US 3885110 A
There are disclosed apparatus and methods for reproducing sounds, such as musical notes, tones and chords, words, voice phonemes or the like, by recording sounds or notes from actual sources and which, upon command, can be instantaneously retrieved for reproduction in any desired sequence and tempo. In accordance with the invention, instantaneous retrieval is accomplished by establishing a predetermined finite initial portion of each recorded note or sound within which the retrieval process can be begun without deleterious loss of fidelity, thereby obviating the costly and in some cases unachievable process of having each commanded sound begin at the precise beginning of the sound track.
Description (OCR text may contain errors)
United States Patent Di Matteo APPARATUS AND METHOD FOR ARTIFICIALLY REPRODUCING AT WILL SOUNDS FROM UNIQUE SOURCES 179/1001 R, 100.1 C, 100.1 PS, 100.3 B, 100.3 D, 15 A, 6 C, 6 R, 6 TA, 100.4 D; 84/1.18
 References Cited UNITED STATES PATENTS 3,496,299 2/1970 Ternmer 179/6 C MODULATOR 20 [451 May 20, 1975 Primary Examiner-Raymond F. Cardillo, Jr. Attorney, Agent, or Firm-Eisenman, Allsopp & Strack 57 ABSTRACT There are disclosed apparatus and methods for reproducing sounds, such as musical notes, tones and chords, words, voice phonemes or the like, by record ing sounds or notes from actual sources and which, upon command, can be instantaneously retrieved for reproduction in any desired sequence and tempo. In accordance with the invention, instantaneous retrieval is accomplished by establishing a predetermined finite initial portion of each recorded note or sound within which the retrieval process can be begun without deleterious loss of fidelity, thereby obviating the costly and in some cases unachievable process of having each commanded sound begin at the precise beginning of the sound track.
8 Claims, 5 Drawing Figures l r 210 I 1 21b 2lr 71 l 21d 1 LIGHT LIGHT LIGHT LI HT SOURCE SOURCE SOURCE SOSRCE l I] 111 II DISC 12b REZFJERENCE C n t,
END OF NOTE sum or NOTE 1 COLLECTION LENS Pf IENTEB MY 2 O ISFS SHEEI 1 BF 2 FIG.
PIITENTEU HAY201975 3,885,110
SHEET 2 OF 2 KEYBOARD LIGHT /22 MODULATOR 20 SELECTOR r 2Ia 21b 2107 2Id LIGHT LIGHT LIGHT LIGHT SOURCE SOURCE SOURCE SOURCE I 11 111 I1 DISC 12b REFERENCE END OF NOTE START OF NOTE cOLLEcTION LENS LIGHT sENsITIvE PICKUP 27 26 I FIG.3
APPARATUS AND METHOD FOR ARTIFICIALLY REPRODUCING AT WILL SOUNDS FROM UNIQUE SOURCES RELATED APPLICATIONS This application is a continuation-in-part of a copending application Ser. No. 245,484, filed Apr. 19, 1972, now abandoned.
BACKGROUND OF THE INVENTION The invention is concerned with apparatus and methods for reproducing instantaneously on command recorded sounds, such as recorded, discrete notes of musical instruments, voice phonemes or the like, without objectionable loss of fidelity.
The artificial reproduction of sound patterns, such as music as played by a musical instrument, human speech and the like, has been achieved in the past through the use of synthesizing or simulation techniques based on tone generators which'are inherently deficient in their ability to match the true source. It is primarily because of the lack of time varying complex harmonic and amplitude characteristics, which inhere in the true or actual sound which emanates from original sources, that the synthesized sounds lack realism. An obvious solution to the problem is to utilize recording techniques in which, for example, sets of actual notes from an instrument are pre-recorded for retrieval on command in any desired order. While good results are theoretically achievable, the complexity of the equipment necessary to retrieve a recorded sound from storage instantaneously upon command has been heretofore thought to exceed practical limits. For example, the notes from a piano might last from 1 to seconds, and the problem of locating the precise beginning of each recorded track in a dynamic system so that the note will occur immediately upon command has been all but unachievable as a practical matter.
SUMMARY OF THE INVENTION The retrieval of stored or recorded sound information has been greatly simplified in accordance with the present invention by the discovery that a part of the initial portion of a recorded sound from a musical instrument can be omitted without perceptible qualitative loss of fidelity, i.e. loss perceptible to a listener. Thus, the reproduction of a musical note can be started at some time later than the actual start of the sound track of that note with no deleterious loss of fidelity.
If, for example, a single note is recorded on a circular and, therefore, endless loop sound track, in which the note begins at one point in the loop and varies in amplitude and harmonic content to substantially the vanishing point at the end of the loop, conventional practice dictates that retrieval on command begin by establishing a scanning and pickup function which coincides precisely with the front end. To locate and reproduce the note beginning exactly at the front end instantaneously on command would require highly complex equipment.
In accordance with the present invention, it has been determined that, as a result of laboratory tests, starting the reproduction at a point spaced, i.e. delayed, significantly from the actual recorded beginning of the note will result in no perceptible loss of fidelity. This delay, which represents an information access tolerance for purposes of reproduction, can be determined as a function of carrier frequency and envelope modulation rate. Thus, assuming a circular sound track recorded, for example, on a rotating disc, pickups can be disposed at a number of relatively widely circumferentially spaced points around the track. Six pickups might be provided around a sound track representing a 4- second note from a piano. That pickup which is closest to the front end of the note is then selectively energized when the sound is to be reproduced. All of the notes from any desired number of different instruments can be recorded and called up for reproduction by means such as a keyboard, and the notes can be reproduced singularly or in combination under the control of the operator, and modulating and timing can be subsequently applied to achieve characteristics closely resembling those of the original instrument.
BRIEF DESCRIPTION OF THE DRAWINGS 2B is a plan view ofa record of modified design having two separate sounds recorded in circumferentially spaced relationship in one endless circular track; and
FIG. 3 is a diagrammatic representation of a system for reproducing the sound of FIG. 1.
DESCRIPTION OF THE PREFERRED EMBODIMENT Referring to FIG. 1, there is illustrated in diagrammatic and simplified form an actual sound, such as a note from a musical instrument, the sound being represented by a sine wave 10 which decays or attenuates with time to define an envelope II. The sine wave, in addition to carrying the fundamental frequency of the sound, has imposed thereon harmonies and overtones which (not shown in FIG. 1) distinguish, for example, the quality of a note from a musical instrument and an artificially generated sound at the same frequency. The envelope 11 has a fast-rise or swell portion Ila and an attenuation or decay portion Ilb. The duration of the sound to a point approaching inaudibility is represented by T, the beginning of the sound by t and the end by 1 To reproduce the sound, the sound track or record is normally scanned from the beginning to the end and, if a shorter duration note is desired in a reproduction system, it is typically clipped at the end in order to insure the reproduction of the front end portion which is richest in harmonics. In synthesized sound instruments such, for example, as electric organs, it has not heretofore been practical to utilize recorded notes from pipe organs. In lieu of such original sounds, electric organs typically utilize tone generators which are at best poor approximations of a true instrument because the com plete, complex time-varying phase and amplitude characteristics of the harmonics and overtones can not readily be synthesized. The concept of using actual recorded sounds as the sources in an instrument is in theory the best way to achieve realism. As a practical matter. however. it has been all but impossible to achieve because of accessing difficulties. It is difficult and costly to start a reproduction of a recorded sound instantaneously upon command beginning at the precise beginning of the record. If the record is continuously moving, pickups would be required at all points along the recorded track and the one at the beginning would be activated when the note was to be instantaneously called forth. If a non-continuous scanning system were to be used, it would be necessary upon command to start the scanning motion and hence the sound reproduction from a standing start. A time lag occurs before full fidelity reproduction is achieved and the instrument can not be played in the manner of a conventional instrument with instantaneous response.
In accordance with the present invention. it has been discovered that a substantial definable portion of the front end of the note can be eliminated without perceptible loss of fidelity. Thus, referring to FlG. l for exam ple, reproduction can be begun at any point between 1,, and r, without degrading to the ears of the listener the quality of the reproduced sound. Given a wide tolerance in the point at which the reproduction of the sound can be begun. it is possible to build. within practical cost limits. a note-selecting and reproducing system. A typical system is illustrated by FIG. 3 of the drawing. and will be described in detail below.
To determine the point t, in accordance with the in vention. the note should be analyzed as to both its car rier rate and its envelope rate. The carrier rate is the fundamental frequency of the note. and the envelope rate is determined by analyzing the frequency spectrum of the sound envelope. The sinusoidal components of the envelope are analyzed with the highest component j}. which might be. for example. 10 cycles per second. being determinative of the maximum envelope rate. This can be otherwise expressed as the highest Fourier component in the envelope. It has been determined experimentally that a period of up to one cycle of the maximum envelope frequency or 0.1 seconds in the frequency example given, can be omitted at the front end of the signal without loss of quality to the listener. The maximum period is represented by one over the highest frequency (l/f in the envelope and. if this factor is multiplied by one'tenth. an interval is determinable at the front end of the note in which virtually perfect reproduction occurs regardless of the precise starting point within that interval. Changes in the quality of the note begin to become apparent at 100% of the maid mum period. Thus, for certain purposes, the access pe riod at the front end of the note can be varied consider ably.
As an example of how a working system can be built around this discovery. actual sounds from original sources. such as the notes of musical instruments. can be recorded on a rotary film disc 12 (FIGS. 2. 2A and 3). A plurality of concentric sound tracks 12a and 12h 121 can be recorded on the disc using a variable area technique. A variable area technique consists of exposing a film emulsion in a track such that a transparent area proportional to the amplitude of the signal plus a bias proportional to one-half the peak-topcak amplitude of the signal is developed within the track width. Thus. the sine wave is reduced to a symmetrically tlf1 (ill dulating track band of varying width. A typical system to effect such recording is illustrated in FIG. 2, in which the disc 12 is rotated in the presence of a collimated light source 13 which is focused through a slotted mask l4 onto the sound track. Eclipsing the slot 14a in the mask 14 is a delta-shaped vane 15 driven transversely of the slot by a galvonometer 16 energized by an amplitier ]7 to which the output of the recording microphone 18 is connected. The recording is arranged so that the beginning of the sound for each track is discretely located and continues circumferentially around the disc to terminate at a point which can be close to the beginning (for the note of longest duration on the record). The location of the beginning, the end and of the intermediate points in the sound track is controlled by an address track 12 which can be located, for example. at the periphery of the record in the form of a digital magnetic recording dividing the record into a large number of discrete sections. Address tracks per se are well known in the art and need not, therefore. be described in greater detail herein. If preferred. in accordance with the invention, more than one note can be located on one circumferential track dependent upon the combined note durations. Such arrangement is shown in FIG. 2B in which the outermost sound track is divided into two sections 12ml and l2a-2 each of slightly less than 180 of are but disposed on the same circle. The address track 12' is correspondingly arranged to indicate either by a special recorded indicator or by means counting the precise beginnings of each of the several discrete sounds recorded in circumferentially spaced relationship around the track. In order to avoid dwells or delays in accessing the desired sounds at the time of playback, additional pickups are used in the playback system. all as described below.
In a typical system. the disc might include recorded thereon all of the notes of many musical instruments including. for example. a piano. When played back in a system shown schematically in FIG. 3. the sound of the piano embodying all of the quality of the original instrument can be achieved. Upon playback. the light that passes through the respective channels or sound tracks on the recording disc 12 will be proportional to the clear or transparent area in the track and, as the disc rotates at constant speed, the light from each channel will describe the signal. The signal from any given track is generated by actuating the appropriate light source associated with the track. In accordance with the invention, a plurality of light sources are associated with each track spaced circumferentially around the disc. with the number of light sources being such that the spacing represents between one-tenth of the envelope rate and the envelope rate of the signal.
The complete system can include an input in the form. for example. of a keyboard 19, with each key comprising an electrical switch. The command signal output of the keyboard can be connected through a modulator 20, the output of which is connected to a plurality of light sources 21a, 21b. 21c and 21d. Also connected to each of the light sources is a light-selector circuit 22 under the control of a reference pickup 23 from the disc 12. The reference pickup means 23, the address track and the light selector 22 are well-known in the art and can take the form, for example. of a magnetic sine wave or sequence of pulse generating indicia recorded on the periphery of the disc and thus movable in synchronism therewith so that. as the disc rotates.
signals are generated in the pickup coil which indicate the precise beginning of the recorded sounds on the sound tracks and which are counted to indicate how far the disc has rotated from the beginning of the sound. The pickup signals also operate the switching or logic circuitry of the light selector 22. With the disc rotating at a constant speed by a drive motor (not shown), when the operator selects at the keyboard a note corresponding to the outer track on the disc, the reference pickup 23 will immediately inform the light selector 22 as to the location of the start of the note on the disc, and that light source which is closest to the starting point will be energized. Thus, in the illustrated arrangement, the light source 21a will be energized and, because its spacing is within the defined limits of the beginning of the track 12a, high fidelity reproduction will result. Using the record of FIG. 28, it will be understood that for equivalent performance, the number oflight sources 21 around the circumference of the track will be doubled, i.e. multiplied by the number of discrete sounds on the track.
The system is completed by means of a collection lens 24 which focuses the light passing through one or more of the sound tracks onto a light-sensitive pickup 25 which transduces the signal into an electronic out put which is passed through an amplifier 26 to a speaker 27. Special amplitude effects can be accomplished by varying the intensity of the light source illuminating the track, this being accomplished by means of the modulator under the control of the operator.
It will be understood that a wide range of optical techniques can be used for focusing and collecting the light transmitted through the sound tracks, including, for example, fiberoptics, prisms, lenses or the like. Also, inputs to the control system can be achieved through keyboard switches, foot pedal devices, auto matic controls or, in general, any device which provides a discrete signal representative of a desired sound track or note. Also, while the invention has been described having reference in particular to a musical instrument, such as a piano, virtually all musical instru ments are capable of being incorporated in the system. The system can also be used to achieve other sound re productions including, for example, the human voice.
The invention should not, therefore, be regarded as limited except as defined in the following claims:
1. In a method for reproducing instantaneously on command recorded sound generated by a unique source including musical notes, vocal phonemes, words and the like, in which the recorded sound envelope highest frequency, as represented by f,,, is low as compared to the recorded sound carrier fundamental frequency ff, the steps of commanding the pickup of the desired sound, picking up a signal representing the recorded sound substantially instantaneously by beginning at any random point in the record within the range from the actual beginning of the sound to a subsequent point substantially proximate to an interval defined by the expression l/f, where f is the highest frequency component in the frequency spectrum in the recorded sound envelope, and converting the signal to an audible sound.
2. The method of creating compositions of sound such as musical compositions, human speech and the like, comprising the steps of establishing a reproducible record of each of the component parts of the sound composition in the form of notes or phonemes from a true source in which the character of each sound varies with time and has an envelope rate which is slow as compared to the carrier rate, selecting desired sounds by substantially instantaneously beginning reproduction of the selected sounds at any random point within a range from the precise beginning of the record for a distance corresponding substantially to l/f,., where f is the highest frequency component in the recorded sound envelope.
3. The method of creating compositions of sound such as musical compositions, human speech and the like, comprising the steps of establishing a reproducible record of each of the component parts of the sound composition in the form of notes or phonemes from a true source in which the character of each sound varies with time and has an envelope rate which is slow as compared to the carrier rate, continuously scanning each record from beginning to end at a plurality of points spaced apart along its length by a predetermined distance corresponding substantially to l/f, where f,, is the highest frequency component in the recorded sound envelope, commanding the desired sequences and combinations of sounds at desired rates, and instantaneously selecting in response to any command the scanning point which is closest to the beginning of the selected sound in a direction opposite to the direction of scanning to derive signals representative of the sound, and converting the signals to audible sound.
4. Apparatus for recreating sound compositions from recorded sound components, comprising, record means carrying discrete data representative of individual sounds from true sources, command means to initiate reproduction of any desired sound or combination of sounds from the record, scanning means instantaneously responsive to the command means to pick up signals from the record beginning randomly at any point from the precise beginning of the recorded sound to a later point which is substantially defined by the expression l/f, where f is the highest frequency component of the envelope rate of the recorded sound.
5. Apparatus as set forth in claim 4, said scanning means including a plurality of scanners related to the record means to afford output signals which correspond to spaced apart points in the original sound, with the spacing being not substantially in excess of a distance representing the envelope rate.
6. Apparatus as set forth in claim 5, said scanning means being responsive to the command means to derive signals from the scanner which is closest to the beginning of the recorded sound in a direction opposite to the direction of scan.
7. Apparatus as set forth in claim 6, the record means for the recorded sound components comprising a single sound track for each component, said plurality of scanners being spaced apart along the sound track.
8. Apparatus as set forth in claim 6, said record