US3467758A - Multiple speaker sound output system for reducing intermodulation distortion - Google Patents

Description

Sept 16, 1969 n. w. MARTlN MULTIPLE SPEAKER soUND OUTPUT SYSTEM FOR REDUCING INTERMODULATION DISTORTION Original Filed March l2, 1964 '7 Sheets-Sheetl l mr Nm M mHH szBLjL-T ammo -zarawnb S; wwowv man. mm

Nnfwm. n NN INVENTOR Damm UJ. MARTIN ATTORNEYS Sept. 16, 1969 D. w. MARTIN 3,467,758

MULTIPLE SPEAKER SOUND OUTPUT SYSTEM`FOR REDUCING INTEHMODULATION DISTORTION Original Filed March l2, 1964 7 Sheets-Sheet 2 Qumran octave FILTER AMPuFlElzS LOUDSPERKER) 1r F02 1F16. 4v Y, IJSH d (45 Mamm a PQEAMP Farsg PLAYBACK Em fw m m 6 m- El m Foa 46552 11 @L T E( H\GHPRSSP\LTEL CB YI "ffoue 83 l Law-Pass mme?.

BELDU) F1 6.5 1 C4 M84 I 512m@ F| LT ERS |09 'AMPLmERs LouosPEnKezs '15 A, M, B HAaMomcs " qg F# 6,6* Je,

HNZWN KS 'T7 al D? E. F "J Hnzmomcs INVENTOR HAEMONICS i ATTORNEYS Sept. 16, 1969 D. w. MARTIN MULTIPLE SPEAKER SOUND OUTPUT SYSTEM FOR REDUCING INTEHMODULATION DISTORTION Original Filed March l2,

7 Sheets-Sheet INVENTOR Damm. LAMARTN ATTORNEYS mmvHm..

ep 16, 1969 D. w. MARTIN MULTIPLE SPEAKER SOUND OUTPUT SYSTEM FOR REDUCING INTERMODULATION DISTORTION Original Filed March 12, 1964 7 Sheets-Sheet 4 INVENTOR DnmEL Ld. MARIN ATTORNEYS Sept. 16, 1969 D, w. MARTIN MULTIPLE SPEAKER SOUND OUTPUT SYSTEM FOR REDUCING INTERMODULATION DISTORTION Original Filed March l2, 1964 '7 Sheets-Sheet 5 mmn w .2.a mm.

o .wud N8.

BAHN T xauNOJ XB'RINOD INVENTOR DnmEL. Mmmm (snmowaw wu) 3mm *D aomaauas A ORNEYS sept. 16,1969

MULTIPLE SPEAKER SOUND OUTPUT SYSTEM FOR B. w. MARTIN 3,467,758

REDUCING INTERMODULATION DISTORTION Original Filed March l2, 1964 7 Sheets-Sheet 6 SINE WAVE CONTINUOUS G ENE RATORS Dnmalu. MARTm Y l L ATTORNEYS Sept. 16, 1969 D. w. MARTIN MULTIPLE SPEAKER SOUND OUTPUT SYSTEM POR REDUCING INTERMODULATION DISTORTION Original Filed March l2, 1964 7 Sheets-Shea?, '7

.v @HMA *NEEM [L (IlllzOS Pauw@ ZQQZD .jwwsw J| 1 l n mw ma @HH mp n VI I -l [L W Pdwd@ .nu U U U F .EWN W n E NF w E m 1E EN, P .I l I I I I I I I I l I l I .IIL 20,@20 d-OIU\| IIWzO-.VGUQJ dwxdwmO-JOI- l.' H HH E Hm m ,m m m .vw o O, N Lv v N vw O. O. o v w N 0 N 0`0 Y o? OH O Wo-Nw S6525 ouwzm oz: 8v zkoow, n

ATTORNEYS United States Patent O U.S. Cl. 84-1.02 12 Claims ABSTRACT F THE DISCLOSURE An audio frequency system for processing musical signals, particularly for reduction of intermodulation distortion produced by simultaneous tones subsisting in relatively small sub-bands of the audio frequency band, Iby subdividing the audio band in terms of quarter octave filters, combining the outputs of those filters which have primarily octavely related partials into separate channels, and transducing the signal content of the channels.

This application is a division of my application Ser. No. 351,427, filed Mar. 12, 1964, and entitled Multiple Speaker Sound Output System for Reducing Intermodulation Distortion. t

The present invention relates generally to systems for processing musical sounds, in its electronic production, recording and/ or reproduction, and more particularly to systems for reduction of intermodulation distortion produced by simultaneous tones subsisting in relatively small sub-bands of the audio frequency band, while enhancing the musical quality of the simultaneous tones.

A most important limitation of present high-fidelity music amplification systems is the necessity to reproduce simultaneously from the same amplifier and loudspeaker tones lying in the same general region of the audio-frequency spectrum. In systems of high quality it has been customary to divide the entire audio frequency range into two, three, or even four broad bands, so as to avoid intermodulation distortion of the type produced by simultaneous tones of widely different frequency. However the other principal form of undesirable intermodulation distortion, (that produced by simultaneous tones within a few octaves of each other, has continued to degrade sound reproduction. These effects are particularly noticeable in the radius and high-frequency ranges of sound, because difference frequencies in intermodulation distortion are much more audible than the summation components. The sense of hearing is rather insensitive to Weak sounds of very low frequency. Hence the tones which (in combination) generate disturbing distortion must be high enough in frequency for their difference to be in the order of 100 cycles or more.

Another problem in sound reproduction systems, especially in large auditoriums, has been to retain the desirable directional characteristics designed into a loudspeaker of limited power-delivering capacity, when the power requirements substantially exceed the unitss capacity. The obvious practice is to use a large number of similar units electrically in parallel, and to neglect the spectral degrada- Mice tion resulting from interference effects between waves from different units, and the highly directional beaming effect produced by a parallel array of similar units.

When music is to be produced (rather than reproduced) electro-acoustically, similar problems exist. One of the inherent economic advantages of electronic organs over pipe organs, for example, is that the tones can all be radiated acoustically from the same sound source. However, there are disadvantages from a tonal standpoint which are inherent in the single sound-source method of tone radiation, such as (l) intermodulation distortion, (2) a deficiency in apparent size of the musical sound source (in comparison to the historical instrument), and (3) a lack of change in sound localization when different notes in the scale are played. It is possible, of course, to have different notes of the same pitch come from audibly different location, eg. when different loudspeakers are used for different divisions of the organ, but this does not provide for any motion of the sound source when the note changes within the division;

One electronic organ system method which is used to increase the apparent physical size of the sources of sound is to use multiple sources in parallel. As in sound reproduction systems, this tends to beam the sound toward the listeners more than a single source would. This is highly undesirable for organ music. Another expedient is to orient the sound sources toward reflecting surfaces, so as to increase the ratio of reflected-to-direct sound. Installing the sources in a moderately reverberant tone chamber coupled to the auditorium is still another approach. These methods reduce acoustical efficiency, and reduce the tonal definition required for full appreciation of musical subtlettes such as transient effects in the tone.

Accordingly it is a principal object of my invention to produce or reproduce musical sounds with a minimum of intermodulation distortion.

A further object is to provide a broad smooth directional characteristic throughout the musical frequency range, so that the tonal spectra at all normal listening positions will closely resemble each other, and will also be similar to the spectrum of the total power radiated from the loudspeakers. This will tend to make the direct sound and the generally reflected sound similar in timbre.

A further object is to, provide motion for the musical source as the frequency changes, and to provide spread for the musical source when various non-octavely related frequencies are present.

A further object is to provide a system for musical spatial modulation effects which are a rapidly changing function of frequency, i.e. which vary greatly within the span of an octave.

The key to the solution of the above problems and to the attainment of the above objects is to be found in the nature and composition of musical sound, and in an acoustically novel subdivision of the audible frequency lrange. Simultaneous octavely related tones occur rather frequently in music. Their intermodulation products are not very serious musically because the distortion products of oc'taves are also octavely related to the original tones. Consequently they are very difiicult to distinguish from the original combination of tones. An ideal radiation system for musical tone, from an intermodulation distortion standpoint would be one in which only octavely related tones are radiated together. Only in very large sound systems would this be economically feasible. Another musical factor, which is pertinent to my invention, is that the frequency of occurrence of closely adjacent musical tones in combination is very small. When such adjacent tones (e.g. E and F in the musical scale) are played, the dissonance effects experienced by 'the listener overshadow intermodulation distortion effects which, by comparison, are minor. Furthermore the more nearly adjacent are the tones played in combination, the lower the difference frequency becomes and (because the ear is quite insensitive to low frequencies) the less audible and objectionable the difference tones will be.

The above described combination of factors makes it advantageous to combine the ideal twelve octavely related groups of musical tones into six, four, or even three groups, and thus effect a reasonable compromise between the requirements for low intermodulation distortion and low cost. Combination into three groups is not too advantageous, because minor thirds are rather common intervals, and are quite consonant in comparison to major second intervals, which are the largest combined intervals (short of an octave) which will occur when there are four major groups of octavely related tones. Electronic and electrocoustical systems designed along these principles of frequency-range division and recombination may be described by the terms octaphonic There are numerous other principles and refinements in the present invention, which will become apparent as the details are disclosed.

Other objects, advantages and features of the invention will also become more fully apparent from the following description of preferred embodimentsjthereof, taken in conjunction with the appended drawings, in which:

FIGURE 1 is a block diagram of an octaphonic system of recording, according to the invention;

FIGURE 2 is a block diagram of a playback system useful with the system of FIGURE 1;

FIGURE 3 is a block diagram of a system for reproducing octaphonically music derived from a conventional one track record;

FIGURE 4 is a block diagram of a modification of the system of FIGURE 3, wherein simplifications have been effected;

FIGURE 5 is a block diagram of a modification of the system of FIGURE 3 employing relatively few multiply resonant filters in place of many singly reasonant lters;

FIGURE 6 is a block diagram of a system employing the octaphonic principle in a musical instrument which generates tone forms directly;

FIGURE 7 is a block diagram of a simplified modification of the system of FIGURE 6;

FIGURE 8 is a block diagram of a formant type of electronic musical instrument employing the octaphonic principle;

FIGURE 9 is a schematic block diagram of a resonator type of electronic musical instrument employing the octaphonic principle;

FIGURE 1'0 is a block diagram of a simplification of the system of FIGURE 9;

FIGURE l1 is a block diagram of a harmonic synthesis type of electronic musical instrument employing the octaphonic principle;

FIGURE 12 is a block diagram of a space modulation system for octaphonic sound production or reproduction equipment;

FIGURE 13 is a block diagram showing the locations of complex tonal centroids relative to locations of the loudspeakers in an octaphonic system; and

FIGURE 14 illustrates an exemplary physical arrangement of the loudspeakers in an installation of a multimanual electronic organ of an octaphonic type.

In sound recording and playback systems it is generally possible to provide microphones and pre-amplifiers with remarkably good linearity of response, so that one need not be particularly concerned with intermodulation distortion in the low-level part of the recording system. However, the recorder amplifiers, the heads, and the medium itself may introduce audible distortion products, and frequently do. Consequently it is highly desirable, if possible, to apply the octaphonic method of frequency-range division ahead of the recorder amplifiers, as shown in FIGURE l.

In the system of' FIGURE 1, sound is picked up by microphone 1. The microphone output signal is amplified by pre-amplifier 2, from which the amplified output is supplied through common input 3 to a set of quarteroctave filters 4-27, inclusive. These filters cover the fundamental frequency range from 32 to 2000 c.p.s. approximately. Another two or three octaves of filters would be needed to fill out the upper portion of the entire audiofrequency spectrum. The transmission characteristics of these filters need not be perfect, from a frequency-range division standpoint. For example, a 10 db discrimination against frequencies in the adajcent passband is quite satisfactory for the purpose of this invention, although 20 db would be desirable. The outputs of filters 4-27 are cornbined through isolating networks (not shown) into four octavely related groups, II, III, IV, V, which are separately amplified by recorder amplifiers 28-31. The outputs lof the recorder amplifiers, together with appropriate signals for biasing the recording medium, are supplied individually to a four-channel head 32., recording on tape 33, and

' comprising individual recording elements 32a, b, c, and d.

In greater detail, the filters 4-27 inclusive are each arranged to pass a frequency span of three semitones of the musical scale. So, filter 4 passes the fundamental of tones A6. At, B6, and nearby partials in the 6th octave of tones in lower octaves, while adjacent filter 5 passes only F46, G6, Gl, filter 6 passes D46, E6, F6 and 7 passes C6, Cll and D6. The four filters 4-7 encompass the entire 6th octave. The four filters 4-7 lead, respectively, to recorder elements 32a, b, c, d.

In essence, all the octaves follow the rule for the sixth octave, so that, for example recorder element 32a records only partials corresponding to A, Ait, B, in all of the octaves.

In playback, FIGURE 2, the tape 33 moves past fourchannel playback head 34, and the four playback signals (octaphonic channels II-V) are amplified separately by playback amplifiers .3S-38 and radiated from separate loudspeakers 39-42. This system provides maximum economy for the purchaser of playback equipment, because the frequency-range division occurs in the recording part of the system. Only the recording studios require the filter set, and the users of the tapes and the playback systems gain the advantages.

The quality of music recordings varies greatly in accordance with the care employed and the equipment used in making the recording. However, a recorded track which is relatively free of intermodulation distortion can be badly degraded by the reproducing system. This is particularly true when the reproduction is at high sound level, or when a very large auditorium or arena must be filled by the sound reproduction. FIGURE 3 illustrates how octaphonic frequency-range division can be advantageously accomplished in reproducing from a singlechannel recording of music. The system can, of course, be duplicated for stereophonic reproduction. Single-channel playback head 43 responds to the single track recorded on tape 44, and preamplifier 45 provides the full-range of the musical spectrum through common input 46 to the inputs of the filters 47-74. In this case the quarter-octave filters 47474 separate the reproduced signal into quarteroctave bands which are octavely combined (through isolation means not shown) in mixing circuits II-V. Again, as in FIGURE l, the four octaphonic channels are amplified and radiated by amplifiers 75-78 and loudspeakers 79-82. Although it is less economical f-or the filter set to be in the reproducing system, there is one great advantage. Such a reproducing system can accept program material from any recorded track. There is no problem of compatibility with the original recording system. However the octaphonic reproducing system cannot correct for TABLE 3 non-linear diStOrtOnSinthe recording. Quarter-Octave Identification of the Partials of a Complex Twenty-Harmonic Tone G1 Group II C, C#, D III D#, E, E

IV Ffh G, G#

V A, A#, B

1 ...A 49.0 C1 IV C505# II t TABLE 1 2 9&0 G2 IV D5 II Quarter-Octave Identification of the Partials of a Complex Thrty- 3 147.0 Da II D5#E5 III Harmonic Tone C1 1 6 o IV F Group gl g3# (re, g 9 G3 5- HI IV 1?#,8 G1' G# 245.0 Ba- V F#5 IV V A, 4* B 294.0 134+ II G5 IV H C! 523.2 H C5 343.0 F4 III G5# IV n o2 555.9 n c#6+ 392'0 G W A5 V Iv G2 588.6 II D5+ 441.0 A4 V A#5 V n o. 821.3 In D#5- 4900 B4" V B5 V III E- 654.0 III E5- IV G3 686.7 III F5- T ABLE 4 V A#a+ 719.4 IV F5F#5 Quarter-Octave Grouping of the Partials of Any G Tone 5 25 1I C4 752 1 IV F# n rv 111 v II Dri- 784.8 IV G5 III El- 817.5 Iv Gif. 1

rv Firm 850.2 Iv G#5A5 2 Iv G. 882.9 v A5 30' 3 rv G#4+ 28 915.6 v M5 4 457.8 v .@#4- 29 948.3 v A#.+ 5.

48o 5 v :B4- 3o 961 o V M5135 6 TABLE 2 13 Quarter-Octave Grouping of the 14 Partials of Any Tone 15 II IV III V 16 It should be pointed out clearly that the octaphonic 7 lter system does not direct all of the partials of a com- 8 plex tone into a single channel, only those which are 9 approximately octavely related. Table l shows an example of the routing of the individual partials of a highly 10 complex musical tone having a fundamental frequency 11 of 32.7 c.p.s. (C1) and 30 harmonic partials. In the first 12 column is the harmonic number. The second column is the frequency in c.p.s. The fourth column identities 13 65 the note nearest to the harmonic in the musical scale. 14 (Plus and minus signs show the direction of deviation 15 when coincidence is poor, and when the partial lies nearly halfway both note letters are shown.) The third column 16 22 19 27 shows which of the four octaphonic channels will trans- 17 23 20 2g 70 mit the harmonic. Table 2 brings the various harmonics 18 24 21 29 together into four octaphonic channels. The number of harmonics in each channel varies from 5 to 10. In this 25 example the II channel, which transmits the fundamental 26 ao frequency, carries five of the first ten harmonics. Channels III and IV each carry two, and V has only one of the first ten. The next three harmonics go to channel IV. Although the lower harmonics are usually of greater amplitude than the higher harmonics, there are some complex tones in which the higher harmonics are significant. For this reason there is some advantage in locating the output of channel IV adjacent to II. These two channels carry ten of the rst 13 harmonics, and 13 of the first 18 harmonics. This is why channels III and IV are transposed in FIGURES 1 and 2, and frequently throughout this application.

Another example will illustrate that it is also advantageous to have channel II nearby when the fundamental frequency lies in channel IV. Table 3 shows a 20-harmonic tone of fundamental frequency 49 c.p.s. (C1), lying in channel IV. Table 4 groups the harmonics into octaphonic channels. It is apparent that channel II has the same relationship to a tone having its fundamental in channel IV as IV did to II in the previous example. The small differences in the high harmonics are because G1 is in the middle of its quarter-octave filter range while C1 is off center. The effects of harmonie distribution among the four channels upon tonal localization will be analyzed hereinafter in connection with FIGURE 11.

FIGURE 4 is a simplification of FIGURE 3, which takes advantage of the normal energy distribution in musical sounds, and of the relative inaudibility of the very low-frequency difference tones generated by combinations of signals in the lowest octaves of the musical scale. Tonal partials lying above C8 (approximately 4000 c.p.s.) are generally quite weak, and can be omitted from the octaphonic filter system without much degradation in performance. The lower harmonics of the lower notes are sparsely distributed along the frequency scale in most music. Furthermore they are not easily localized by listeners and their difference tones are hardly audible. Therefore they can be combined as shown in FIGURE 4, with small degradation in purity of reproduction. Highpass filter 83 and low-pass filter 84 can be connected to amplifiers and loudspeakers designed specifically for their frequency ranges or, for the sake of economy, can be combined and mixed into all four of the octaphonic channels. Preferably the loudspeakers in this case would be Woofer-Tweeter combinations with dividing networks, in order to avoid intermodulation distortion between widely separated frequencies.

An alternative means for separating the tonal components into -octaphonic channels is the use of multi-resonant filters. An example is a bank of suitably damped flexible, vibrating strings tuned to the twelve equally tempered frequencies in a low octave of the scale. In FIGURE such a set of string filters 84-95 is shown schematically, with separate input transducers 96-107 and output transducers 108-119. The electro-mechanical transducer usage could be simplified. A single input transducer T of high mechanical impedance could drive all twelve strings, and each group of three strings could have a common output transducer, leading to channel amplifiers 75-78 and the corresponding loudspeakers 79-82. This system has the advantage that any given amplification channel Iwill transmit all of the harmonics of a tone lying in its group designation. However some of the higher harmonics will also be reproduced in other channels as well, so it will be necessary to attenuate the high-frequency response of all channels to compensate for the multiple reproduction of the higher frequencies. The chief advantage of this system is that only twelve filters are required, and they are of a type which can be sharply tuned at each resonance frequency at relatively low cost.

The simplest application of the octaphonic principle to musical instruments is to instruments which generate each individual musical -wave form separately. The pipe organ is an example. If a small pipe organ were to be amplified so as to fill a large arena, it would be a advantageous from an intermodulation distortion standpoint to have all of the C, Cif, D pipes so arranged as to be picked up by one microphone; all of the Dfi, E, and F pipes by another microphone; and so forth. Each microphone would be connected to its own amplifier and loudspeaker system.

The counterpart for the various electrical and electronic Waveform-generating types of instruments is shown in FIGURE 6. Generators in bank 120, keyed on from the organ clavier, may be any type of complete waveform generator, such as an electronic generator of a complex tone spectrum, an electrostatic scanner of a variable capacitance generator, a magnetic scanner of a specially distributed or recorded magnetic waveform, a photoelectric pickup of a photographic image of a musical waveform, and the like. In all of these the outputs from all of the Cs, Cits, and Ds in each octave of each stop would be combined (with reasonable care in isolation and loading) and amplified in amplifiers A and radiated separately from the other three groups of tones, in radiators R. Note in FIGURE 6 that the loudspeakers have acoustically separate back spaces because they are mounted close together. Preferably the loudspeakers would be spaced apart as well as partially enclosed so that intermodulation distortion resulting from acoustical coupling between loudspeakers will be minimized.

Although the explanation of this form of octaphonic system for organs is simple and straightforward, this form of the system has the disadvantage that each individual tone has less spatial spread than in some of the later systems (e.g. FIGURE 9). Patent No. 2,596,258 separates the tones of the scale into two groups for separate amplification and radiation, with alternate notes in the equally tempered scale going to different channels. The purpose of that invention was to prevent electrical beating effects between tones having equally tempered intervals of a perfect 4th or a perfect 5th (e.g. C-F and C-G). The present invention differs from Patent No. 2,596,258 in that the tones have a greater variety of source locations. Furthermore the musical possibilities for intermodulation distortion are negligible in comparison to Patent No. 2,596,258, in which a number of frequently used combinations of tones (eg. major third, augmented fifth) are amplified and radiated together with attendant intermodulation distortion.

A simplification of the system of FIGURE 6 is shown in FIGURE 7. Here all the lower tones in the scale are combined and supplied to amplifier 121 and loudspeaker 122, specifically designed for low tones. In this range of playing the simultaneous tones are usually (although not always) octavely related anyway, in contrast to the upper octaves where chords are common.

In the system of FIGURE 7, the octaphonic division of tones is followed for all tones above the lower tones, the division point being relatively arbitrary. The amplifiers and radiators are accordingly denoted A and R, as in FIGURE 6.

In the formant type 0f electronic musical instrument complex waveforms are selected from a musical scale of generators (either keyed-on or keyed-output, the latter being indicated in FIGURE 8) and are supplied in combination to common stop (or tone-color) circuits prior to amplification. In other words the same electrical waveshaping net-Work is used regardless of fundamental frequency. This method of producing tone colors offers wide variety of timbre at relatively low cost, because the tone color filters do not have to be reproduced for each note or each octave. For the advantages of the octaphonic principle to be realized in the formant type of instrument it is necessary only to wire each octave of tone outputs in four groups, and to combine them octavely in mixing circuits as shown in FIGURE 7. Each octaphonic channel of signals is passed through an individual stop filter before separate amplification. Connections from coupled Scales of generators or key switches are made in the usual manner. Each stop (Y, for example) would employ four similar (but not necessarily identical) tone color filters Y2, YB, YQ, Y5 each receiving a different quarter of every 9 octave. Stops X, W, U, and Y would also have 4 filters each. Stop switches, whether on the input or the output f the filters, would be of the 4-pole type. Thus the octaphonic principle can be applied to the formant instrument simply by quadrupling the least expensive part of the electronic system, the tone color filters.

Resonator organs of the type described in application for U.S. Patent Ser. No. 46,704, filed in the name of Wayne, are ideally suited to the octaphonic principle, because the organ already contains (for other purposes) an ideal set of filters for octaphonic use. Two notes of such an organ are shown in FIGURE 9. Organ generator C1 supplies all the necessary harmonics of a complex wave to common input point 124. Generator G1 does the same for common input 125. From these two common input points are connected many resistors 12Go, b, c, etc., and 127a, b, c, etc. which determine the amount of signal current supplied to the appropriate singly resonant circuits (or resonators) in circuit bank 128. The circuits selected from this bank or resonant at or near harmonic frequencies of the input tone waves. Thus the resistors 126a, b, c, etc. are spectrum shaping resistors for the note C1. Different stops have different spectrum shaping resistor values for the same fundamental frequency. Only are stop is shown here for simplicity. However, no matter how many input circuits are connected to the resonator bank 128, the outputs of the individual resonators are connected octaphonically in groups II-V, and amplified and radiated separately for minimum intermodulation distortion.

Referring now to the second object of the invention, assume that each loudspeaker R in FIGURE 9 has been designed to have the best possible directional characteristic throughout its frequency range. The multiplicity of channels, and the fact that no note is played on more than one loudspeaker, allow the angular distribution of energy from the system to be just as uniform as one loudspeaker can provide. Each loudspeaker is radiating a different set of frequencies, and can be driven beyond the usual limitations due to non-linearity before audibly spurious frequencies are generated. The necessity to combine loudspeaker outputs in beaming arrays is obviated. When higher powers are required, the number of octaphonic channels can be increased to six or even twelve before any one channel duplicates the frequency of another.

FIGURE 10 is a simplification of FIGURE 9, in which all of the lower partials, irrespective of location within the octave, are combined for separate amplification by amplifier 121 and radiation by special low-frequency loudspeaker 122. This simplification reduces the number of very low-frequency loudspeaker units required. Such units are usually large and require large enclosurw, so reduction of their number minimizes space requirements and cost. This simplification is possible because chords are seldom played in the pedal section of the organ (except by virtuosos), and the difference tones generated by such low-frequency partials are too low in frequency to be audible.

The harmonic-synthesis type of electronic organ in FIGURE l1 uses a bank 123 of continuous sine-wave generators which are tuned to the equally tempered scale. The generator outputs are wired to ganged key-switches which provide simultaneous switching of a number of approximately harmonically related frequencies, eight in the case of FIGURE 1l. Because of the complexity in wiring of such a system, only three ganged key-switches are shown for the simple case of a major triad, C2, E2, G2. The principal commercial type of harmonic synthesis organ employs harmonic drawbars. All of the first harmonics are normally collected and supplied to drawbar No. 1, all of the second harmonics to drawbar No. 2, and so forth. FIGURE 11 shows how the octaphonic principle can be applied to a harmonic synthesis organ, by using four sets of drawbars which (for convenience to the player) would be ganted, so that a single motion would be required for setting the relative amplitude of each liarmonic, just as in present harmonic synthesis organs. Note that for C2 the output of the first harmonic switch goes to the first drawbar of the lowest set, and the output of the second harmonic switch goes to the second harmonic drawbar of the same set. However the output of the third harmonic switch goes to the third drawbar of the next higher set. The 4th harmonic, being octavely related, goes to the lowest set of drawbars, but the 5th harmonic goes to the 5th drawbar in the third set. The seventh harmonic of C2 (frequently omitted in synthesis organs because the nearest equally tempered tone is grossly inharmonic) would be routed to the 7th drawbar of the top set. Similar connections are shown for E2 and G2. This enables the partials of a harmonic synthesis organ to be amplified and radiated octaphonically with audible intermodulation distortion practically eliminated.

In the electronic organ field there is an entire class of modulating devices and circuits which make the electronically produced tone more interesting, more realistic, or more desirable aesthetically to the listener. Examples are phase vibrato circuits, chorus or celeste circuits, reverberation circuits or systems, and random or non-periodic modulators. It has frequently been found desirable to modulate differently in different octaves, as in Wayne Patent No. 3,004,460, and to radiate the different modulations separately from each other and from the original unmodulated signal, asin Wayne application Ser. No. 45,609. The octaphonic principle of frequency-range division provides advantages over previous systems employing modulations in that the different (or uncorrelated but similar) modulation effects are applied to different frequencies within the same octave. FIGURE 12 is an example in which, in addition, each individual complex tone will have space modulation within itself. This is because some of the harmonics are being modulated differently from the basic group of harmonics which are approximately octavely related. In FIGURE 12 the bank of singly resonant circuits 128 of previous FIGURES 9 and l0 is the source of signals to be combined octaphonically. The scale of generators 120 in FIGURE 6, or the scale of key switches 123 of FIGURE 8, could just as easily have been used. Had they been, however, only the first advantage, i.e. different or uncorrelated modulation of different notes Within the same octave, would have been achieved. In the case of the resonator organ the second advantage of modulation enhancement within the partial structure of a single tone is also achieved.

Assume for discussion that the channel modulators in FIGURE 12 are phase shift circuits 129-132, and that the channel modulators are under the control of suitably filtered sources 13S-136 of random noise. These could, of course, be dlifferent bands of noise from a single noise source for simplicity and economy. The important thing is for the modulations to be independent or uncorrelated. If a triad C, E, G, of pure flute tones is played, in which only the first harmonic partial is of audible amplitude, the corresponding tones will be radiated by loudspeakers 79, 81, and respectively. The phase modulations of each tone will be independent in magnitude and time. At any instant any pair of tones may have phase shifts which are even opposite in direction. When these phase shifts are reflected into the acoustical standing wave system of the listening room, it will sound to the listener as if the three tones in the triad are quite independent of each other. When more complex tones (such as diapason, string, -or reed types) emerge from the resonator bank 128, a majority of the energy of any given tone will generally go to one channel. Howe-ver smaller parts of the energy of the same tone will be routed to other channels (refer to Table 4), and will therefore be modulated somewhat differently. Thus for complex tones the independent modulation of sub-groups of partials, combined With the radiation of these partials from spaced loudspeakers, gives each individual complex tone an apparent location l1 which is variable in time and space about the normal location which would be observed if the modulations were turned off. This adds greatly to the liveness of the musical tone heard by the listener. Moderate amounts of modulation make the tone resemble pipe organ tone. Greater modulation gives a novel, musically interesting elfect.

When the complete audio-frequency spectrum, such as the output of a single division of an electronic organ, is radiated octaphonically, the apparent location of the individual tones will depend upon (a) the number of octaphonic channels (four, six, or twelve), (b) the type of electronic organ (waveform generator, formant, resonator, harmonic synthesis), and (if the organ is a resonator type) (c) the degree of complexity of the tone spectrum radiated. The listeners ability to localize sounds coming from each of the channel loudspeakers will depend upon the angular spread of the loudspeakers, as seen by the listener, and upon the acoustics of the listening space.

For the waveform generating yand formant types of organ all of the harmonics of a given fundamental frequency will be radiated by the same loudspeaker system. Consequently the listener will localize the tonal sources more definitely than he would when listening to the octaphonic output of a resonator organ, for which the motion effects of the one organ division are shown in FIGURE 13. Each square in the diagram represents the location of a channel loudspeaker identified by the channel number underneath. No Ivertical motion is in- Itended to be implied by the four horizontal bands lwithin each square. This is simply a convenience for showing what the situation is with regard to each of the four note-groups in a four-channel octaphonic system. Table 4 has previously shown how the harmonics are distributed among the four output channels for a complex tone having twenty harmonics. If only the fundamental has an audibly appreciable amplitude, the listening situation is the same as for waveform generating or formant types of organs. .Each tone will come from the particular loudspeaker associated with its note-group. At the other extreme are tones having many prominent harmonics. Seldom is the spectrum so highly developed as a sawtooth tone valve, in which the spectrum level has a negative slope of six db per octave. More stringent tones than the saw-tooth Wave tend to be musically unpleasant, and the unfiltered saw-tooth itself borders upon this condition. In FIGURE 13 the saw-tooth wave has been assumed as a kind of upper limit on the distribution of energy into the higher harmonics for calculating a center of gravity for the power distributed ywithin a highly complex tone wave, taking into account the harmonic distribution shown in Table 4. For notes in Groups III and IV 84% of the power radiates from the loudspeaker of the same designation. Ten percent of the power radiates from the adjacent end loudspeaker, and four percent from the other end loudspeaker, with a mere two percent from the fourth loudspeaker. Although the spectral energy is somewhat spread, the center of gravity is squarely in the center of the appropriate loudspeaker. The situation is slightly diiferent for channels II and V. Here there are no loudspeakers on one side to provide a balancing contribution. Consequently the centers of gravity for a highly complex tone lie slightly inside the actual loudspeaker locations. For less complex tones, in general, the center of gravity will lie between the center of the end loudspeakers and the dots shown for the saw- The calculations on energy center of gravity are not intended to imply that a listener will localize the sounds strictly on a basis of energy centroid. The ability of listeners to localize sound distributed in space, and especially the mechanisms by which this is accomplished, are still not fully understood. However, the difference between an octaphonic tone radiation system and a fourparallel-channel tone radiation system (in which all of the tones are radiated equally from all of the loudspeakers) can easily be heard and demonstrated. In the octaphonic system musical denition is greatly enhanced, and the sounds are more interesting, even to the musically naive observer. This advantage is in addition to the freedom from intermodulation distortion, .and the freedom from the undesirable effects of arrays of multiple loudspeakers radiating the same sounds.

FIGURE 14 is an example of the use of the octaphonic principle in a multi-manual electronic organ. It shows an advantageous distribution of the loudspeakers in a three-manual organ consisting of Great, Swell, and Choir divisions. The Pedal division has been omitted from the discussion, but it could be combined with the Great division, or it could have its own widely distributed loudspeakers. The purpose here is to make the Great division sound large in comparison to the Choir division, with the Swell division intermediate in apparent size. The dashed lines enclose the Choir division and the Swe-ll division. The horizontal spread is much greater in the Swell division than in the Choir division. A vertical spread has also been provided for the Great division by duplicating its loudspeakers at a lower altitude. However an even better arrangement than the one shown in FIGURE 14 would have the Great division divided into six or twelve octaphonic channels, and spread over the same area shown in FIGURE 14. This would conserve the advantage in directional characteristics while enlarging the Great divsions apparent size. Note that the group numbers have the reverse numerical order, left to right, in the Great division. This balances one division against the others when the full organ is played, utilizing the full solid angle available, no matter what chords or progressions are most probable in the key of the musical selection being played.

Many other refinements and extensions of the invention within the scope thereof, will occur to those skilled in the art. It is intended therefore that the scope of the present invention not be limited by the vforegoing disclosure, but rather by the appended claims.

Iclaim:

1. In a loudspeaker arrangement for an organ system having a great division and at least one small organ division, comprising an extended array of octaphonically arranged speakers for said small organ division, an array of octaphonically arranged speakers for said great division comprising four speakers arranged -two on one side and two on another side of said first mentioned array.

2. The combination according to claim 1 wherein the rst mentioned array has a predetermined octaphonic group number order and said second mentioned array has the reverse octaphonic group number order.

3. In a music system, a source of music signals comprising a source of plural complex tones, a plurality of string iilters connected in parallel with each other and in cascade with said source of plural complex tone, and means for acoustcally radiating the outputs of said string filters, wherein is provided means for octaphonically combining the outputs of said string filters in separate channels, and means for acoustically radiating the outputs of said channels via separate loudspeakers.

4. In a musical system, a transducer of music signals extending over a wide multiple octave audio spectrum, p

a plurality of quarter octave filters connected in cascade with said source and in parallel with each other,

means for combining in separate channels, outputs of said lters having primarily octavely related partials, and

means for separately transducing the signal content of said separate channels.

5. The combination according to claim 4, wherein said means for separately transducing are loudspeakers.

6. The combination according to claim 4, wherein said 'means for transducing are magnetic tape recording means.

7. The combination according to claim 4, wherein said transducer includes plural magnetic tape recorders, one for each of said quarter octave lters.

8. In a music system,

a transducer of music signals extending over a Wide multiple octave audio spectrum,

means for separating from said music signals into separate channels spectra having primarily octavely related partials which are also immediately adjacent in frequency and which extend over no more than one third of an octave, and

means for separately transducing the signal content of said separate channels, respectively.

9. In a magnetic tape recording and reproducing system,

a source of a Wide band of frequencies representing audio, lter means for dividing said wide band of signals into sub-bands, each of said sub-bands including a quarter octave of frequencies and said sub-bands being substantially non-overlapping and immediately adjacent,

means for combining in separate channels the outputs of those of said filter means which have primarily octavely related partials, means for separately magnetically recording the outputs of said channels as separate tracks of a tape,

means for separately reproducing the signal content of said tracks including a separate reproducer head for reading out each of said tracks, and

a separate loudspeaker coupled to each of said reproducer heads.

10. The combination according to claim 9, wherein said loudspeakers are arranged in a spatial array such that only two of said loudspeakers handling immediately adjacent frequencies are immediately adjacent each other.

11. In a magnetic tape reproduction system, for reproducing from a magnetic tape having a track containing a recorded multi-octave audio signal,

means for reproducing the content of said track into a single channel as a broad band electrical signal,

iilter means for dividing said broad band electrical j References Cited UNITED STATES PATENTS 2,596,258 5/1952 Leslie 84-1.01 2,982,819 5/1961 Meinema et al. 333-71 X 3,327,043 6/1967 Martin 84-1.01 1,550,684 8/1925 Espenschied. 1,961,410 6/1934 Wegel 333-71 X 2,803,800 8/1957 Vilbig 333-71 X HERMAN K. SAALBACH, Primary Examiner S. CHATMAN, JR., Assistant Examiner U.S. Cl. X.R.

4 slt- 1.03, 1,14

Images