Electronic music synthesis for recordings

Electronic music synthesisfor recordings

Employed in the production of practically every form of recordedmusic today, electronic music synthesis in some way affects95 percent of the multibillion-dollar recording industry

Harry F. Olson RCA Laboratories

Electronic music synthesis is a process whereby sev- tion, combination of existing tones, or the production oferal parts or elements of a musical composition that tones by manual or programmed synthesis, the founda-are performed or produced as separate entities are tion for all synthesis involves the basic physical propertiescombined to form the entire or rendered composition. of a musical tone-frequency, intensity, waveform, andSince the entire process can seldom be carried out time. A more meaningful and useful description of ain real time, the synthesis of music has not been a musical tone can be given in terms of frequency, intensity,performing-type rendition; therefore, the final product growth, steady state, decay, duration, portamento,takes the form of a record that can be reproduced at timbre (spectrum), vibrato, and deviation. These proper-any given time. Electronic music synthesis includes ties of a musical tone are depicted in Fig. 1.the modification and combination of conventional and The following definitions of tone properties are descrip-original sound sources, manual and programmed tive rather than absolutely rigorous presentations of theelectronic music synthesizers, and digital computers. formal and somewhat abstruse language of the standards.

Frequency of a sound wave is the number of cycles oc-curring per unit of time, measured in hertz. The subjec-tive counterpart of frequency is pitch.

To dispel the confusion that sometimes arises with Intensity of a sound wave is energy transmitted perrespect to the definition of electronic music synthesis, unit of time. The intensity is usually expressed in decibelsthe author would like to state that the term is not synony- above the threshold of hearing at 1000 Hz, which ismous with music that is produced by an electronic music 0 dB = 10-11 W/cm2. The subjective counterpart ofsynthesizer. In the process of electronic music synthesis, intensity is loudness.the individual elements of a particular music score are Growth is the time required for a sound to build up toperformed separately by independent means and later some fraction of the ultimate value.combined electronically to form the complete composi- Steady state of a sound is the length of time in whichtion. With this definition in mind, it should come as no there is no change in the intensity.surprise that almost all popular and contemporary Decay is the time required for sound to fall to some("rock") music that is recorded today is in fact electron- fraction of the original value. Note that growth, steadyically synthesized. state, and decay lumped together become the envelope

of a tone.Properties of a musical tone'.2 Duration is the length of time that a sound persistsThe ultimate objective of electronic music synthesis without interruption or discontinuity in the output.

is the production of desired musical tones. In electronic Portamnento is a uniform glide in frequency from amusic synthesis, whether the process involves modifica- sound of one frequency to a sound of another frequency.

18 IEEE spectrum APRIL 1971

Frequency Portamento is also termed a frequency glide.(pitch) A complex sound wave is made up of the fundamental

30 100 1000 10 000 15 000 tone and overtones. The timbre or spectrum of a tone isFrequency, hertz expressed in the number, intensity, and phase relations

of the components; that is, the fundamental and over-

90I tones or partials.,ntensiy 90 _Vibrato is a low-frequency modulation of a musical(ioudness) 60- tone. This may result from either frequency modulation

30 _ or amplitude modulation or a combination of both.O In general, the modulation frequency is of the order of

7 Hz. Tremolo, a special case of vibrato, is created by< Duration amplitude modulation only.

Growth X Growth Steady state Decay Deviation is a departure from the regular and is one ofSteadystate the beautiful and artistic characteristics of some types ofDuration music.

/__________________________ With reference to the preceding descriptions, many ofTime, seconds the properties of a tone are interdependent. For example,

timbre is influenced by the attack, decay, portamento,vibrato, etc. When the properties of a tone as just defined

Portamento and depicted in Fig. I are specified, the tone can be com-U.

Time,_seconds pletely described. Furthermore, the tone can be producedTime, seconds from these specifications by providing electronic means

for generating its characteristic properties.Timbre Ii

100 1000 10 000Frequency, hertz

VibratonA . A A n nnA FIGURE 1. Properties of a musical tone.Vibrato ANA iWv|VJ V

Amplitude Frequencymodulation modulation

Deviation ' ' ' ' ' FIGURE 2. Perspective view of a recording studio depictingDLevLi LL.. LLLa. Vueb.r the sound pickup of eight musicians by means of separate

microphones and the monitoring room with director andrecording engineers operating the console and recorder.

Olo-letoi mscsntei frrcrdns1Talk-back

Monitoring \ Monto/in

Cogneesole Mntoig n

~~~~''/recordingroo

Director \/ ~~~~~~~~~~~~~~~Rcrdn

MloniEetorinicmscsnhssgo eodns

Sound Electronic music synthesis___ ___(S) (E) sources from conventional sound sources8-

Microphones Since the introduction of magnetic tape recording anduntil only a few years ago, master tape recordings wererecorded on two- and three-channel machines. The

J E)1(¾) (~6) (So)sun recording of original sounds on 8-, 16-, and 24-channelMicrophones machines is a recent innovation brought about by the re-

Amplifiers quirements of new sounds in popular and contemporarymusic in order to produce commercially successful

Volume records. The use of the 8-, 16-, and 24-channel magneticI I I I I I I Itape recorder makes it possible to record each instrument

N7 77 7 7Amplifiers or performer on a separate channel as depicted by the

M8(((i Volume perspective view of the studio, monitoring room, musi-indicators cians, conductor, and engineers in Fig. 2. A schematic

diagram of the elements used in the recording system isVolume shown in Fig. 3.controls The eight-channel system is used in this article for theAmplifiers sake of simplicity. In order to record each and every

Loudspeakers sound source, that is, record each instrument or per-former separately when playing as a group, there shouldbe a minimal acoustic crosstalk between the differentsound sources. To reduce the crosstalk, either bidirec-tional or unidirectional microphones are employed. In

8-channel addition, a studio with a very low reverberation time is0 magnetic tape used, and each sound source is fed to a separate track so

that the isolation is preserved. In turn, each sound sourceFIGURE 3. Schematic diagram of the recording of eight is monitored by means of volume indicators and loud-sound sources, each with a pickup microphone enablingthe outputs of the sound sources to be individually re- speakers. The use of a loudspeaker in each channel makescorded on eight separate tracks of a tape recorder. it possible to detect acoustic crosstalk and other un-

desirable effects.

FIGURE 4. The mixing, modifying, and conversion to two-channel stereophonic magnetic tape from the eight soundsources recorded in Figs. 2 and 3.

8-channel( 0} { o } magnetic tape

reproducer

Amplifiers

Volume controls

Tone modifiers

Amplifiers

M M M MMM M M M M M M M M M M ~~~~~~~~~~~~~~~~~~Volumeindicators

Isolating resistors

recorderm cntol


Recording each sound source on a separate track 3 (no delay) to anywhere between 3 and 5, dependingrepresents the ultimate for electronic music synthesis. upon the amount of delay. (A large delay will produceIn some performances, the recording of two or more location 5.) If the delay in the two channels is the samesound sources on one track may be desirable from the but the signal in the left channel is attenuated, then thestandpoint of simplifying the procedure of electronic sound source can be moved from location 3 (no attenua-music synthesis. In any event, the synthesis procedure tion) to anywhere between 3 and 5, depending upon theis essentially the same. amount of attenuation. (Location 5 results from con-A schematic diagram for electronic synthesizing of the siderable attenuation.) Employing both delay and at-

recording produced by the system of Figs. 2 and 3 is tenuation, therefore, places the sound source in anyshown in Fig. 4. In this particular case, the final product position at or between the loudspeakers. The attenuationis a two-channel stereophonic record. The system of Fig. and delay provide a powerful means for producing the4 provides a facility for placing any of the eight sound desired auditory perspective.sources in either the left or right channel or any mixture Let us return now to the tone modifier of Fig. 5: Thein both channels. The important element in this system frequency-response modifier consists of frequency-selec-is the tone modifier, which is diagramed in Fig. 5. There tive networks that change the original timbre (waveform)are 16 of these tone modifiers in the system of Fig. 4. by accentuating or attenuating frequency bands.

Signal delay in combination with signal level makes it The timbre modifier of Fig. 5 is somewhat similar topossible to specify the auditory perspective of any sound the frequency-response modifier save that new frequencysource, that is, place any sound source in reproduction components may be added.at either the right or left loudspeaker or any position The vibrato or tremolo generator modulates, by fre-between. The control of the auditory perspective is quency or amplitude modulation or a combination ofillustrated in Fig. 6. If the entire audio signal is fed to the both, the original audio signal, with a modulation fre-left loudspeaker and no audio signal to the right loud- quency of the order of 7 Hz.speaker, the sound source will appear to be located at Since the original sound as depicted in Figs. 2 and 3 isposition 1. In the same way, if the entire signal is fed recorded without any reverberation, a means must beto the right loudspeaker and no signal to the left loud- provided to introduce the effect artificially.speaker, the sound source will appear to be located at In general, the goal of all sound reproduction is to re-position 5. If the same audio signal is fed to both loud- duce nonlinear distortion. However, in some instancesspeakers, the sound source will appear to be located at the subjective response can be heightened by the intro-position 3. However, if delay is introduced in the left duction of an appropriate nonlinear distortion. Ac-channel but the same amplitude is fed to both loud- cordingly, means are provided for introducing such non-speakers, the sound source will be moved from location linear distortion.

The fuzz producer is a method of generating high har-monics and frequency-modified random noise. Here againis another example of the introduction ofwhat is normallyconsidered undesirable- noise-to heighten the artistic

FIGURE 5. An electronic modifier.

FIGURE 6. System of delayers and attenuators for placingthe apparent sound of an audio signal at any point be-tween two stereophonic loudspeakers.

Audiosignalinput

Attenuators

Delay Delay

Left Rightloudspeaker loudspeaker

Apparent location ofreproduced sound source

_ ..> . j~, NM .........Output 0

| L0.DiM.*.^S. 0 I , Listener

Olson-Electronic music synthesis for recordings 21

aspects of the recorded sound. to a certain input level. Beyond that level, the output doesThe volume compressor of Fig. 5 is an electronic system not increase. The limiter is extremely useful for protection

that reduces the amplification in a gradual manner when- against sudden overloads in the sound system.ever the signal input attains a certain level. Compressors An extension of the tone modifier is shown in Fig. 7.are often used to reduce the volume or amplitude range This particular modifier makes it possible to extend thein such systems as sound motion pictures, magnetic tape frequency range of an instrument and modify its timbre.and phonograph disk recording, sound broadcasting, The fundamental selector searches and selects the funda-and sound reinforcing. A reduction in the volume range mental frequency of the instrument. This fundamentalof radio, magnetic tape, and phonograph sound reproduc- frequency is then multiplied or divided to extend thetion makes it possible to reproduce the wide amplitude frequency range of the instrument or to produce morerange of orchestra music in the home without excessive than one instrument (as in Fig. 7). Under these condi-top levels. The use of the compressor improves the signal- tions, the original sound source or instrument suppliesto-noise ratio, thereby improving the intelligibility of the fundamental frequency (pitch), the envelope, andspeech and enhancing music sound reproduction when the intensity (loudness).the ambient noise is high. The systems of Figs. 5, 6, and 7 are very powerful toolsA limiter is similar to the compressor except that the for modifying the original tones and auditory perspective

relationship between the input and output is constant up of the instruments recorded in Figs. 2 and 3. The generalidea is to provide new sounds that will increase the sub-jective response to the ultimate product-the record-when it is reproduced.In the preceding descriptions, all of the performers and

Volume instruments played at the same time. Therefore, the en-controls tire unmodified composition was performed in real time.

However, in many cases, the vocalist is introduced afterthe instrumental part has been recorded. This method is

Input Outputillustrated in Fig. 8. The magnetic tape containing theinstrumental part is reproduced and fed to the vocalisteither by loudspeaker or by earphone. The vocalist thenperforms along with the instrumental music, and his voiceis recorded along with the instrumental part.The "overdub" process depicted in Fig. 8 is often re-

FIGURE 7. A system for selecting the fundamental fre- placed by the "sel-sync" method, in which the voice isquency of an audio signal, multiplying or dividing it by recorded on one track of a multiple track recorder whilewhole numbers, and injecting new timbre to produce sev-eral new instruments from the input of one instrument. the other tracks, which have been recorded previously,

are reproduced on the loudspeaker or earphones forsynchronization. In some recording procedures, the entireinstrumental group may not be recorded at the same time.

FIGURE 8. A method of combining and recording a vocalist In this case, the "overdub" or the "sel-sync" process iswith a recorded instrumental number. used to combine other instrumental parts at a later time.

Magnetic tape The system of Fig. 4 converts eight channels of originalreproducer sound to two channels of stereophonic sound. In the past

instrumental few years, practically all records have been produced inmagnetic tape the stereophonic form. For the production of mono-

phonic records, however, only one section of the systemLoudspeaher in Fig. 4 is used. Moreover, by the addition of two more

sections of eight channels each (as shown in Fig. 9), the

> Earphone eight channels may be allocated and recorded on four-Earphone channel quadriphonic sound. Reverberation envelope,

sound in motion, and other spatial effects may be ob-a"> , tained in quadriphonic sound.

Amplitiers Microphone Vocalist All of the conversion of the eight channels to mono-

phonic, stereophonic, and quadriphonic sound, includ-, Attenuators 5 ing the modifications, are carried out in a room simulat-

ing the acoustics of the average room in a residence. TheM Volume indicators loudspeaker arrangement for the three systems will be

described in the next section.The preceding exposition, together with Figs. 1 through

9, shows that the combined recording of instruments andperformers, along with the modifying processes, in fact

T 1 constitutes an electronic music synthesis in which theoriginal sounds supply the basic elements but the elec-tronic modifications supply the artistic embellishment so

Loudspeaker necessary for the production of a recording that willMagnetic tape receive a high order of consumer acceptance and, as arecorder result, become a commercially successful product.


8-channelmagnetic tapereproducer

_2 3 4 5- 6 78

I AmplifiersVvolume controlsTone ModifiersVolume controls

5 7 fj7 fj7Ampli Isolators

MastervolumecontrolsAmplifiers

Volumecontrols

> > ~~~~~~~~~4-channelmagnetic tape Amplifiers

recorderLoudspeakers

LF LR RF RR

FIGURE 9. The mixing, modifying, and conversion to four-channel quadriphonic sound from the eight sound sourcesrecorded in Figs. 2 and 3.

Stereophonic records can also be played on a mono-phonic reproducer, in which case the two channels arecombined into one channel. With the advent of quadri-

quadriphonic sound reproduction systems phonic sound, there will invariably be a shift in emphasisfrom the stereophonic form because of the many addi-

Systems for the reproduction of synthesized sound may tional spatial sound cffcctsbe classified as monophonic, stereophonic, and quadri-phonic. The three systems are described in Fig. 10. Synchronized magnetic tape recordersThe monophonic sound system is of the field typc em- In motion picture recording and in cases where a singleploying a single channel; the stereophonic sound system series of tones are produced by a programmed electronicis of the field type employing two channels; and the music synthesizer, the recording is made with a synchro-quadriphonic sound system is of the field type employing nized multichannel magnetic tape recorder. In most cases,four channels. ine magnetic tape is in the form of 35-mm magneticallyThe monophonic sound system can reproduce the coated film with sprocket holes as shown in Fig. 11. In

entire audio-frequency range with low noise and distor- this way, the magnetic tape recorder can be synchronizedtion. In turn, the stereophonic sound system can repro-duce the entire audio-frequency range with low noiseand distortion and in auditory perspective. Auditoryperspective provides the subjective location of the re-produced sound sources. Finally, the quadriphonic*sound system can reproduce the entire audio-frequency FIGURE 10. The auditory effects of monophonic, stereo-range with low noise and distortion, auditory perspective, phonic, and quadriphonic sound-reproducing systems.and a reverberation envelope and spatial effects. Such a Monophonic Stereophonic Q psystem may be employed to provide all manner of reproducer reproducer epacoustic spatial effects such as sound in motion aroundthe room or bouncing back and forth around the room.

Electronic music synthesis has been released as re-corded music in monophonic, stereophonic, and quadri-phonic form. However, in the past few years practicallyall records have been produced in stereophonic form.

*There has been no standardization of the designation for a four-channel sound reproduction system. However, the terms quadri-phonic and quadraphonic appear to be the most common in theliterature.


7 different audio inputs with any picture or program. The synchronized recorderis employed in all types of programmed electronic musicsynthesis. One design of the synchronized magnetic taperecorder is of the seven-channel type, where seven audioinputs are recorded on seven separate tracks (Fig. 11).

Amplifiers In Fig. 12(A), a magnetic tape with seven separateaudio signals recorded in the manner of Fig. 11 is repro-duced and the audio signals in the seven channels aremixed and rerecorded on a single track of a seven-track

Attenuators recorder. In turn, if each of the seven tracks of this re-corder is itself filled with seven different operations, thetotal result of the seven tracks if recorded as in Fig.12(A) represents a 49-tone sequence, as demonstrated in

Amplifiers Fig. 12(B). If the recorder in Fig. 12(B) is also capable ofseven channels and all the tracks are again filled, thetotal final result will represent a 343-tone sequence. It isobvious that this procedure can be repeated over and

.________,_:_-_::::- over again, producing tone sequences that are multiplesSprocket drive of seven.

Electronic music synthesis by the7-channel

generation and modification of original sounds6-9synchronized Electronic synthesis of music may be created by themagnetic tape generation and modification of all sorts of original

sounds, which was the first method used for the synthesisFIGURE 11. Schematic diagram of a synchronized seven- of music by electronic means. A catalog of natural soundschannel magnetic tape recorder for recording sevenseparate audio inputs or tone sequences on seven tracks recorded on magnetic tape may also be augmented byof the magnetic tape. the sound created on all kinds of electronic generators.

Modifying and shuffling these -sounds creates a varietyof new sounds that may be classified as discrete units

FIGURE 12. A-The seven audio inputs or tone sequences or series of units. These units and combinations of unitsrecorded in Fig. 11 are combined into one magnetic track. may be assembled to create musical productions. Hence,In turn, if the recorder has a seven-track capacity and all there are an infinite number of ways in which electronicthe magnetic tracks are filled, the final product then con-tains 49 separate audio signals or tone sequences. B- synthesis of music may be carried out by the generationIf the seven magnetic tracks of A, each containing seven- and modification of original sounds.tone sequences, are combined to form one magnetic track Electronic synthesis of music by these means has beenof the seven-track recorder, then when all the magnetic termed musique concrete in France, electronic music intracks are filled the final product will contain seven mag-netic tracks with 49-tone sequences each, which results Germany, Italy, Holland, and Japan, and tape music inin a 343-tone sequence tape. These 343-tone sequences the United States. The differences in various schoolscan again be combined into one magnetic track to repeat exist in the specific types of approach employed to carrythe process over and over again. out the process. For example, the source and nature of the

Multichannel A B Multichannel original sounds and modification processes are areas inmagnetic magnetic which the various schools differ. It is the purpose of thistape tape

reproducer reproducer section to describe the basic process of the electronico R 6;4 0g synthesis of original sounds.

A generalized diagram describing the main elementsinvolved in mixing and modifying original sources ofsound is shown in Fig. 13. The original sources may be

Amplifiers Amplifiers recordings on magnetic tape, oscillators, and noisegenerators. Recordings on magnetic tape may be the

Attenuators Attenuators sounds of voice, musical instruments, nature, etc. In somecases, the output of more than one magnetic tape repro-

Amplifiers Amplifiers ducer may be mixed (see Fig. 13). The various waveshapesAttenuator Attenuator of oscillators may also be employed. For example, theAmplifier Amplifier output of an oscillator with a sawtooth waveshape con-

ei- ^ A tains the fundamental and all the harmonics; the outputo Q3o2) o o of an oscillator with a triangular or rectangular wave-Single Single shape contains the fundamental and all the odd har-ortchne ortchne monics. The noise generator is designed to produce whitetape tape noise and other signals of a random nature.recorder recorder ~~~~~~~Forour purposes, a mixing system is used to combine

(i) (g ~~~~~~~~~~~~theoutput of the sound sources in the desired arrange-Final product: Final product: ment as determined by the musician. The output of the

7-tone sequences 49-tone sequences mixing system is then fed to the input of the modifiers.on each track on each track The modifiers include elements for introducing frequency


shift, vibrato, tremolo, portamento, timbre change, The decay controller determines the decay characteristicgrowth, steady state, decay, and reverberation. of the tone. The decay follows the steady-state durationThe frequency shifter is capable of changing the fre- of the tone. The growth, steady-state, and decay con-

quency by a fixed amount as in the case of a single-side- trollers are interconnected because the three character-band system. If this type of frequency shifting is applied istics follow a direct sequence of events. For example, theto a signal containing the fundamental and all the har- steady-state time may be zero, which means that decaymonics, the overtones will no longer be harmonics. The of the tone follows immediately after growth.frequency shifter may change all the frequencies of every Reverberation is an artistic (and acoustic) embellish-component by a multiplicative or divisive factor, with ment that plays an important part in blending a series ofmultiple relations of frequency overtones maintained. tones. Subjective considerations have established thatThe vibrato modulator provides a frequency modula- for each type of music there is an optimum value of re-

tion of the input frequency. The modulation frequency is verberation. A reverberator is included in the system ofusually somewhere below 10 Hz. Fig. 13 to supply artificial reverberation to the sound.The tremolo modulator provides an amplitude modu- The final product of electronic music synthesis is re-

lation of the input frequency. The modulation frequency corded on magnetic tape or disk for rendition of theis usually below the audio-frequency range. music in an auditorium or home.The portamento glider supplies a continuous frequency A rather highly sophisticated system for modifying a

glide in changing from a tone of one frequency to a tone tone has just been described for the purpose of illustratingof another frequency. the process. As a matter of fact, innumerable changes canThe timbre modifier consists of frequency-selective be made in the components and arrangement of the

networks that accentuate or attenuate the various com- system depicted in Fig. 13 for the electronic synthesis ofponents of the overtone structure of a complex tone input. music by the generation and modification of originalThe growth controller determines the growth or attack sounds, depending upon the composer's requirements.

characteristic of the tone.The steady-state controller determines the time length Formal electronic music synthesizers2

of the steady state of the tone after the growth has been In the preceding sections, the electronic synthesis ofestablished. music has been carried out by modification of conven-

Magnetic Magnetic FIGURE 13. Schematic diagramtape tape of the apparatus for the pro-reproducer reproducer duction of electronic music by_Interlock--- mixing and modifying original

sources of sound.

t t Atte~~~~~nuators

Attenuator

Multichanneltape

recorder

Olsoni-Electronic inusic synthesis for recordings 25

Magnetic tional and original sounds. The formal electronic musictape synthesizer, however, as exemplified by the manual, pro-recorder Loudspeaker grammed, and computer operation, provides the most

sophisticated means for producing recorded music. Insuch synthesizers, a single series of tones is generated atone time.

Hence, as a result, all the characteristics of a tone(as depicted in Fig. 1) can be applied to each series of

Fl r = = = n _ _ tones. This, of course, is impossible in the case of con-ventional musical instruments or, for that matter, of all

l l l /I I l electronic musical instruments because, in general, more

:11| l _ | L 5 1 _ than one tone is generated at a time. As a matter of fact,the entire composition is produced in one run. Thus itwilt be seen that synthesizers provide the most powerfulsystem available for the generation of musical sound andthe resultant production of recorded music. The sectionsthat follow will provide descriptions of the principalformal electronic music synthesizers.

Frequency and noise Manual electronic music synthesizers'0."As know, the properties

of a tone are frequency,intensity, growth, steady state, decay, portamento,

KeyboardI IT TTIITTT T timbre, vibrato, and deviation. When these properties of

Keyboard a tone are completely specified, the tone can be com-pletely described. The manually operated electronic musicsynthesizer, outlined in Fig. 14, is based upon the genera-tion of the these properties to produce the musical tone.The question may be asked as to the difference between a

Pedal Pedal highly sophisticated electric organ and the manual elec-control control tronic musical synthesizer of Fig. 14. In the synthesizer

FIGURE 14. The elements of a formal manual electronic of Fig. 14, only one series of tones is produced at a time.music synthesizer. However, in Fig. 14, each series of tones is not only re-

corded as in Fig. 11, but the various different series oftones are combined as described by Figs. 11 and 12.

FIGURE 15. The elements of a formal programmed Hence, the synthesizer in Fig. 14 can produce a tre-electronic music synthesizer. mendous range of tones with a maximum variation in

Magnetic each series that cannot be accomplished by the electrictape organ. For example, the performer is able to change the

Selsyn recorder Loudspeakerrelsyne recorder Loudspeaker volume, timbre, and some other characteristics with onehand while he plays on the keyboard with the other hand.The feet are also used to control some of the functions.

Sprocket drive Therefore, great physical dexterity and considerablemusical talent are required to produce acceptable musicby means of the manual electronic music synthesizer, themain reason being that each series of tones must possessdistinct musical qualities in order to produce musical ef-fects that are beyond the conventional. In addition, theremust be a perfect synchronism of the different series oftones that constitute the complete production. Such asynchronizing task is also carried out subjectively, thatis, by "ear." It is interesting to note that synthetic re-verberation may be introduced either at each note or overthe entire composition.Human engineering has been applied to the manual

electronic music synthesizer in order to provide thesynthesist with all possible assists in producing each

Sensor | | | g _ series of tones. Preset functions may be employed for dif-\ l l llDrum ferent music events, and various means of interconnection

may be provided to augment and extend the manualoperation. As the ultimate in assist, program systems'2

Coded | ll _ X @s I I are being developed to provide some of the character-paper Motor istics of tone. Such a program system is indicated by the

|1 | _ lGe ar) dashed lines in Fig. 14. The musician plays the keyboardw driv Alternating Selsyn of the instrument and the program system provides the

current transmitter desired tone characteristic at the proper time. The pro-


n L *-la <

Sif

G5 Ab Ab B Bb Gs

ob A

0Q

127 frequencies 16 voiumes f growths 16 timbres 2 vibratos FIGURE 16. Coded paper record+ 1 noise 81 steady state 'for the programmed synthe-

Idecays 4 portamentos sizer of Fig. 15.

FIGURE 17. System for sensing a track of the paper recordof Fig. 16. The electrical output represents the signal out-

gramming of the system must be carried out by the put of the sensing system for a section of the record.musician for each series of tones. As probably surmised, ..Electric outputtechnical aspects of synchronizing the pprogrammingsystem and the performing artist become quite complex. Photocell - =The manual electronic music synthesizer has great Paper_________________________

appeal for the accomplished performing musician be-cause he can hear the series of tones as he executes them. Sectional view of the sensing systemHe can go over and modify the series of tones until he issatisfied with the result, at which time he then carries out / ignal trackthe recording process for that series of tones. In terms of =.variety, manual synthesizers from very simple instru- W=_ ~ments to exceedingly complex and sophisticated elec-tronic systems have been developed and produced. Top view of coded paper record

Programmed electronic music synthesizersx3-16 Electric - - i nThe use of a programmed electronic music synthesizer Electric system operation Time.

opens up an entirely new field for the production of re-corded music. A programmed synthesizer is detailed inFig. 15. In order to synthesize any musical tone whatso-ever, the synthesizer must possess the following facilities:a means of producing tones with any fundamental fre- The program for each series of musical tones is re-quency within the audio-frequency range; a means of corded on a coded paper record, which controls all theproducing tones with any overtone structure; a means of functions of the electronic music synthesizer. The in-producing a tone with any growth, steady state, or decay formation is recorded and stored on the paper record incharacteristic; a means of introducing a vibrato; a means the form of black ink lines* as shown in Fig. 16. Whenof changing the intensity of the tone; a means of providinga portamento or glide from one frequency to another fre-quency;~ ~~ -an a men of inrduig intedsie.n * The original RCA eicctronic music synthesizers Mark X andquencyandmeansOI mtoducm, m tn uesleu m- Mark II employ a punched paper record instead of marked paper.

stances, various deviations of these characteristics. The marked paper record is a contribution of D. Friend.


the paper is actually run through the machine, positioned from four tracks, 16 volumes from four tmcks, and 127sensors scan the paper record as indicated in Fig. 17. frequencies plus one noise from seven tracks. The 88-noteThe operation of the sensor is as follows: A lamp illumi- player piano requires 88 tracks. With this system, 128nates the paper record. When the sensing system passes notes can be obtained from seven tracks, clearly demon-over a black line, the light received by the photocell (solid strating the saving in paper brought about by the binarystate) is reduced and so is the resulting current level. In system.effect, the sensor is actuated when detecting dark por- The preparation of the record of Fig. 16 is simple. Thetions of the paper. musician merely marks the paper with a wide pen to

It is clear that the paper record of Fig. 16 indicates use obtain the type of sound he desires. This marking of theof a binary code system. Employing this system, two dif- paper can be done anywhere. After the record is com-ferent functions may be obtained from one track, four pleted, he runs it through the synthesizer and listens tofrom two tracks, eight from three tracks, 16 from four the series of tones by means of a loudspeaker. If he is nottracks, etc. The gate tree for the binary system is demon- satisfied, he can make changes by creating new lines orstrated in Fig. 18. Here, any one of eight different inputs deleting existing lines with a white marker.can be selected by means of three sensors. Conversion from the musical score is straightforward,

Referring to the paper record of Fig. 16, it will be seen the frequencies being taken directly from the notes on thethat two vibratos are obtained from one track, four music score. The synthesist usually knows what he wantsportamentos from two tracks, eight growths, steady states, in the growth, steady-state, and decay characteristic. Theand decays from three tracks, 16 timbres are obtained overall interval is determined by the length of the note

itself.In general, the determination of the desired volume is

straightforward. This leaves the timbre, which can bedetermined by trial and error if the synthesist has notcarried out work on the electronic music synthesizer be-

1 Gates 2gatefore. The portamento is used only in gliding from onenote to another in a continuous frequency change. When

, -2, 3, or 4 the vibrato is used, it is a low-frequency modulation ofabout 7 Hz.

ffi 4 _ , , . Gate When a series of tones as recorded on the paper recordis found to be satisfactory, the result is recorded on one

t 5 or 6 1,2,3,4, 5, 6, 7, or 8 track of a multiple-channel synchronized magnetic tapeco 3 _ _ recorder (see Fig. 15) as used in Figs. 11 and 12. This

5 6, 7, orprocess is carried out for each series of tones until all ofthe series of the composition have been completed.

or70r8 Figures 11 and 12 amply describe the method of com-bining the series of tones. It should be noted that syn-

[L, ]l C ] Cthetic reverberation is usually introduced throughout theSensor Sensor Sensor completed composition.Sensor Sensor Sensor ~~~The use of the programmed electronic music syn-

FIGURE 18. A binary sensing system in which any of the thesizer for the production of musical sounds opens aneight inputs can be selected by three actuating sensors. enir fieldh i recorded music. Fpe, te

entirely new field in recorded music. For example, thereis the possibility of entirely new tone complexes and

FIGURE 19. Schematic diagram depicting the conversion combinations that cannot be achieved with conventionalof a sequence of numbers stored in a digital computer instruments. Furthermore, in the case of conventionalmemory into a sound pressure waveform. The sampling instruments, the musician is limited to the use of lips,rate is 10 000 numbers per second, yielding a bandwidth of mouth, ten fingers, two hands, and two feet to perform the5000 Hz for the sound wave. different functions. In the programmed synthesizer, there

are no limitations. If the composer or synthesist has inMemo -Loudspeaker mind what he wants to achieve, the effects can be obtained

6,13,16, 12, 11, 15, 12, 5, P _ by means of the programmed synthesizer regardless of-4,-6,-5, -1, 2 * * * _ _ _ _ whether he can play a musical instrument or not. In addi-

tion, in the case of manual synthesizer, there is the prob-lem of synchronizing the various series of tones. In the

6 13 16 12 11 15 12 5 -4-6 -5 -1 2 . . . programmed synthesizer, however, there is no problemsynchronizing these series of tones, as is evident from

m < 11 1llL t the direct selsyn drive between the paper record and theE § 1CA I I K magnetic tape recorder (Fig. 15).

XZE = a | | | S | | | 111Electronic music synthesis by digital computer's-l'dhr 0 AM

l |7'UI *111tAs already mentioned, music may be synthesized byS. I means of a digital computer. By use of a digital-to-ox analog converter, the numbers that a digital computer-10 o.0.0ox02 generates can be converted to electrical waves and then

0 0.001 0002sound waves by means of a loudspeaker. The numbers

Time, seconds stored in the computer program represent the desired


sound waves to be generated. A schematic diagram of the bers on each note card determine the starting time of theprocess of generating sound waves from a digital com- note, the instrument on which to play the note, the dura-puter is shown in Fig. 19. The sequence of numbers from tion, the amplitude, 1/(duration), and the frequency, inthe digital computer are converted to pulses of constant that order. The particular significance of these numberswidth, with the amplitude of the pulses corresponding to depends on the particular instrument on which the notethe numbers emitted by the digital computer. The pulses is to be played and the composer has a great deal ofare smoothed by a low-pass filter to obtain the input for flexibility in setting up his orchestra. Instruments ofa loudspeaker (Fig. 19), and a sound wave with frequen- greater versatility and complexity may be developed bycies from 0 to N Hz can be generated by 2N pulses per the addition of vibrato generation, a portamento glider,second. If the sampling rate is 10 000 per second, the top and other components.frequency emitted by the loudspeaker will be 5000 Hz. To "play" the music, the computer reads a line fromAssuming each sample is produced from a three-decimal- the score, inserting into an "instrument" at the properdigit number, the signal-to-noise ratio will be of the order time the parameters chosen by the "composer," thusof 60 dB. Hence, within the limits of the frequency range activating the instrument with the generated numbersand signal-to-noise ratio, the computer can produce any equivalent to the duration of the note. The sequence ofsound whatsoever provided the appropriate sequence of musical events is automatically taken care of by the pro-digital samples can be generated. gram and need not concern the composer.The basic procedure of listing 10000 numbers per As in the case of the programmed electronic music

second by the composer does provide a high order of synthesizer described in the preceding section, thegenerality. However, such a procedure is impossibly process is simplified if the composer listens to one seriestedious and, at the present time, practically out of the of notes at a time, makes suitable modifications, and thenquestion. Furthermore, with the present input means groups the combination of tones after each one has beenavailable, such a procedure does not result in a practical approved. Minor modifications can then be made in thesolution leading to effective control of the parameters group if such a procedure appears to be desirable.involved in the synthesis of music by means of a com- The description in this section has given one processputer. Finally, as in the synthesis of music described in for the electronic synthesis of music employing a digitalpreceding sections, the computer synthesis of music must computer. As programming for computers is simplified,20also be based upon the fundamental properties of a musi- the general procedure will be simplified. However, thecal tone as depicted in Fig. 1. composer operating the synthesizer, regardless of theThe basic form of the generating program is a scheme method employed for the electronic synthesis of music,

for producing a sequence of sounds representing in- is the final judge of the rendition. The analog and digitaldividual "instruments," which are formed by combining a synthesizers can produce any tonie whatsoever regardlessset of basic building blocks termed unit generators. Ap-propriate combinations of these unit generators canproduce sounds of almost any desired complexity. Thecompiling program is greatly simplified by the use ofmacro instructions, which specify a sequence of com- FIGURE 20. Instrument with attack and decay. A-Flowputer instructions by a single statement. In this way, each diagram. B-Envelope function. C-Waveform function.unit generator can be specified by a single macro state- D-Conventional music score. E-Computer score.

ment. (After Mathews)The first step in producing a musical selection is to P5 P6 P7 Fl

punch a set of cards that specify the "instruments" of the"orchestra." By way of example, a very simple orchestracontaining only one instrument is represented in Fig. 20. w vThe cards that define both the instrument and the score rsc 1are typified by Fig. 20(E). Cards 1 through 5 (lines 1-5) Fl \ = 60define the instrument as two interconnected oscillators, 0 20 491 511each oscillator having two inputs and one output. The 2 A Bleft-hand input specifies the amplitude of the output; the mSC F2 4 Jright-hand input, the frequency. The waveshape of the F2 Doscillation is specified by stored functions Fl and F2,which in general are not sinusoidal. The inputs P5, P6, B2 1and P7 to the instrument are set by the program to specific o 511,_values at the beginning of each note. P5 determines the iUTamplitude of the note and P7 the frequency. P6 is set to -11/(note duration); thus oscillator Fl goes through exactlyone cycle during each note. In this way, Fl becomes an Eattack and decay function that multiplies the waveshape-produced by oscillator F2, which is a slightly modified 1 N 0 1;-square wave whose shape determines the harmonic con- 3 0SC B2 P7 B2 F2 P29;tent ofthe output waveform. 4 0UT B2B1;

After the instrument has been defined, the next two 6 GEN 0 1 0 0 .99 20° .99 4v91 0.9 306 -.99 461 0 511;score cards (6 and 7) cause the functions Fl and F2 to be 8 NO0T 0 1 2 1000 .0128 6.70;calculated and stored in the computer memory. The 9 N0T 2 1 1 1000 .0256 8.44;actual notes are specified by cards 8 and 9. The six num--_______________________


of whether it has ever been produced before or not. 10. Moog, R. A., "Voltage controlled electronic music modules."Therefore, the method selected should be the one that J. AudioRE C..Soc etvo Mog1synhesi3r,"Audi, vo965311. EhIe, R. C., "Inside the Moog synthesizer," Audio, vol. 53,provides the greatest ease of translating what the com- p. 30, Dec. 1969.poser has in mind into the final product. 12. Ghent, E., "The coordinome in relation to electronic music,"

Electron. Music Rev., vol. 7, p. 33, Jan. 1967.A progress report 13. Olson, H. F., and Belar, H., "Electronic music synthesizer,"

J. Acoust. Soc. Am., vol. 27, no. 3, p. 595, 1955.In the year 1970, the disk record industry did a $1.4 14. Olson, H. F., Belar, H., and Timmens, J., "Electronic music

billion business, with prerecorded magnetic tape netting synthesis," J. Acoust. Soc. Am., vol. 32, no. 3, p. 311, 1960.$500 million. In short, prerecorded music for the con- 15. Logcmann, G. W., "Techniques for programmed electronicsumer is a very large business indeed. music synthesis," Electron. Music Rev., vol. 7, p. 45, Jan. 1967.

Clsia ufdisk and pre- 16. Hiller, L. A., and Isaacson, L. M., Experimental Music. NewClassical music records in the form of dsanpr- York: McGraw-Hill, 1959.recorded magnetic tape constitute about 5 percent of the 17. Mathews, M. V., "An acoustic compiler for music and psy-recorded music business. Popular, country, western, and chological stimuli," Bell System Tech. J., vol. 40, no. 3, p. 677,contemporary ("rock") music represent the remainder 1961.

and threfor~ll te entre buiness.At prsent, 18. Mathews, M. V., "The digital computer as a musical instru-and therefore practicaly the entire business. At present, ment," Science, vol. 142, no. 12, p. 553, 1963.classical music is recorded and produced without any 19. Mathews, M. V., The Technology of Computer Mutsic. Cam-electronic modifications; however, the remainder of the bridge, Mass.: M.I.T. Press, 1969.record industry (95 percent) uses modification by elec- 20. Mathews, M. V., and Moore, F. R., "GRooVE-A computer

program for real-time music and sound synthcsis, J. Acoust. Soc.tronic means to some degree. Am., vol. 47, no. I, p. 132, 1970.Exhaustive studies have been made of formal electronic

music synthesis, in which conventional musical instru-ments are not used in any form. In the past decade, more Reprints of this article are being made available tothan 1000 music compositions have been released in the readers. Please use the order form on page 10, whichform of records produced by formal synthesis, and the gives information and prices.field is expanding at a very rapid rate. It is beyond thescope of this article to review all the fine records thathave been produced by formal synthesis; however, the Harry F. Olson (F) attended the University of Iowa, whereauthrwuldbe en-issif some of the outstandingly suc- he received the B.E. in 1924, the M.S. in 1925, the Ph.D. inauthor would be remiss if some of the outstandmGly suc- 1928, and the professional degree of electrical engineer in

cessful ones were not mentioned. 1932. He also received the honorary D.Sc. degree from IowaOne overwhelming success in the field of formal Wesleyan in 1959. In 1928, Dr. Olson joined the Radio Cor-

manual electronic music synthesis is entitled "Switched- poration of America as a member of the Research Depart-On Bach" by composer Walter Carlos and released by ment. After being placed in charge of acoustical research

Columia Rcord (Streo S719). Tis cmpostion in 1934, he continued as head of RCA's acoustic researchColumbia Records (Stereo MS7194). This composition activities during the ensuing years, retiring as staff vicewas produced on the Moog Electronic Music Synthesizer, president, acoustical and electromechanical research, RCAand has sold well beyond one million copies. In another Laboratories, in 1967. At present, he is serving as an ad-instance, composer Charles Wuorinen was the winner of visor to RCA Laboratories.a 1970 Pulitzer Prize for his "Time's Encomium," the An early and important contribution of Dr. Olson's long

awardindicting career was the development of velocity and cardioid uni-first all-electronic work to receive the award-indicating directional microphones. He has also made pioneering con-that electronically synthesized music has attained major tributions to loudspeaker development, improvement ofstatus in the music field. "Time's Encomium" was re- photograph pickup and recording equipment, underwaterleased for the home listening audience by Nonesuch sound equipment, and motion picture sound and reen-forcement systems. In addition, he has guided and con.Records (Stereo H-71125), and was performed on the tributed substantially to such novel systems as the elec-programmed RCA Electronic Music Synthesizer at the tronic music synthesizer, speech processing, and theColumbia Princeton Music Center, New York City. phonetic typewriter.

A frequent contributor to professional journals, Dr. Olsonhas written several books and holds more than 100 U.S.

The author appreciates the valuable comments and suggestions patents on devices and systems in the acoustical field. Hefrom David Friend, Tonus Inc.; Max V. Mathews, Bell Telephone is a member of Tau Beta Pi, Sigma Xi, and the NationalLaboratorics, Inc.; Robert A. Moog, R. A. Moog Inc.; and John Acdemy of Sciences a FM. Woram, RCA Records. Aciemy of Scie a Fellow of the American Physical

Society, the SMPTE, and the Acoustical Society of America,of which he is a past president, and an Honorary Member

REFERENCES of the Audio Engineering Society, of which he is a past1. Seashore, C. E., Psychology of Music. New York: McGraw- president, and the American Society of Swedish Engineers.Hill, 1938. For his contributions to the field of audio engineering,2. Olson, H. F., Music, Physics and Engineering. New York: Dr. Olson has received many honors, including the ModernDover, 1967. Pioneer Award of the NAM (1940), the John H. Potts Medal3. Woram, J. M., "A pop recording session," Audio, vol. 53, p. 21, of the Audio Engineering Society (1952), the Samuel L.May 1969; also vol. 53, p. 30, June 1969. Warner Medal of the SMPTE4. Kent, E. L., "Musical instruments with electronic amplification (1955), the John Scott Medal ofor tone modification," J. Audio Eng. Soc., vol. 17, no. 3, p. 316, the City of Philadelphia (1956),1969. the Achievement Award of the5. Martin, D. W., "Electronic music 1988," J. Audio Eng. Soc., IRE Professional Group onvol. 17, no. 4, p. 386, 1969. Audio (1956), the John Ericsson6. Douglas, A., The Electrical Production of Music. New York: Medal of the ASSE (1963), thePhilosophical Library, 1957. Emile Berliner Award (1965), and7. Ussachevsky, V. A., J. Audio Eng. Soc., vol. 6, no. 3, p. 202, the Institute of Electrical and1958. Electronics Engineers' Mervin J.8. Badings, H., and de Bruyn, J. W., Philips Tech. Rev., vol. 19, Kelly Award (1967), Consumerno. 6, p. 191, 1957/58. Electronics Award (1969), and9. Searight, J., Radio-Electron., vol. 36, no. 6, p. 36, 1965. Lamme Medal (1970).

30 Olson-Electronic music synthesis for recordings

Documents

Electronic music synthesis for recordings