U.S. patent number 3,809,792 [Application Number 05/321,231] was granted by the patent office on 1974-05-07 for production of celeste in a computor organ.
This patent grant is currently assigned to Nippon Gakki Seizo Kabushiki Kaisha. Invention is credited to Ralph Deutsch.
United States Patent |
3,809,792 |
Deutsch |
May 7, 1974 |
PRODUCTION OF CELESTE IN A COMPUTOR ORGAN
Abstract
Apparatus is disclosed for producing a celeste effect in a
computor organ of the type wherein musical notes are generated by
computing the amplitudes at successive sample points of a musical
waveshape and converting the amplitudes to notes as the
computations are carried out in real time. Each amplitude is
computed during a regular time interval t.sub.x by individually
calculating and combining at least two sets of discrete Fourier
components. The first set includes harmonically related components,
generally the true pitch fundamental and overtones of each selected
note. Components of the second set are offset slightly higher in
frequency from those in the first set. The resultant synthesized
sound resembles an organ celeste stop wherein two organ pipes, one
tuned slightly sharp with respect to the other, are sounded when a
note is played. In one illustrative embodiment, each set contains
the same number of components, each component in the second set
being slightly higher in frequency than the corresponding component
of the first set. In another embodiment, the first set includes
plural harmonic components, the second set contains only one
component slightly offset from the fundamental of the first
set.
Inventors: |
Deutsch; Ralph (Sherman Oaks,
CA) |
Assignee: |
Nippon Gakki Seizo Kabushiki
Kaisha (Hamamatsu, JA)
|
Family
ID: |
23249745 |
Appl.
No.: |
05/321,231 |
Filed: |
January 5, 1973 |
Current U.S.
Class: |
84/631; 84/DIG.4;
984/397; 84/664; 984/326 |
Current CPC
Class: |
G10H
7/105 (20130101); G10H 1/10 (20130101); Y10S
84/04 (20130101) |
Current International
Class: |
G10H
1/06 (20060101); G10H 1/10 (20060101); G10H
7/10 (20060101); G10H 7/08 (20060101); G10h
001/02 (); G10h 005/02 () |
Field of
Search: |
;84/1.01,1.03,1.22-1.24,DIG.4,DIG.5 |
References Cited
[Referenced By]
U.S. Patent Documents
Primary Examiner: Wilkinson; Richard B.
Assistant Examiner: Witkowski; Stanley J.
Attorney, Agent or Firm: Flam & Flam
Claims
Intending to claim all novel, useful and unobvious features shown
or
1. Apparatus for production of celeste in a computor organ
comprising:
first means, operative during repetitive computation intervals, for
separately calculating a first set of Fourier components associated
with the musical waveshape of a first note of one pitch; and second
means, also operative during said repetitive computation intervals,
for separately calculating a second set of Fourier components
associated with the musical waveshape of a second note having a
pitch slightly offset in frequency with respect to said first
note,
means for combining the calculated components of said first and
said second sets within each computation interval to establish a
sample point amplitude of a resultant musical waveshape the shape
of which varies in time as a result of the frequency difference
between said first and second notes,
means operative at the end of each computation interval for
incrementing the effective sample point for which said resultant
waveshape amplitude is established, and
means for converting said resultant waveshape amplitudes to sounds
in real
2. Celeste production apparatus according to claim 1 wherein said
calculating and combining is performed digitally, wherein said
means for converting includes a digital-to-analog converter and a
sound system for reproducing the output from said converter, and
wherein each component amplitude is established by a set of
coefficients stored digitally, the relative amplitudes of said
components establishing the tonal quality of
3. Celeste production apparatus according to claim 1 wherein
components of said first set are calculated at effective waveshape
sample points separated by qR wherein R is a frequency number
establishing the fundamental period of said first note and q is an
integer incremented at the end of each computation interval and
wherein components of said second set are calculated at effective
waveshape sample points separated by q(R + .delta.) wherein .delta.
is a value designating the amount of frequency
4. Celeste production apparatus according to claim 1
comprising:
a clock for establishing said repetitive computation intervals,
a frequency number memory storing values R which establish the
effective waveshape sample point separation for corresponding
notes,
a .delta. memory storing harmonic offset values .delta. for
selectable notes,
a keyboard for selecting notes to be produced by said apparatus,
actuation of a key on said keyboard causing memory readout of the R
and .delta. values for the selected note,
note interval adders for establishing values of (qR) and q(R +
.delta.) for selected notes during successive computation
intervals, where q is an integer incremented by said means for
incrementing, and wherein said first means comprises
circuitry, cooperating with said note interval adders, for
evaluating said first set F.sub.A.sup.(n) of Fourier components in
accordance with the relationship
F.sub.A.sup.(n) = C.sub.n sin (2.pi./N) nqR
and wherein said second means comprises circuitry, also cooperating
with said note interval adders, for evaluating said second set
F.sub.B.sup.(n) of Fourier components in accordance with the
relationship
F.sub.B.sup.(n) = C.sub.n ' sin (2.pi./N) nq (R + .delta.)
wherein A and B represent the number of Fourier components includes
in said respective first and second sets, components in said second
set being shifted in frequency with respect to said first set by an
amount established by said values .delta., wherein C.sub.n and
C.sub.n ' are coefficients indicating the relative amplitude of the
corresponding n.sup.th component in the respective first and second
set, and wherein N
5. As a musical instrument exhibiting a:
first means for computing at regular time intervals t.sub.x the
amplitudes x.sub.o (qR) of a waveshape, where q is an integer
incremented each time interval t.sub.x, in accordance with the
relationship ##SPC3##
wherein A and B represent the number of Fourier components included
in respective first and second sets defining said waveshape,
components in said second set being shifted in frequency with
respect to components of said first set by an amount established by
.delta., wherein C.sub.n and C.sub.n ' are coefficients
establishing the relative amplitudes of the corresponding n.sup.th
components in the respective first and second sets, wherein R is a
number specifying the period of said waveshape, and wherein N is a
system constant, said first means comprising;
a coefficient memory storing said harmonic coefficients C.sub.n and
C.sub.n ',
a sinusoid table comprising a memory storing values of sin
(2.pi./N) .theta. for 0 .ltoreq. .theta. .ltoreq. N at intervals of
D where D is a resolution constant,
a frequency number memory containing values of R associated with
selectable musical notes, a .delta. memory containing values of
.delta. associated with said notes, and note selection circuitry
for accessing from said frequency number and .delta. memories the
values R and .delta. for each selected note,
harmonic component evaluation circuitry utilizing said coefficient
memory and said sinusoid table to calculate
F.sub.A.sup.(n) = C.sub.n sin (2.pi./N) nqR (n = 1,2, . . .A)
for each of the A Fourier components in said first set in
accordance with the selected value R, and to calculate
F.sub.B.sup.(n) = C.sub.n ' sin (2.pi./N) nq(R + .delta.) (n = 1,2,
. . .B)
for each of the B Fourier components in said second set in
accordance with the selected values R and .delta., and
an accumulator for algebraically summing the calculated values
F.sub.A.sup.(n) and F.sub.B.sup.(n) to obtain each waveshape
amplitude X.sub.o (qR), and
second means responsive to said first means for providing celeste
tones
6. A musical instrument according to claim 5 wherein said
calculations are performed digitally, wherein said second means
includes a digital-to-analog converter and a sound system for
converting said obtained waveshape amplitudes to musical sounds
exhibiting a celeste effect, successive cycles of said obtained
waveshape being of different
7. A musical instrument according to claim 5 wherein said first
means includes:
a clock and counter defining calculation subintervals within said
regular interval t.sub.x, components of said first and second sets
being
8. A musical instrument according to claim 5 together with means
for preventing calculation of components in said second set when
notes having
10. A musical instrument according to claim 9 wherein C.sub.n =
C.sub.n '
11. A musical instrument according to claim 9 wherein C.sub.n
.noteq.
12. A musical instrument according to claim 5 wherein B = 1, the
frequency of the single component in said second set being slightly
higher than the
13. A musical instrument according to claim 5 wherein N represents
the number of waveshape sample points for the tone of lowest
fundamental
14. A musical instrument according to claim 5 wherein the
components of said first set are harmonically related in frequency
to the true pitch of a selected note and wherein each component of
said second set is offset slightly higher in frequency from the
corresponding component of said
15. A musical instrument according to claim 5 wherein the values
.delta. are selected so that the frequency offset of the (n =
1).sup.th component
16. A musical instrument according to claim 5 wherein said first
means includes parallel processing channels for concurrently
calculating components of said first and second sets.
Description
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to production of celeste in a
computor organ.
2. Related Applications
The present invention is related to the inventor's copending U.S.
Pat. applications No. 225,883, filed on Feb. 14, 1972, entitled
COMPUTOR ORGAN and No. 298,365, filed on Oct. 17, 1972, entitled
COMPUTOR ORGAN USING PARALLEL PROCESSING. Those disclosures are
incorporated herein by reference.
3. Description of the Prior Art.
The celeste tones of a pipe organ are produced by a multi-rank set
of pipes. One rank is set to true pitch, producing tones at the
nominally correct 8-foot frequencies. The second rank consists of
like sounding pipes, but tuned sharp with respect to true pitch.
The frequency offset of the second rank is not consistent over the
manual, but typically ranges from about 2 Hz at C.sub.3 (the note
of C in the third octave) to about 4 Hz at C.sub.5. When a note is
played, the listener perceives a pleasant beat note as the sounds
from the two ranks interact. This gives the tone a considerable
warmth.
In conventional electronic organs a celeste effect is obtained
using a separate set of oscillators tuned sharp with respect to the
usual analog tone generators. When mixed electrically or
acoustically, the combined generator output produce a reasonable
semblance of celeste. In another approach, a pseudo-celeste effect
is achieved acoustically by using a slowly rotating speaker to
reproduce the organ tones.
Celeste cannot easily be produced in a digital organ of the type
wherein a stored musical waveshape is repeatedly read from memory
at a rate determined by the fundamental frequency of the note being
generated. (An instrument of this type is shown in the inventor's
U.S. Pat. No. 3,515,792 entitled DIGITAL ORGAN.) A fundamental
characteristic of celeste is an interference or beat effect which
occurs between sounds of slightly different frequencies. To
synthesize this effect requires production of a waveshape which
changes in time. To achieve such synthesis in a system which
repeatedly reproduces the same stored wave form requires two
separate digital organs, one generating a note of true pitch, the
other producing a note of slightly higher pitch. The two notes are
combined, either electrically or acoustically, to produce celeste.
Obviously, such implementation may double the system cost.
The principal object of the present invention is to produce a
celeste effect in a computor organ of the type wherein musical
notes are generated by individually calculating and combining the
Fourier components comprising that note. To accomplish this, at
least two sets of Fourier components, offset slightly in frequency
from each other, are calculated and combined to synthesize each
celeste tone. In effect, this corresponds to generating two notes,
one at the true pitch and another tuned sharp. The resultant
waveshape is not uniformly repititious, but changes in time; it may
be thought of as the superposition of separate waveshapes
associated with two notes of slightly different frequency. When
this resultant waveshape is reproduced acoustically, a remarkably
realistic celeste effect results.
SUMMARY OF THE INVENTION
As described in the above mentioned patent application entitled
COMPUTOR ORGAN, musical notes are produced by computing in real
time the amplitudes at successive sample points of a musical
waveshape, and converting these amplitudes to notes as the
computations are carried out. In accordance with the present
invention, the amplitude at each sample point is obtained by
summing at least two sets of Fourier components, one associated
with the true pitch of the selected note, the other set being
offset, generally slightly higher in frequency therefrom. The two
sets of Fourier components thus may be considered as synthesizing
respectively the true pitch and tuned-sharp ranks of a pipe organ
celeste stop.
In one typical implementation, described in conjunction with FIGS.
1 and 2 below, the first set of Fourier components includes the
fundamental and second through eighth harmonics of the selected
note. These true pitch components are illustrated by the solid
lines in the spectrum of FIG. 2. The second set of Fourier
components includes a fundamental having a frequency slightly
higher than that of the first set, and seven overtones harmonically
related to this shifted fundamental, and hence all offset in
frequency with respect to the first set. The offset or
frequency-shifted components are indicated by broken lines in the
spectra of FIG. 2.
The circuitry of FIG. 1 calculates both the true-pitch and
frequency-offset Fourier components during each computation time
interval t.sub.x. The components are summed to obtain the waveshape
amplitude at the sample point currently being evaluated. The
computations are repeated during successive time intervals t.sub.x
to generate a waveshape which when acoustically reproduced yields a
realistic celeste sound. The use of two component sets each having
eight harmonics is quite satisfactory to synthesize a flute or soft
string voice.
In the alternative embodiment of FIG. 3, a greater number of true
pitch harmonics are generated, as indicated by the solid lines in
the spectrum of FIG. 4. A rich string voice can be synthesized. The
celeste effect is produced by a single harmonic component (shown as
a broken line in FIG. 4) having a frequency slightly higher than
the true-pitch fundamental. The resultant offset celeste rank has a
"sinusoidal" waveform tuned sharp with respect to the first
rank.
BRIEF DESCRIPTION OF THE DRAWINGS
A detailed description of the invention will be made with reference
to the accompanying drawings, wherein like numerals designate
corresponding parts in the several figures.
FIG. 1 is an electrical block diagram of a computor organ
configured to produce a celeste effect with an equal number of
Fourier components in the true-pitch and offset-frequency sets.
FIG. 2 is a harmonic spectrum associated with the computor organ of
FIG. 1.
FIG. 3 is an electrical block diagram of a computor organ
configured for production of celeste and wherein only a single
frequency-shifted component is generated.
FIG. 4 is a harmonic spectrum associated with the computor organ of
FIG. 3.
FIG. 5 is a simplified electrical block diagram of circuitry useful
in conjunction with the computor organ of FIG. 3 for inhibiting
production of celeste for certain selected notes.
FIG. 6 is an electrical block diagram showing celeste generation in
a parallel processing computor organ.
DESCRIPTION OF THE PREFERRED EMBODIMENTS
The following detailed description is of the best presently
contemplated modes of carrying out the invention. This description
is not to be taken in a limiting sense, but is made merely for the
purpose of illustrating the general principles of the invention
since the scope of the invention best is defined by the appended
claims.
Structural and operational characteristics attributed to forms of
the invention first described shall also be attributed to forms
later described, unless such characteristics are obviously
inapplicable or unless specific exception is made.
The computor organ 10 of FIG. 1 produces via a sound system 11
musical notes having a celeste quality. For each note selected by
the keyboard switches 12, the computor organ 10 computes the
amplitudes at successive sample points of a waveshape
characterizing the selected note. Each amplitude is obtained by
calculating two sets of discrete Fourier components as illustrated
in FIG. 2.
Both sets of components are summed algebraically in an accumulator
13 which, at the end of each computation time interval t.sub.x
contains the amplitude for the current sample point. This amplitude
is provided via a gate 14, enabled by the t.sub.x signal on a line
15, to a digital-to-analog converter 16 which supplies to the sound
system 11 a voltage corresponding to the waveshape amplitude just
computed. Computation of the amplitude for the next sample point
subsequently is initiated, so that the analog voltage supplied from
the converter 16 comprises a musical waveshape generated in real
time. The resultant sound, synthesized from true pitch and
frequency-offset harmonic components, realistically simulates a
multi-rank celeste tone.
The period of the computed waveshape, and hence the fundamental
frequency of the generated note, is established by a frequency
number R selected by the keyboard switches 12. A set of such
frequency numbers corresponding to the notes of the instrument is
stored in a frequency number memory 17. Each true pitch Fourier
component F.sub.A.sup.(n) is calculated in accordance with the
following equation:
F.sub.A.sup.(n) = C.sub.n sin (2.pi./N) nqR for q = 1, 2, 3 (1)
where R is the frequency number mentioned above, and n = 1, 2, 3, .
. . A designates the Fourier component during evaluated. The value
n = 1 corresponds to the fundamental, n = 2 to the second harmonic,
n = 3 to the third harmonic, and so forth. The harmonic coefficient
C.sub.n specifies the relative amplitude of n.sup.th Fourier
component. The value of R designates each sample point of the
waveshape being generated.
Similarly, each frequency-offset Fourier component F.sub.B.sup.(n)
is calculated in accordance with the following equation:
F.sub.B.sup.(n) = C.sub.n ' sin (2.pi./N) nq (R + .delta.) for q =
1 2, 3 (2)
where again n = 1, 2, 3, . . . B designates which order Fourier
component is being evaluated. The harmonic coefficient C.sub.n '
specifies the relative amplitude of the n.sup.th Fourier component
in the shifted-frequency set. The value .delta. determines the
extent of frequency-offset with respect to the corresponding
true-pitch component. This value .delta. may be the same for all
notes, or may be different for each note or groups of notes.
Appropriate values of .delta. are stored in a memory 18 (FIG. 1)
accessed in unison with the frequency number memory 17 as each
keyboard switch 12 is selected.
The value N designates the number of amplitude sample points
computed for the note of lowest pitch (fundamental frequency) of
the computor organ 10. Satisfactory synthesis of pipe organ sounds
is achieved using 32 such sample points (N = 32). Preferably the
total number (A + B) of components calculated to synthesize the
waveshape is equal to or less than N/2. This will satisfy the well
known sampling rate requirements (related to the Nyquist criteria)
of a sampled data system. In the embodiment of FIG. 1, the computor
organ 10 calculates eight Fourier components (A = 8, B = 8) for
each of the two sets combined to obtain each waveshape sample point
amplitude. Accordingly, the sample point amplitude X.sub.o (qR) is
given by the relationship: ##SPC1##
which is a form of discrete Fourier representation of a sampled
periodic complex waveshape.
In the embodiment of FIG. 1, equation 3 is implemented by computing
the amplitude value x.sub.o (qR) for each sample point during a
fixed time interval t.sub.x established by a clock 20 and a counter
21. During each interval t.sub.x individual Fourier components are
calculated in successive time intervals designated t.sub.cp1
through t.sub.cp16 respectively. During the first eight intervals
t.sub.cp1 through t.sub.cp8 .sub.' the eight true-pitch components
(solid lines in FIG. 2) are calculated in accordance with equation
(1) above. The eight frequency-shifted components (broken lines in
FIG. 2) are calculated during the subsequent calculation intervals
t.sub.cp9 through t.sub.cp16 .sub.' in accordance with equation (2)
above. All of the calculated components are summed in the
accumulator 13, the contents of which, representing the amplitude
value x.sub.o (qR), is gated to the digital-to-analog converter 16
at the end of the computation cycle t.sub.x.
To this end, the clock 20 provides timing pulses at intervals
t.sub.cp via a line 22 to the counter 21. The counter 21 preferably
is of modulo 16, and provides outputs t.sub.cp1 through t.sub.cp16
on the lines designated with corresponding numbers. The signals
t.sub.cp1 through t.sub.cp8 all are provided via an OR-gate 23 onto
a line 24 to control calculation of the true-pitch components.
Similarly, the signals t.sub.cp9 through t.sub.cp16 all are
supplied via an OR-gate 25 to a line 26 which controls calculation
of the frequency-offset components. The t.sub.cp16 signal, slightly
delayed in a delay unit 27, provides the t.sub.x signal on the line
15 indicating the end of the computation cycle.
To calculate each true-pitch harmonic component, the frequency
number R associated with a selected note is supplied from the
memory 17 via a line 28 and a gate 29 to a note interval adder 30.
The gate 29 is enabled by the t.sub.x signal, so that the contents
of the adder 30 is incremented each computation interval, and
represents the value (qR) designating the waveshape sample point
currently being evaluated.
At each interval t.sub.cp1 through t.sub.cp8 .sub.' the value (qR)
is gated from the adder 30 via a line 32 and a gate 33 to a
harmonic interval adder 34 which is cleared by the t.sub.x signal
at the beginning of each computation cycle. Accordingly, during the
first eight calculation cycles, the contents of the adder 34
represents the value nqR (for n = 1, 2, 3, . . . 8) designating
which true-pitch harmonic component currently is being
evaluated.
An address decoder 35 accesses from a sinusoid table 36 the value
sin (2.pi./N) nqR corresponding to the argument nqR received via a
line 35' from the harmonic interval adder 34. The sinusoid table 36
may comprise a read only memory storing values of sin (2.pi./N)
.theta. for 0 .ltoreq. .theta. .ltoreq. N at intervals of D, where
D is called the resolution constant of the memory.
The value sin (2.pi./N) nqR, supplied via a line 37, is multiplied
by the coefficient C.sub.n for the corresponding n.sup.th harmonic
by a multiplier 38. The multiplication product represents the
amplitude F.sub.A.sup.(n) of the n.sup.th true-pitch harmonic
component, and is supplied via a line 39 to the accumulator 13. The
appropriate coefficient C.sub.n is accessed from a harmonic
coefficient memory 41, described in more detail below, under
direction of a memory address control unit 42 also receiving the
computation interval signals t.sub.cp1 through t.sub.cp8 from the
line 24.
After the eighth true-pitch component has been calculated, the
harmonic interval adder 34 is cleared. To accomplish this, the the
t.sub.cp8 signal, slightly delayed by a delay unit 44, is supplied
via a line 45 to the "clear" input of the adder 34.
To compute the frequency-offset components, the value .delta.
associated with the selected note is accessed from the memory 18
and added to the frequency number R for that note by an adder
circuit 46. The sum (R + .delta.) is supplied to a second note
interval adder 47 via a gate 48 actuated by the computation
interval signal t.sub.x on the line 15. Accordingly, the note
interval adder 47 during each computation interval will contain the
sum q(R + .delta.). This value q(R + .delta.) in effect represents
the sample point of a waveshape having a fundamental slightly
higher in frequency, by an amount designated by .delta., than the
true-pitch fundamental of the same note.
At each interval t.sub.cp9 through t.sub.cp16 the value q(R +
.delta.) is supplied via a line 49 and a gate 50 to the harmonic
interval adder 34. Accordingly, the contents of the adder 34
represents a quanitity nq(R + .delta.) for n = 1, 2, 3, . . . 8
where n now indicates the harmonic order of the frequency-shifted
Fourier components illustrated by the broken lines in FIG. 2.
The memory address decoder 35 now accesses from the sinusoid table
36 the value sin (2.pi./N) nq(R + .delta.) corresponding to the
argument nq(R + .delta.) received from the harmonic interval adder
34 on the line 37. This sin value, supplied via the line 37, is
multiplied by the appropriate harmonic coefficient C.sub.n '
obtained from the harmonic coefficient memory 41. The memory
address control 42 now receives the signals t.sub.cp9 through
t.sub.cp16 on the line 26, insuring that the appropriate values
C.sub.n ' are supplied to the multiplier 38.
The output of the multiplier 38 on the line 39 represents the value
F.sub.B.sup.(n) of the frequency-offset component currently being
calculated. This value is supplied to the accumulator 13 where it
is summed with the previously calculated true-pitch and
frequency-shifted components. When all eight frequency-shifted
components have been evaluated (i.e., after interval t.sub.cp16)
the contents of the accumulator 13 represents the value x.sub.o
(qR) as given by equation (3) above. The t.sub.x signal gates this
value x.sub.o (qR) via the digital-to-analog converter 16 to the
sound system 11, and clears the accumulator 13 in readiness for
computation of the next sample point amplitude. As the computations
are carried out, the sound produced by the system 11 corresponds to
the selected notes with a pleasing celeste effect.
The memory 41 advantageously comprises a read only memory
containing harmonic coefficient values C.sub.n and C.sub.n '
appropriate to produce a note of desired tonal quality. The values
C.sub.n may be the same as, or different from the values C.sub.n '
for like harmonics. In the former instance (C.sub.n = C.sub.n ')
each frequency-offset harmonic component (broken lines in FIG. 2)
will have an amplitude equal to the corresponding true-pitch
component. This in effect will synthesize a pipe organ sound
wherein both celeste ranks are of like tonal quality.
Alternatively, the values C.sub.n may differ from the corresponding
value C.sub.n ', producing a sound wherein the two celeste ranks
have different voices.
The following Table I indicates typical values of C.sub.n and
C.sub.n ' for a flute voice and a soft string voice respectively
wherein both celeste ranks are of like voice (C.sub.n = C.sub.n ')
and for a celeste stop having ranks of different tonal quality
(C.sub.n .noteq. C.sub.n ').
TABLE I
__________________________________________________________________________
Value Stored in Memory
__________________________________________________________________________
Flute Soft String Mixed Voice Harmonic Coefficient (Relative
Amplitude) Decibel Equivalent (Relative Amplitude) db Equivalent
(Relative Amplitude) db
__________________________________________________________________________
Equivalent C.sub.1 127 0 127 0 127 0 C.sub.2 0 -50 40 -10 40 -10
C.sub.3 4 -30 16 -18 16 -18 C.sub.4 0 -50 36 -11 36 -11 C.sub.5 0
-50 6 -27 6 -27 C.sub.6 0 -50 4 -30 4 -30 C.sub.7 0 -50 5 -29 5 -29
C.sub.8 0 -50 1 -44 1 -44 C.sub.1 ' 127 0 C.sub.2 ' Same Same 0 -50
C.sub.3 ' as as 4 -30 C.sub.4 ' C.sub.1 --C.sub.8 C.sub.1 --C.sub.8
0 -50 C.sub.5 ' respectively respectively 0 -50 C.sub.6 ' 0 -50
C.sub.7 ' 0 -50 C.sub.8 ' 0 -50
__________________________________________________________________________
The harmonic coefficient memory 41 and address control 42 together
may be implemented using a single integrated circuit read only
memory such as the Signetics type 8223. Such unit accepts a binary
coded addressing signal. Correspondingly, the counter 21 may
comprise a Signetics type 8281 16-state binary counter, the binary
output of which may be supplied directly to the address control
input of the type 8223 memory. A Signetics type 8250
binary-to-octal decoder may be used in conjunction with the type
8281 counter to provide the separate t.sub.cp1 through t.sub.cp16
signal lines shown in FIG. 1. The type 8223 memory may be
programmed to store the harmonic coefficients listed in Table I
above, or other values of C.sub.n and C.sub.n ' appropriate to
produce other celeste voices.
The frequency number memory 17 and the .delta. memory 18 likewise
may be implemented using the same or separate conventional
integrated circuit read only memories such as the Signetics type
8223. The following table shows typical values for the frequency
number R and .delta. values for the notes between C.sub.3 and
C.sub.5.
TABLE II ______________________________________ Note R .delta.
Frequency Offset of Shifted Fundamental(Hertz)
______________________________________ C.sub.3 0.0341 0.005 2.00 C
.sub.3 0.0361 0.005 2.10 D.sub.3 0.0382 0.006 2.20 D .sub.3 0.0405
0.006 2.25 E.sub.3 0.0429 0.006 2.35 F.sub.3 0.0455 0.006 2.45 F
.sub.3 0.0482 0.006 2.50 G.sub.3 0.0510 0.007 2.60 G .sub.3 0.0541
0.007 2.70 A.sub.3 0.0573 0.007 2.75 A .sub.3 0.0607 0.007 2.85
B.sub.3 0.0643 0.008 2.95 C.sub.4 0.0681 0.008 3.00 C .sub.4 0.0722
0.008 3.10 D.sub.4 0.0765 0.008 3.20 D .sub.4 0.0810 0.009 3.30
E.sub.4 0.0858 0.009 3.40 F.sub.4 0.0909 0.009 3.45 F .sub.4 0.0963
0.009 3.55 G.sub.4 0.1021 0.009 3.60 G .sub.4 0.1081 0.010 3.70
A.sub.4 0.1146 0.010 3.75 A .sub.4 F 0.010 3.85 B.sub.4 0.1286
0.010 3.90 C.sub.5 0.1362 0.011 4.00
______________________________________
in the foregoing table, the frequency numbers are based on N = 32
sample points per period for the note C.sub.7, and assume a
monophonic instrument as shown in FIG. 1. The listed .delta. values
will provide the frequency-offset between the true-pitch and
frequency-shifted fundamental components also specified in Table
II. The .delta. values are a design choice selected to provide a
pleasing celeste. In the example of Table II, different groups of
notes have like frequency offset. As mentioned before, this is not
necessary, and all notes could have the same offset, or each note
could have a different frequency offset.
In the alternative embodiment of FIG. 3, the computor organ 10'
calculates 15 true-pitch Fourier components F.sub.A.sup.(n) (for n
= 1, 2, 3, . . . ,15) and a single component F.sub.B.sup.(1) offset
slightly higher in frequency than the true-pitch fundamental. The
associated harmonic spectrum is shown in FIG. 4. The true-pitch
components are calculated during the time intervals t.sub.cp1
through t.sub.cp15.sub.' and the offset component is evaluated at
the calculation interval t.sub.cp16.
To this end, the corresponding t.sub.cp1 through t.sub.cp15 outputs
from the counter 21' are supplied via an OR-gate 52 and a line 53
to the gate 33. Thus the value nqR in the harmonic interval adder
34' is incremented at each of these 15 consecutive calculation
intervals. Accordingly, the true-pitch component values
F.sub.A.sup.(n) for n = 1,2, . . .,15 successively are provided on
the line 39' for summation in the accumulator 13. After the 15th
true-pitch component F.sub.A.sup.(15) has been calculated, the
harmonic interval adder 34' is cleared by the t.sub.15 signal,
slightly delayed by a delay unit 54.
The single frequency-offset component is calculated during the
interval t.sub.cp16. At the beginning of each computation cycle,
the value .delta. associated with the selected note is accessed
from the memory 18' and supplied via a gate 55 to an interval adder
56. The value .delta. is added to the previous contents of the
interval adder 56, so that the output on a line 57 represents the
value q.delta.. This is summed with the value qR from the note
interval adder 30 by an adder 58 to obtain the value q(R +
.delta.). At the calculation interval t.sub.cp16, the value q(R +
.delta.) is supplied from the adder 58 via a gate 59 to the
harmonic interval adder 34' upon occurrence of the t.sub.cp16
signal on a line 60. Since the adder 34' previously was cleared by
the delayed t.sub.15 signal, the resultant contents of the adder
34' will be simply q(R + .delta.).
The memory address decoder 35 then accesses from the sinusoid table
36 the value sin (2/N) q(R + .delta.) corresponding to the argument
q(R + .delta.) received from the adder 34'. That sin value,
provided via the line 37', is multiplied by the corresponding
coefficient C.sub.1 ' to provide the value F.sub.B.sup.(1) =
C.sub.1 ' sin (2.pi./N) q(R + .delta.). This value F.sub.B.sup.(1)
is added in the accumulator 13 to the sum of the previously
calculated 15 true-pitch components, to provide the sample point
amplitude ##SPC2##
This value of x.sub.o (qR) then is gated via the digital-to-analog
converter 16 to the sound system 11. Again there results a note
having pleasant celeste characteristics.
FIG. 4 shows a typical harmonic spectrum of the celeste sound
produced by the computor organ 10' of FIG. 3. The 15 true-pitch
components are indicated by the solid lines, and the single
frequency-offset component by the broken line. The relative
amplitudes of the various components of course determine the tonal
quality of the produced sound. By way of example, a rich string
sound may be produced using the harmonic component values C.sub.n
and C.sub.1' listed in the following Table III. These values are
stored in the harmonic coefficient memory 41' and appropriately
accessed by the memory address control 42' which receives the
calculation interval signals on the lines 53 and 60.
TABLE III ______________________________________ Harmonic Value
Stored in Memory Coefficient Rich String Voice (Relative (Decibel
Amplitude) Equivalent) ______________________________________
C.sub.1 80 -4 C.sub.2 80 -4 C.sub.3 101 -2 C.sub.4 127 0 C.sub.5 32
-12 C.sub.6 36 -11 C.sub.7 25 -14 C.sub.8 18 -17 C.sub.9 28 -13
C.sub.10 16 -18 C.sub.11 13 -20 C.sub.12 7 -25 C.sub.13 5 -28
C.sub.14 3 -33 C.sub.15 3 -33 C.sub.1 ' 127 0
______________________________________
Celeste may be implemented for all notes of the organ, or only for
some notes. Thus in the embodiment of FIG. 3, celeste is produced
for each note between C.sub.3 and C.sub.5. Celeste may be
inhibited, as by appropriate logic 62, when a note between C.sub.1
and B.sub.3 or between D.sub.5 and C.sub.7 is selected.
Illustrative celeste inhibit circuitry 62 is shown in FIG. 5. The
lines C.sub.1 and B.sub.3 and B.sub.5 through C.sub.7 from the
corresponding keyboard (or pedal) switches 12 are supplied on an
OR-gate 63. When a note between C.sub.3 and C.sub.5 is played, a
low output is present on the line 64 from the OR-gate 63,
indicating that celeste is to be implemented. This low signal is
inverted by an inverter 65 to produce on a line 66 a high signal
which enables a pair of AND-gates 67, 68. The gates 67, 68 thus
provide the t.sub.cp15 and t.sub.cp16 signals respectively to the
delay unit 54 and the gate 59, as shown in FIG. 3. Normal celeste
production occurs.
When a note between C.sub.1 and B.sub.3, or between B.sub.5 and
C.sub.7 is played, the output of the OR-gate 63 on the line 64 is
high. This functions as described below to inhibit celeste
production. During the calculation interval t.sub.cp16 the offset
harmonic component F.sub.B.sup.(1) is not generated. Instead, a
16th (n = 16) true-pitch harmonic F.sub.A.sup.(16) is produced.
When the output of the OR-gate 63 is high, the output of the
inverter 65 is low, and the AND-gates 67, 68 are disabled. The
t.sub.cp15 is not supplied to the delay unit 54, hence the harmonic
interval adder 34' is not cleared at the end of the t.sub.cp15
interval. Further, the high signal on the line 64 enables an
AND-gate 69, which provides the t.sub.cp16 pulse via an OR-gate 70
to the gate 33. As a result, during the time interval t.sub.cp16
the value (qR) is added to the contents of the harmonic interval
adder 34', so that the contents becomes nqR = 16qR. As a result,
the sin value corresponding to that argument (16qR) is accessed
from the sinusoid table 36 and to the harmonic amplitude multiplier
38.
Similarly, the t.sub.cp16 signal is provided via the AND-gate 69 to
the memory access control 42'. This causes access from the harmonic
coefficient memory 41' of the value C.sub.16 (that is, the harmonic
coefficient for the 16th true-pitch harmonic). As a result, the
true-pitch harmonic F.sub.A.sup.(16) is provided to the accumulator
13. The resultant waveshape is obtained from 16 true-pitch
harmonics and no frequency-offset components; this corresponds
exactly to the production of a true-pitch note without celeste.
As shown in FIG. 6, production of a celeste readily is implemented
in a computor organ 75 using parallel processing. The organ 75,
like the instrument of FIG. 1, calculates the same number of
true-pitch and frequency-shifted components. The advantage of using
parallel processing is that both sets of Fourier components are
calculated concurrently, so that the system clock rate may be
one-half that required for the computor organ 10 of FIG. 1. As
discussed in the above mentioned patent application entitled
COMPUTOR ORGAN USING PARALLEL PROCESSING, this significant
reduction in computation rate more readily permits the computor
organ to be implemented using conventional integrated
circuitry.
Referring to FIG. 6, the computor organ 75 includes a first
processing channel 76a in which the values F.sub.A.sup.(n) for the
true-pitch components are calculated, and a second, like parallel
processing channel 76b wherein the values F.sub.B.sup.(n) are
calculated for the frequency-shifted components. System timing is
established by a clock 77 having a rate one-half that of the clock
20 in FIG. 1. The output pulses t.sub.cp ' from the clock 77
advance a binary counter 78 of modulo 8. The output of the counter
78 on the lines 79a, 79b, 79c comprises a binary signal
representing the respective counts t.sub.cp1 ' through t.sub.cp8
'.
At the first interval t.sub.cp1 ' the low order, true-pitch Fourier
component F.sub.A.sup.(1) is calculated in the channel 76a and
concurrently the low order frequency-shifted component
F.sub.B.sup.(1) is calculated in the channel 76b. These components,
present on the respective lines 80, 81 are summed by an adder 82
and supplied via a line 83 to an accumulator 13, gate 14,
digital-to-analog converter 16 and sound system 11 like that of
FIG. 1. At consecutive intervals t.sub.cp2 ' through t.sub.cp8 '
successive pairs of true-pitch and frequency-shifted components
F.sub.A.sup.(n) and F.sub.B.sup.(n) for values n = 2,3, . . . 8 are
computed, summed in the adder 82 and supplied to the accumulator
13. In this manner, both sets of Fourier components are computed
during eight time intervals t.sub.cp ', each of which intervals
t.sub.cp ' is twice as long as the calculation interval t.sub.cp of
the FIG. 1 system.
The various components of the parallel processing organ 75 will be
recognized by reference to FIG. 1. However, separate harmonic
interval adders 34a, 34b are used to accumulate the totals nqR and
nq(R + .delta.) respectively. Both adders 34a, 34b are cleared by
the t.sub.x signal derived via a delay unit 84 from the t.sub.cp8 '
signal. The values qR from the note interval adder 30a and q(R +
.delta.) from the note interval adder 30b respectively are gated to
the harmonic interval adders 34a and 34b via gates 33a and 33b
enabled at each calculation interval t.sub.cp1 ' through t.sub.cp8
'.
The timing signals t.sub.cp1 ' through t.sub.cp8 ' are derived from
the binary counter 78 output using a binary-to-octal decoder 85.
The eight lines from the decoder 85, containing the respective
signals t.sub.cp1 ' through t.sub.cp8 ' all are connected to an
OR-gate 86 the output of which, on a line 87, enables the gates 33a
and 33b.
Separate harmonic coefficient memories 41a, 41b and associated
address control units 42a, 42b are used in the respective channels
76a, 76b. Each may be implemented using a Signetics type 8223 read
only memory or the equivalent, the address control portion of which
directly receives the binary coded count on the lines 79a - 79c.
The memory 41a contains the true-pitch harmonic coefficients
C.sub.n and the memory 41b stores the coefficients C.sub.n ' for
the frequency-shifted components. These values may correspond to
those set forth above in Table I.
Although the embodiments shown in the drawings each calculate two
sets of Fourier components, the invention is not so limited. Thus
three or more sets of components could be evaluated and summed to
obtain each sample point amplitude. In such case, all three sets
may be slightly offset in frequency from each other. Further, even
in the two set embodiments, it is not required that the components
of either set correspond in frequency to the true pitch of the
selected note. Thus, e.g., one set may be tuned slightly below true
pitch, the other slightly above. Advantageously, but not
necessarily, the musical instruments disclosed herein are
implemented digitally.
TABLE A
__________________________________________________________________________
Conventional Inte- Component grated Circuit* (or (FIG. 1) other
reference) Remarks
__________________________________________________________________________
Frequency (a)SIG 8223 field-pro- number grammable read only memory
17 memory (ROM) [p. 37] Typical values of R (b)TI SN5488A, SN7488A
and .delta. are 256-bit ROM [p. 9-235] listed in Table II of
specification .delta. memory 18 Note interval (a)SIG. 8260
arithmetic adders 30,47 logic element [p. 37] (b)SIG. 8268 gated
full adder [p. 97] (c)TI SN5483, SN7483 4-bit binary full adders
[p. 9-271] (may be connected as shown in Flores.sup.1 Section 11.1
to accumulate sum) Harmonic Same as note interval interval adder 30
adder 34 R+.delta. Adder 46 SIG. 8268 gated full adder Gates
14,29,33, TI SN5408, SN5409 48,50 quadruple AND gates [p. 6-17]
Sinusoid table (a)TI TMS4405 sinusoid 36 and memory table and
addressing address decoder 35 circuitry Roundoff, if required, may
(b)TI TMS4400 ROM be implemented containing 512 words per
Ledley.sup.2 of eight-bits [p. 14-188] section 4-6. programmed to
store sin values Harmonic (a)SIG 8223 read only coefficient memory
which in- See Table I for memory 41 cludes address examplary and
memory control circuitry contents address control 42 (b)TI SN54166
series shift registers [p.9-134] Harmonic (a)May be implemented
Amplitude as shown in appli- Multiplier cation sheet, SIG 38
catalog, p.28 using SIG 8202 buffer registers and 8260 arithmetic
element (b)Also can be implemented using SIG 8243 scaler [p.65]
Accumulator 13 (a)SIG 8268 or TI SN5483, SN7483 full adders
connected as shown in Flores.sup.1, section 11.1 "Accumulators".
Counter 21 (a)SIG 8281 sixteen-state binary counter [p. 123], and a
SIG 8250 binary- to-octal decoder Clock 20 Any oscillator
__________________________________________________________________________
*TI=Texas Instrument Co. [Page references are to the TI "Integrated
Circuits Catalog for Design Engineers", First Edition, January,
SIG=Signetics, Sunnyvale, California [Page references are to the
SIG "Digital 8000 Series TTL/MSI" catalog, copyright 1971 Flores,
Ivan "Computer Logic" Prentice-Hall, 1960 Ledley, Robert "Digital
Computer and Control Engineering" McGraw-Hill, 1960
* * * * *