:
Ŷ q(m,k)=G(m,l)·Y q(m−1,k), kεsubband(l). - The estimated gain on each spectral band greatly benefit from smoothing both in the time domain (smoothing in m) as well as in the frequency domain (smoothing in l) by the use of, for example, low pass filtering in the time and the frequency domain or a polynomial fit in the frequency domain and low pass filtering in the time domain.
- The sub-band embodiment is especially useful if the spectral missing coefficients are uniformly spread over the frequency axis. In certain situations, the spectral coefficients of a previously assigned sub-band grouping may all be lost. In this case, it is possible to merge the neighboring sub-band groupings or determine the gain associated with the sub-band as the average of the gain estimated in the neighboring sub-bands. Another strategy involves re-using the previous gain, i.e. G(m,l)=G(m−1,l), or a frequency domain filtered version, i.e.
Other strategies may of course be used without departing from the basic idea. - For the case when all spectral coefficients are lost, then the adaptive gain matching may be estimated either by using the previous two frames, or by using the previous adaptive gain matching, i.e. G(m,l)=G(m−1,l).
- More sophisticated, but more complex, means may be used for the gain prediction. For instance, a linear adaptive gain predictor may be used. The prediction may then be formed by:
where the predictor coefficients Apt(m,k) are adaptively adjusted, for example in some least error sense such as the least mean square.
Phase Prediction - The phase prediction is more critical, since if the phase of the predicted spectral coefficients is far from the true one, a phase mismatch on the overlap sections leads to severe audible artifacts. In the article “Improved Phase Vocoder Time-Scale Modification of Audio” by Laroche and Dolson [10], they mention that in the context of time-stretching phase vocoders one of the main reasons for artifacts is the lack of phase coherence.
- Preferably, the phase prediction technique proposed by the present invention uses redundancies of the information signal in both time and frequency. A particularly advantageous model is based on approximate group delay matching. This comes from the observation in audio applications that for a stationary single tone the derivative of the phase with respect to the frequency, i.e. the group delay, is approximately constant over time. This is justified in theory for a constant amplitude complex tone:
x(m,n)=A·e jω0 n+jφu
x(m+1,n)=A·e jω0 n+jφu +jω0 L
where L is the amount of overlap. - The windowed DFT (Discrete Fourier Transform) of both signal sections is given by;
X(m,ω)=AH(ω−ω0)e jφ0
X(m+1,ω)=AH(ω−ω0)e jφ0 +jω0 L
and it is easily seen that the group delay of both signal sections is the same
which shows that the group delay is constant and does not depend on m. This result can be shown to hold approximately for multiple tones, depending on how good is the window rejection band. - Therefore, estimating the derivative of the phase from the previous frame(s) allows the estimation of the missing spectral components phase by extrapolation.
- A simple way of performing the phase prediction based on group delay matching is to first estimate the derivative of the phase in the previous frame. This can be done by using simple finite differences:
Δφ(m−1,k)=φ(m−1,k)−φ(m−1,k−1). - Other ways of obtaining an estimate of the group delay may of course be used. The idea then is to approximately restore the same group delay for each missing spectral component. This can be achieved by computing the predicted phases such that they minimize an error function, for example:
where the unkmown parameters are φ(m,k) such that kεS, i.e. the phase of the lost spectral coefficients, and W(k) are positive weighting coefficients. - It is advantageous that the weighting coefficients are set proportional to the magnitude of the spectrum of the previous frame, or the predicted magnitude of the current frame, or a smoothed spectral envelope. This allows to emphasize the importance of the spectral peaks and to filter out the bad estimates of the phase derivative introduced by noise in the spectral valleys.
- In other words, the phase prediction is preferably based on estimating the group delay from at least one other (previous) frame, and determining the spectral phase of the erroneous coefficient such that the group delay associated with the erroneous coefficient gets as close as possible to the estimated group delay, according to some approximation criterion.
- An example of a solution in the case of W(k)=1 is given. As shown in FIGS. 9A-B, the lost coefficients are between bin K and bin K+N.
- The minimization of the error criterion leads to the following recursive solution for the extrapolated-predicted phase:
{circumflex over (φ)}(m,k)={circumflex over (φ)}(m,k−1)+Δφ(m−1,k)+Δφc , k=K+1, . . . ,K+N−1,
where
Δφc=(1/N)·(φ(m,K+N)−φ(m−1, K+N)−φ(m,K)+φ(m−1,K)). - In this solution, it is quite obvious that {circumflex over (φ)}(m,K)=φ(m,K) is used to start the recursion.
- For the case when all spectral coefficients are lost, then a secondary phase predictor is used to allow initialization of the above recursion.
- More sophisticated, but more complex means may be used for the phase prediction without departing from the basic idea of group delay matching/conservation. For instance, by additional exploitation of time-domain redundancies with the group delay conservation.
-
FIG. 10 is a schematic block diagram of an estimator for complex coefficients according to a preferred embodiment of the invention. Theestimator 56 basically comprises astorage unit 60 for storing coding coefficients belonging to a selectable number of frames, and aunit 70 for performing the necessary calculations to estimate new replacement coefficients. Thestorage unit 60 receives the extracted coefficients of the current frame and stores these together with non-erroneous/recovered coding coefficients belonging to one one previous frames. Thecalculation unit 70 receives information S on which coefficients to estimate, and calculates corresponding replacement coefficients based on the stored coefficients accessed from thestorage unit 60. In a preferred embodiment of the invention, adapted for complex transform coefficients, thecalculation unit 70 comprises anamplitude estimation unit 72 that operates based on the previously described energy matching principles, aphase estimation unit 74 that operates based on the previously described group delay matching principles, as well as acombination unit 76 for combination of the estimated phase and amplitude components into complex coefficients. - It should though be understood that the advanced phase and amplitude estimation techniques proposed by the invention can be used independently. For example, phase could be estimated based on group delay matching as indicated above, with a simpler estimate of amplitude. On the other hand, amplitude could be estimated based on spectral energy matching as indicated above, with a simpler estimate of phase.
- Direct Coefficient Prediction
- In this embodiment, complex spectral coefficients are predicted directly. The predictor output ŷq(m,k) is preferably dependent on at least the previous spectral coefficient(s) of the same bin as well as previous and/or current spectral coefficients of other bins.
- In general, this can be represented by a time-dependent adaptive predictor function fm,k such that:
ŷ q(m,k)=f m,k( y q(m−1), . . . , y q(m−q),y(m,k 1),y(m,k 1), . . . ,y(m,k p)).
where k, k2 . . . , kp denote the indices of the non-erroneous spectral coefficients. The predictor function can, for instance, take the form of a linear predictor. - The embodiments described above are merely given as examples, and it should be understood that the present invention is not limited thereto. Further modifications, changes and improvements which retain the basic underlying principles disclosed and claimed herein are within the scope of the invention.
-
- [1] S. J. Godsill, P. J. W. Rayner, “Digital Audio Restoration”, Springer, 1998.
- [2] J. J. K. O Ruanaidh, W. J. Fitzgerald, “Numerical Bayesian Methods Applied to Signal Processing”, Springer 1998.
- [3] R. Veldhuis “Restauration of lost samples in digital signals”, Prentice Hall, 1990.
- [4] J. Herre, E. Eberlein, “Error Concealment in the spectral domain”, 93rd AES Convention, 1992 Oct., 1-4, preprint 3364.
- [5] J. Herre, E. Eberlein, “Evaluation of concealment techniques for compressed digital audio”, 94th AES Convention, 1993 Oct., 1-4, preprint 3364.
- [6] U.S. Pat. No. 6,421,802-B1
- [7] EP-0 574 288-B1
- [8] U.S. Pat. No. 6,351,728-B1
- [9] A. M. Kondoz, “Digital Speech: Coding For Low Bit Rate Communication”, Wiley (1994), pp. 123-128.
- [10] J. Laroche, M. Dolson, “Improved Phase Vocoder Time-Scale Modification of Audio”, IEEE transactions on speech and audio processing, 323-332, Vol. 7, No 3, May 1999.
- claims
Claims (33)
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US11/011,780 US7356748B2 (en) | 2003-12-19 | 2004-12-15 | Partial spectral loss concealment in transform codecs |
Applications Claiming Priority (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US53065303P | 2003-12-19 | 2003-12-19 | |
SE0400418-0 | 2004-02-20 | ||
SE0400418A SE527669C2 (en) | 2003-12-19 | 2004-02-20 | Improved error masking in the frequency domain |
US11/011,780 US7356748B2 (en) | 2003-12-19 | 2004-12-15 | Partial spectral loss concealment in transform codecs |
Related Child Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US14/634,726 Continuation US10270800B2 (en) | 2003-12-11 | 2015-02-28 | Method for computer security based on message and message sender |
Publications (3)
Publication Number | Publication Date |
---|---|
US20050163234A1 US20050163234A1 (en) | 2005-07-28 |
US20060093048A9 true US20060093048A9 (en) | 2006-05-04 |
US7356748B2 US7356748B2 (en) | 2008-04-08 |
Family
ID=34799359
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US11/011,780 Active 2026-05-15 US7356748B2 (en) | 2003-12-19 | 2004-12-15 | Partial spectral loss concealment in transform codecs |
Country Status (1)
Country | Link |
---|---|
US (1) | US7356748B2 (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20080126096A1 (en) * | 2006-11-24 | 2008-05-29 | Samsung Electronics Co., Ltd. | Error concealment method and apparatus for audio signal and decoding method and apparatus for audio signal using the same |
EP2112653A1 (en) * | 2007-05-24 | 2009-10-28 | Panasonic Corporation | Audio decoding device, audio decoding method, program, and integrated circuit |
US20100115370A1 (en) * | 2008-06-13 | 2010-05-06 | Nokia Corporation | Method and apparatus for error concealment of encoded audio data |
US20110070892A1 (en) * | 2009-09-21 | 2011-03-24 | Andrew Llc | System and method for a high throughput gsm location solution |
US20150279380A1 (en) * | 2006-11-30 | 2015-10-01 | Samsung Electronics Co., Ltd. | Frame error concealment method and apparatus and error concealment scheme construction method and apparatus |
US10784988B2 (en) | 2018-12-21 | 2020-09-22 | Microsoft Technology Licensing, Llc | Conditional forward error correction for network data |
US10803876B2 (en) * | 2018-12-21 | 2020-10-13 | Microsoft Technology Licensing, Llc | Combined forward and backward extrapolation of lost network data |
Families Citing this family (86)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7725324B2 (en) * | 2003-12-19 | 2010-05-25 | Telefonaktiebolaget Lm Ericsson (Publ) | Constrained filter encoding of polyphonic signals |
US7668712B2 (en) * | 2004-03-31 | 2010-02-23 | Microsoft Corporation | Audio encoding and decoding with intra frames and adaptive forward error correction |
JP4679513B2 (en) * | 2004-04-28 | 2011-04-27 | パナソニック株式会社 | Hierarchical coding apparatus and hierarchical coding method |
WO2006079349A1 (en) | 2005-01-31 | 2006-08-03 | Sonorit Aps | Method for weighted overlap-add |
US9626973B2 (en) * | 2005-02-23 | 2017-04-18 | Telefonaktiebolaget L M Ericsson (Publ) | Adaptive bit allocation for multi-channel audio encoding |
US7627467B2 (en) * | 2005-03-01 | 2009-12-01 | Microsoft Corporation | Packet loss concealment for overlapped transform codecs |
US7418394B2 (en) * | 2005-04-28 | 2008-08-26 | Dolby Laboratories Licensing Corporation | Method and system for operating audio encoders utilizing data from overlapping audio segments |
US7177804B2 (en) * | 2005-05-31 | 2007-02-13 | Microsoft Corporation | Sub-band voice codec with multi-stage codebooks and redundant coding |
US7831421B2 (en) * | 2005-05-31 | 2010-11-09 | Microsoft Corporation | Robust decoder |
US7707034B2 (en) * | 2005-05-31 | 2010-04-27 | Microsoft Corporation | Audio codec post-filter |
US8620644B2 (en) * | 2005-10-26 | 2013-12-31 | Qualcomm Incorporated | Encoder-assisted frame loss concealment techniques for audio coding |
US7590523B2 (en) * | 2006-03-20 | 2009-09-15 | Mindspeed Technologies, Inc. | Speech post-processing using MDCT coefficients |
US8798172B2 (en) * | 2006-05-16 | 2014-08-05 | Samsung Electronics Co., Ltd. | Method and apparatus to conceal error in decoded audio signal |
KR100883652B1 (en) * | 2006-08-03 | 2009-02-18 | 삼성전자주식회사 | Method and apparatus for speech/silence interval identification using dynamic programming, and speech recognition system thereof |
US8165872B2 (en) * | 2007-02-01 | 2012-04-24 | Broadcom Corporation | Method and system for improving speech quality |
KR101086435B1 (en) * | 2007-03-29 | 2011-11-25 | 삼성전자주식회사 | Method for detecting errors from image data stream and apparatus thereof |
US10002189B2 (en) | 2007-12-20 | 2018-06-19 | Apple Inc. | Method and apparatus for searching using an active ontology |
US9330720B2 (en) | 2008-01-03 | 2016-05-03 | Apple Inc. | Methods and apparatus for altering audio output signals |
US20100030549A1 (en) | 2008-07-31 | 2010-02-04 | Lee Michael M | Mobile device having human language translation capability with positional feedback |
US8676904B2 (en) | 2008-10-02 | 2014-03-18 | Apple Inc. | Electronic devices with voice command and contextual data processing capabilities |
CN101989425B (en) * | 2009-07-30 | 2012-05-23 | 华为终端有限公司 | Method, device and system for multiple description voice frequency coding and decoding |
US8682667B2 (en) | 2010-02-25 | 2014-03-25 | Apple Inc. | User profiling for selecting user specific voice input processing information |
US9262612B2 (en) | 2011-03-21 | 2016-02-16 | Apple Inc. | Device access using voice authentication |
US10057736B2 (en) | 2011-06-03 | 2018-08-21 | Apple Inc. | Active transport based notifications |
US10134385B2 (en) | 2012-03-02 | 2018-11-20 | Apple Inc. | Systems and methods for name pronunciation |
CN103714821A (en) | 2012-09-28 | 2014-04-09 | 杜比实验室特许公司 | Mixed domain data packet loss concealment based on position |
WO2014197335A1 (en) | 2013-06-08 | 2014-12-11 | Apple Inc. | Interpreting and acting upon commands that involve sharing information with remote devices |
US10176167B2 (en) | 2013-06-09 | 2019-01-08 | Apple Inc. | System and method for inferring user intent from speech inputs |
CN110442699A (en) | 2013-06-09 | 2019-11-12 | 苹果公司 | Operate method, computer-readable medium, electronic equipment and the system of digital assistants |
PT3011556T (en) | 2013-06-21 | 2017-07-13 | Fraunhofer Ges Forschung | Method and apparatus for obtaining spectrum coefficients for a replacement frame of an audio signal, audio decoder, audio receiver and system for transmitting audio signals |
US10296160B2 (en) | 2013-12-06 | 2019-05-21 | Apple Inc. | Method for extracting salient dialog usage from live data |
US10170123B2 (en) | 2014-05-30 | 2019-01-01 | Apple Inc. | Intelligent assistant for home automation |
US9633004B2 (en) | 2014-05-30 | 2017-04-25 | Apple Inc. | Better resolution when referencing to concepts |
US9430463B2 (en) | 2014-05-30 | 2016-08-30 | Apple Inc. | Exemplar-based natural language processing |
EP3480811A1 (en) | 2014-05-30 | 2019-05-08 | Apple Inc. | Multi-command single utterance input method |
US9818400B2 (en) | 2014-09-11 | 2017-11-14 | Apple Inc. | Method and apparatus for discovering trending terms in speech requests |
US9668121B2 (en) | 2014-09-30 | 2017-05-30 | Apple Inc. | Social reminders |
US10127911B2 (en) | 2014-09-30 | 2018-11-13 | Apple Inc. | Speaker identification and unsupervised speaker adaptation techniques |
US10074360B2 (en) | 2014-09-30 | 2018-09-11 | Apple Inc. | Providing an indication of the suitability of speech recognition |
US10152299B2 (en) | 2015-03-06 | 2018-12-11 | Apple Inc. | Reducing response latency of intelligent automated assistants |
US9886953B2 (en) | 2015-03-08 | 2018-02-06 | Apple Inc. | Virtual assistant activation |
US9721566B2 (en) | 2015-03-08 | 2017-08-01 | Apple Inc. | Competing devices responding to voice triggers |
WO2016142002A1 (en) * | 2015-03-09 | 2016-09-15 | Fraunhofer-Gesellschaft Zur Foerderung Der Angewandten Forschung E.V. | Audio encoder, audio decoder, method for encoding an audio signal and method for decoding an encoded audio signal |
US10083688B2 (en) | 2015-05-27 | 2018-09-25 | Apple Inc. | Device voice control for selecting a displayed affordance |
US11025565B2 (en) | 2015-06-07 | 2021-06-01 | Apple Inc. | Personalized prediction of responses for instant messaging |
US10049668B2 (en) | 2015-12-02 | 2018-08-14 | Apple Inc. | Applying neural network language models to weighted finite state transducers for automatic speech recognition |
WO2017153299A2 (en) * | 2016-03-07 | 2017-09-14 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Error concealment unit, audio decoder, and related method and computer program fading out a concealed audio frame out according to different damping factors for different frequency bands |
US10049663B2 (en) | 2016-06-08 | 2018-08-14 | Apple, Inc. | Intelligent automated assistant for media exploration |
US10586535B2 (en) | 2016-06-10 | 2020-03-10 | Apple Inc. | Intelligent digital assistant in a multi-tasking environment |
DK201670540A1 (en) | 2016-06-11 | 2018-01-08 | Apple Inc | Application integration with a digital assistant |
DK179415B1 (en) | 2016-06-11 | 2018-06-14 | Apple Inc | Intelligent device arbitration and control |
US10474753B2 (en) | 2016-09-07 | 2019-11-12 | Apple Inc. | Language identification using recurrent neural networks |
US10043516B2 (en) | 2016-09-23 | 2018-08-07 | Apple Inc. | Intelligent automated assistant |
US11281993B2 (en) | 2016-12-05 | 2022-03-22 | Apple Inc. | Model and ensemble compression for metric learning |
US11204787B2 (en) | 2017-01-09 | 2021-12-21 | Apple Inc. | Application integration with a digital assistant |
US10417266B2 (en) | 2017-05-09 | 2019-09-17 | Apple Inc. | Context-aware ranking of intelligent response suggestions |
DK201770383A1 (en) | 2017-05-09 | 2018-12-14 | Apple Inc. | User interface for correcting recognition errors |
US10395654B2 (en) | 2017-05-11 | 2019-08-27 | Apple Inc. | Text normalization based on a data-driven learning network |
US10726832B2 (en) | 2017-05-11 | 2020-07-28 | Apple Inc. | Maintaining privacy of personal information |
DK201770428A1 (en) | 2017-05-12 | 2019-02-18 | Apple Inc. | Low-latency intelligent automated assistant |
US11301477B2 (en) | 2017-05-12 | 2022-04-12 | Apple Inc. | Feedback analysis of a digital assistant |
US10311144B2 (en) | 2017-05-16 | 2019-06-04 | Apple Inc. | Emoji word sense disambiguation |
US20180336275A1 (en) | 2017-05-16 | 2018-11-22 | Apple Inc. | Intelligent automated assistant for media exploration |
US10403278B2 (en) | 2017-05-16 | 2019-09-03 | Apple Inc. | Methods and systems for phonetic matching in digital assistant services |
US10657328B2 (en) | 2017-06-02 | 2020-05-19 | Apple Inc. | Multi-task recurrent neural network architecture for efficient morphology handling in neural language modeling |
US10628165B2 (en) * | 2017-08-17 | 2020-04-21 | Agora Lab, Inc. | Gain control for multiple description coding |
WO2019053228A1 (en) * | 2017-09-15 | 2019-03-21 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Communication apparatus, method and cellular network usable in a localization of a user equipment using a phase estimate |
US10445429B2 (en) | 2017-09-21 | 2019-10-15 | Apple Inc. | Natural language understanding using vocabularies with compressed serialized tries |
US10755051B2 (en) | 2017-09-29 | 2020-08-25 | Apple Inc. | Rule-based natural language processing |
US10636424B2 (en) | 2017-11-30 | 2020-04-28 | Apple Inc. | Multi-turn canned dialog |
US10733982B2 (en) | 2018-01-08 | 2020-08-04 | Apple Inc. | Multi-directional dialog |
US10733375B2 (en) | 2018-01-31 | 2020-08-04 | Apple Inc. | Knowledge-based framework for improving natural language understanding |
US10789959B2 (en) | 2018-03-02 | 2020-09-29 | Apple Inc. | Training speaker recognition models for digital assistants |
US10592604B2 (en) | 2018-03-12 | 2020-03-17 | Apple Inc. | Inverse text normalization for automatic speech recognition |
US10818288B2 (en) | 2018-03-26 | 2020-10-27 | Apple Inc. | Natural assistant interaction |
US10909331B2 (en) | 2018-03-30 | 2021-02-02 | Apple Inc. | Implicit identification of translation payload with neural machine translation |
US11145294B2 (en) | 2018-05-07 | 2021-10-12 | Apple Inc. | Intelligent automated assistant for delivering content from user experiences |
US10928918B2 (en) | 2018-05-07 | 2021-02-23 | Apple Inc. | Raise to speak |
US10984780B2 (en) | 2018-05-21 | 2021-04-20 | Apple Inc. | Global semantic word embeddings using bi-directional recurrent neural networks |
DK201870355A1 (en) | 2018-06-01 | 2019-12-16 | Apple Inc. | Virtual assistant operation in multi-device environments |
DK179822B1 (en) | 2018-06-01 | 2019-07-12 | Apple Inc. | Voice interaction at a primary device to access call functionality of a companion device |
US10892996B2 (en) | 2018-06-01 | 2021-01-12 | Apple Inc. | Variable latency device coordination |
DK180639B1 (en) | 2018-06-01 | 2021-11-04 | Apple Inc | DISABILITY OF ATTENTION-ATTENTIVE VIRTUAL ASSISTANT |
US11386266B2 (en) | 2018-06-01 | 2022-07-12 | Apple Inc. | Text correction |
US11076039B2 (en) | 2018-06-03 | 2021-07-27 | Apple Inc. | Accelerated task performance |
WO2020164751A1 (en) * | 2019-02-13 | 2020-08-20 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Decoder and decoding method for lc3 concealment including full frame loss concealment and partial frame loss concealment |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5673363A (en) * | 1994-12-21 | 1997-09-30 | Samsung Electronics Co., Ltd. | Error concealment method and apparatus of audio signals |
US5752225A (en) * | 1989-01-27 | 1998-05-12 | Dolby Laboratories Licensing Corporation | Method and apparatus for split-band encoding and split-band decoding of audio information using adaptive bit allocation to adjacent subbands |
US5781888A (en) * | 1996-01-16 | 1998-07-14 | Lucent Technologies Inc. | Perceptual noise shaping in the time domain via LPC prediction in the frequency domain |
US6351728B1 (en) * | 1991-04-05 | 2002-02-26 | Starguide Digital Networks, Inc. | Error concealment in digital transmissions |
US6418408B1 (en) * | 1999-04-05 | 2002-07-09 | Hughes Electronics Corporation | Frequency domain interpolative speech codec system |
US6421802B1 (en) * | 1997-04-23 | 2002-07-16 | Fraunhofer-Gesellschaft Zur Forderung Der Angewandten Forschung E.V. | Method for masking defects in a stream of audio data |
US6597961B1 (en) * | 1999-04-27 | 2003-07-22 | Realnetworks, Inc. | System and method for concealing errors in an audio transmission |
US6904404B1 (en) * | 1996-07-01 | 2005-06-07 | Matsushita Electric Industrial Co., Ltd. | Multistage inverse quantization having the plurality of frequency bands |
US7003448B1 (en) * | 1999-05-07 | 2006-02-21 | Fraunhofer-Gesellschaft Zur Foerderung Der Angewandten Forschung E.V. | Method and device for error concealment in an encoded audio-signal and method and device for decoding an encoded audio signal |
-
2004
- 2004-12-15 US US11/011,780 patent/US7356748B2/en active Active
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5752225A (en) * | 1989-01-27 | 1998-05-12 | Dolby Laboratories Licensing Corporation | Method and apparatus for split-band encoding and split-band decoding of audio information using adaptive bit allocation to adjacent subbands |
US6351728B1 (en) * | 1991-04-05 | 2002-02-26 | Starguide Digital Networks, Inc. | Error concealment in digital transmissions |
US5673363A (en) * | 1994-12-21 | 1997-09-30 | Samsung Electronics Co., Ltd. | Error concealment method and apparatus of audio signals |
US5781888A (en) * | 1996-01-16 | 1998-07-14 | Lucent Technologies Inc. | Perceptual noise shaping in the time domain via LPC prediction in the frequency domain |
US6904404B1 (en) * | 1996-07-01 | 2005-06-07 | Matsushita Electric Industrial Co., Ltd. | Multistage inverse quantization having the plurality of frequency bands |
US6421802B1 (en) * | 1997-04-23 | 2002-07-16 | Fraunhofer-Gesellschaft Zur Forderung Der Angewandten Forschung E.V. | Method for masking defects in a stream of audio data |
US6418408B1 (en) * | 1999-04-05 | 2002-07-09 | Hughes Electronics Corporation | Frequency domain interpolative speech codec system |
US6597961B1 (en) * | 1999-04-27 | 2003-07-22 | Realnetworks, Inc. | System and method for concealing errors in an audio transmission |
US7003448B1 (en) * | 1999-05-07 | 2006-02-21 | Fraunhofer-Gesellschaft Zur Foerderung Der Angewandten Forschung E.V. | Method and device for error concealment in an encoded audio-signal and method and device for decoding an encoded audio signal |
Cited By (20)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10283125B2 (en) | 2006-11-24 | 2019-05-07 | Samsung Electronics Co., Ltd. | Error concealment method and apparatus for audio signal and decoding method and apparatus for audio signal using the same |
US8219393B2 (en) * | 2006-11-24 | 2012-07-10 | Samsung Electronics Co., Ltd. | Error concealment method and apparatus for audio signal and decoding method and apparatus for audio signal using the same |
US9704492B2 (en) | 2006-11-24 | 2017-07-11 | Samsung Electronics Co., Ltd. | Error concealment method and apparatus for audio signal and decoding method and apparatus for audio signal using the same |
US9373331B2 (en) | 2006-11-24 | 2016-06-21 | Samsung Electronics Co., Ltd. | Error concealment method and apparatus for audio signal and decoding method and apparatus for audio signal using the same |
US20080126096A1 (en) * | 2006-11-24 | 2008-05-29 | Samsung Electronics Co., Ltd. | Error concealment method and apparatus for audio signal and decoding method and apparatus for audio signal using the same |
US20150279380A1 (en) * | 2006-11-30 | 2015-10-01 | Samsung Electronics Co., Ltd. | Frame error concealment method and apparatus and error concealment scheme construction method and apparatus |
US10325604B2 (en) | 2006-11-30 | 2019-06-18 | Samsung Electronics Co., Ltd. | Frame error concealment method and apparatus and error concealment scheme construction method and apparatus |
US9858933B2 (en) | 2006-11-30 | 2018-01-02 | Samsung Electronics Co., Ltd. | Frame error concealment method and apparatus and error concealment scheme construction method and apparatus |
US9478220B2 (en) * | 2006-11-30 | 2016-10-25 | Samsung Electronics Co., Ltd. | Frame error concealment method and apparatus and error concealment scheme construction method and apparatus |
EP2112653A1 (en) * | 2007-05-24 | 2009-10-28 | Panasonic Corporation | Audio decoding device, audio decoding method, program, and integrated circuit |
EP2112653A4 (en) * | 2007-05-24 | 2013-09-11 | Panasonic Corp | Audio decoding device, audio decoding method, program, and integrated circuit |
TWI466102B (en) * | 2008-06-13 | 2014-12-21 | Nokia Corp | Method and apparatus for error concealment of encoded audio data |
US8397117B2 (en) * | 2008-06-13 | 2013-03-12 | Nokia Corporation | Method and apparatus for error concealment of encoded audio data |
KR101228165B1 (en) | 2008-06-13 | 2013-01-30 | 노키아 코포레이션 | Method and apparatus for error concealment of encoded audio data |
US20100115370A1 (en) * | 2008-06-13 | 2010-05-06 | Nokia Corporation | Method and apparatus for error concealment of encoded audio data |
US8463293B2 (en) | 2009-09-21 | 2013-06-11 | Andrew Llc | System and method for a high throughput GSM location solution |
US8340683B2 (en) * | 2009-09-21 | 2012-12-25 | Andrew, Llc | System and method for a high throughput GSM location solution |
US20110070892A1 (en) * | 2009-09-21 | 2011-03-24 | Andrew Llc | System and method for a high throughput gsm location solution |
US10784988B2 (en) | 2018-12-21 | 2020-09-22 | Microsoft Technology Licensing, Llc | Conditional forward error correction for network data |
US10803876B2 (en) * | 2018-12-21 | 2020-10-13 | Microsoft Technology Licensing, Llc | Combined forward and backward extrapolation of lost network data |
Also Published As
Publication number | Publication date |
---|---|
US7356748B2 (en) | 2008-04-08 |
US20050163234A1 (en) | 2005-07-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US7356748B2 (en) | Partial spectral loss concealment in transform codecs | |
EP1697927B1 (en) | Improved frequency-domain error concealment | |
US5701390A (en) | Synthesis of MBE-based coded speech using regenerated phase information | |
US5754974A (en) | Spectral magnitude representation for multi-band excitation speech coders | |
US5226084A (en) | Methods for speech quantization and error correction | |
US6377916B1 (en) | Multiband harmonic transform coder | |
KR101376762B1 (en) | Method for trained discrimination and attenuation of echoes of a digital signal in a decoder and corresponding device | |
US8050933B2 (en) | Audio coding system using temporal shape of a decoded signal to adapt synthesized spectral components | |
US8315860B2 (en) | Interoperable vocoder | |
US8843798B2 (en) | Frame error concealment method and apparatus and decoding method and apparatus using the same | |
US20070156397A1 (en) | Coding equipment | |
EP0927988A2 (en) | Encoding speech | |
US6732075B1 (en) | Sound synthesizing apparatus and method, telephone apparatus, and program service medium | |
KR20160018497A (en) | Device and method for bandwidth extension for audio signals | |
US7363216B2 (en) | Method and system for parametric characterization of transient audio signals | |
WO2020169754A1 (en) | Methods for phase ecu f0 interpolation split and related controller | |
ZA200603725B (en) | Improved frequency-domain error concealment | |
KR100220783B1 (en) | Speech quantization and error correction method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AS | Assignment |
Owner name: TELEFONAKTIEBOLAGET LM ERICSSON (PUBL), SWEDEN Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:TALEB, ANISSE;REEL/FRAME:016442/0198 Effective date: 20031221 |
|
STCF | Information on status: patent grant |
Free format text: PATENTED CASE |
|
FPAY | Fee payment |
Year of fee payment: 4 |
|
CC | Certificate of correction | ||
FPAY | Fee payment |
Year of fee payment: 8 |
|
MAFP | Maintenance fee payment |
Free format text: PAYMENT OF MAINTENANCE FEE, 12TH YEAR, LARGE ENTITY (ORIGINAL EVENT CODE: M1553); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY Year of fee payment: 12 |