Chowning FM - Synthesispaper 1973
Chowning FM - Synthesispaper 1973
Chowning FM - Synthesispaper 1973
JOHN M. CHOWNING
(first published: J. M. Chowning. The Synthesis of Complex Audio Spectra by Means of Frequency Modulation. J. Audio Eng. Soc. 21, 7, 1973.)
.
INTRODUCTION FREQUENCY MODULATION
Of interest in both acoustical re-search and electronic music is Frequency modulation (FM) is well understood as applied in
the synthesis of natural sound. For the researcher, it is the ultimate radio transmission, but the relevant equations have not been
test of acoustical theory, while for the composer of electronic applied in any significant way to the generation of audio spectra
music it is an extraordinarily rich point of departure in the domain where both the carrier and the modulating frequencies are in the
of timbre, or tone quality. The synthesis of natural sounds has audio band and the side frequencies form the spectrum directly.
been elusive; however, recent research in computer analysis and In FM, the instantaneous frequency of a carrier wave is varied
synthesis of some tones of musical instruments has yielded an according to a modulating wave, such that the rate at which the
insight which may prove to have general relevance in all natural carrier varies is the frequency of the modulating wave, or
sounds: the character of the temporal evolution of the spectral modulating frequency. The amount the carrier varies around its
components is of critical importance in the determination of average, or peak frequency deviation, is proportional to the
timbre.1 amplitude of the modulating wave. The parameters of a
In natural sounds the amplitudes of the frequency components frequency-modulated signal are
of the spectrum are time-variant, or dynamic. The energy of the
components often evolves in complicated ways, in particular, c = carrier frequency or average frequency
during the attack and decay portions of the sound. The temporal m = modulating frequency
evolution of the spectrum is in some cases easily followed as with d = peak deviation.
bells, whereas in other cases not, because the evolution occurs in
a very short time period, but it is nevertheless perceived and is an The equation for a frequency-modulated wave of peak amplitude
important cue in the recognition of timbre. Many natural sounds A where both the carrier and modulating waves are sinusoids is
seem to have characteristic spectral evolutions that, in addition to
providing their "signature," are largely responsible for what we e = Asin (!t + I sin "t ) (1)
judge to be their lively quality. In contrast, it is largely the fixed where
proportion spectrum of most synthesized sounds that so readily e = the instantaneous amplitude of the modulated carrier
imparts to the listener the electronic cue and lifeless quality. ! = the carrier frequency in rad/s
The special application of the equation for frequency ! = the modulating frequency in rad/s
modulation, described below, allows the production of complex
I = d/m = the modulation index, the ratio of the peak
spectra with very great simplicity. The fact that the temporal
deviation to the modulating frequency.
evolution of the frequency components of the spectrum can be
easily controlled is perhaps the most striking attribute of the
It is obvious that when I = 0 the frequency deviation must also be
technique, for dynamic spectra are achieved only with
zero and there is no modulation. When I is greater than zero,
considerable difficulty using current techniques of synthesis. At
however, frequencies occur above and below the carrier
the end of this paper some simulations of brass, woodwind, and
frequency at intervals of the modulating frequency. The number
percussive sounds are given. The importance of these simulations
of side frequencies that occur is related to the modulation index in
is as much in their elegance and simplicity as it is in their
such a way that as I increases from zero, energy is "stolen" from
accuracy. This frequency modulation technique, although not a
the carrier and distributed among an increasing number of side
physical model for natural sound, is shown to be a very powerful
perceptual model for at least some.
AMPLITUDE for an index greater than 2.5, the Bessel functions (Fig. 2) will
yield a negative scaling coefficient for some components.
I=0 0.5 Ordinarily, these negative signs are ignored in plotting spectra, as
FREQ. in Fig. 1, since they simply indicate a phase inversion of the
a. frequency component, —sin(θ) = sin(—θ). In the application of
c
FM described below, this phase information is significant and
must be considered in plotting spectra.
I=1 0.5
b. 1.0
I=2 0.5
0.6 First order side-band component J1
c.
c-4m … c … c+4m 0.4
0.5 0.2
d.
0
c-5m … c … c+5m
e.
-0.4
0.5 c = 100 Hz
m = 100 Hz
I=4 I=4
f 2f 3f 4f 5f 6f 7f 8f 9f
-m c +m
I=1
f 2f 3f 4f 5f 6f 7f 8f 9f
-m c +m
I=2
f 2f 3f 4f 5f 6f 7f 8f 9f
-m c +m
I=3
f 2f 3f 4f 5f 6f 7f 8f 9f
-m c +m
I=4
f 2f 3f 4f 5f 6f 7f 8f 9f
-m c +m
DYNAMIC SPECTRA
Fig. 5. Plot of spectrum where the ratio of c/m is 4/1. As As demonstrated above, the equation for FM has an inherent
the index increases, the reflected lower side frequencies and, as will be shown, most useful characteristic: the complexity
begin to affect the spectrum when I = 3, where the of the spectrum is related to the modulation index in such a way
fundamental, c—3m, is noticeably greater than the 7th that as the index increases, the bandwidth of the spectrum also
harmonic, c+3m. increases (see Fig. 5). If, then, the modulation index were made
to be a function of time, the evolution of the bandwidth of the
an increasing index. The fundamental only becomes significant spectrum could be generally described by the shape of the
when the index is greater than two. function. The evolution of each of the components of the
Inharmonic spectra will result from ratios of irrational numbers, spectrum, however, is determined by the shape of the Bessel
e.g., c / m = 1 2 , ! 3 , 1 e . In this case, the reflected side functions. Therefore, if the index increases with time the overall
frequencies will fall between the positive components, thus bandwidth will also increase, but a given component will either
forming a spectrum whose components are not in a relation of increase or decrease in amplitude depending on the slope of the
simple ratios. Fig. 6 shows an adjusted spectrum where the ratio Bessel function at that index range. Fig. 7 is a three dimensional
c ! 1 and the index = 4.* representation of a dynamic FM spectrum where c/m = 1/1 and
m 2 the modulation index increases in time from O to 4. If the index
sweeps over a very large range, for example from 2 to 10, the
* amplitudes of the components will oscillate around O amplitude
Fig. 6 has been expanded to show the reflected side as the bandwidth of the spectrum increases.
frequencies as in Fig. 4. [1-27-2010 jc]
IMPLEMENTATION
The research described here was done using a Digital
Equipment Corporation PDP-10 computer for which there is a
special sound synthesis program designed to make optimum use
of the time-sharing capability of the machine. Implementation of
this research, however, will be described for MUSIC V, a sound
The presence of reflected side frequencies in a dynamic synthesis program that is both well documented and generally
spectrum enormously complicates the evolution of the individual available.4
components, to the extent that it is difficult to visualize them MUSIC V is a program that generates samples or a numerical
through time with any precision. It is possible to gain an intuitive representation of a sound pressure wave according to data, which
feeling for their tendency of change, which in the research specify the physical characteristics of the sound. The samples are
presented here, has proven to be largely sufficient.† Certainly the stored on a memory device as they are computed. On completion
complexity in the evolution of each of the components of the of the computation, the samples are passed at a fixed rate
spectrum makes an important contribution to the lively quality of (sampling rate, which is typically 10 000 to 30 000 samples/sec)
†
A dynamic computer display program was very helpful in to a digital-to-analog converter, which generates a sequence of
visualizing the spectra that result from a changing index and voltage pulses whose amplitudes are proportional to the samples.
reflected side frequencies. Given a ratio of carrier to modulating The pulses are smoothed by a low-pass filter and passed to an
frequencies and an initial and terminal index the program plots audio system.
the changing spectrum.
The program is designed so that the computation of the samples in Fig. 9, must be able to produce a wave that results from taking
is done by program blocks called unit generators. A typical unit the sine of an angle that decreases as well as increases with time.‡
generator is the oscillator that has two inputs, an output, and a In order to specify the modulation index as a function of time
stored waveshape function. The first input specifies the amplitude and control the attack and decay of the modulated carrier, it is
of the output, the second input the frequency of the output, and necessary to alter the instrument, Fig. 9, by adding three more
the function determines the shape of the output. The value of an unit generators. In Fig. 10, u.g. 4 and u.g. 5 are time-domain
input can either be specified by the user or can be the output from function generators (oscillators or envelope generators in MUSIC
another unit generator, thereby allowing multilevel operations on V). U.g. 4 imposes an amplitude envelope on the modulated
waveforms. A collection of interconnected unit generators is carrier and u.g. 5 and u.g. 6
called an instrument, which is supplied data through a set of
parameters, P1 to Pn, set by the user. P1 and P3 are reserved for
beginning time and duration of the note the instrument is to play P4 1/P3 P7XP6 (P8-P7)XP6 P6
AMPLITUDE 1/DUR DEV1 DEV2 MODULATING
and P2 is reserved for the instrument number. The remaining FREQ.
parameters are assigned their function by the user.
Shown in Fig. 9, is an instrument diagram that consists of three
P5
unit generators, two oscillators and an adder. The function for CARRIER
each oscillator is defined to be a sinusoid. This instrument is FREQ.
capable of producing complex FM spectra such as the one in Fig.
4 where the values are now assigned to parameters.
+
P4 = 1000 = amplitude of modulated carrier (arbitrary u.g.6
u.g.4
scaling)
P5 = 100 Hz = carrier frequency
P6 = 100 Hz = modulating frequency
P7 = 400 Hz = frequency deviation, for l = 4. u.g.1
P4 P5 P7 P6
AMPLITUDE CARRIER FREQ. MODULATING u.g.3
FREQ. DEVIATION FREQ.
OUTPUT
u.g.1
Fig. 10. FM circuit to produce dynamic spectra. Two
function generators are added, u.g. 4 and u.g. 5, to
produce an amplitude envelope and a modulation
+ index envelope that causes the bandwidth to vary.
u.g.2
u.g.3
‡
The change in code to the oscillator in MUSIC V to allow for a
OUTPUT
decreasing angle is:
Fig. 9. Simple FM circuit as represented in MUSIC V for
notation. 290 IF(SUM—XNFUN) 288, 287, 287
287 SUM=SUM—XNFUN
In the case above, which is typical for this application of FM, the substitute
instantaneous frequency of the modulated carrier at times 290 IF(SUM.GE.XNFUN) GO TO 287
becomes negative. That is, from Eq. 1, the sum of at, a ramp IF(SUM.LT. 0.0) GO TO 289
function, and I sin βt, a sinusoid with amplitude I, can produce a and for
curve that has a negative slope at certain points and, therefore, a GO TO 293
phase angle that decreases with time! This condition occurs when 292 J6=L1 +J3—1
either the ratio of the carrier to the modulating frequency is very substitute
small or the modulation index is very large. The oscillator, u.g. 3, GO TO 293
287 SUM=SUM—XNFUN
GO TO 288
289 SUM=SUM+XNFUN
GO TO 288
292 J6=LI+J3—1.
P5 = 900 Hz
P6 = 300 Hz
1.00
P7 = 0 0
P8 = 2.
P5 = 500 Hz 15 sec
P6 = lOO Hz
P7 = 0 Fig. 14. Exponentially decaying envelope for bell-like
P8= 1.5. timbres
Another reed quality can be produced by choosing a ratio of c/ m Bell-like sounds can be produced by making the change of the
that yields the odd harmonics. The parameters index directly proportional to the amplitude envelope. Fig. 14,
then, is the function for the amplitude and index. The parameters
P5 = 900 Hz can be the following:
P6 = 600 Hz
P7 = 4 P3 = 15 seconds
P8 = 2 P4 = 1000
P5 = 200 Hz
will produce a clarinet-like timbre where 300 Hz is the P6 = 280 Hz
fundamental and the index is inversely proportional to the p7 = 0
amplitude function. The bandwidth of the spectrum will decrease P8= 10.
as the amplitude increases during the attack.
In all of the above examples, small alterations can be made The ratio c/m = 1/1.4 results in an inharmonic relation of the
which make the sounds more interesting and/or realistic. A frequency components. With the large initial index, the spectrum
particularly useful alteration is the addition of a small constant to is dense and as the amplitude decreases the spectrum becomes
the modulating frequency. If the value 0.5 Hz were added, for gradually simple. As the amplitude approaches 0, the predominant
example, the reflected lower side frequencies would not fall frequency is the carrier at 200 Hz. By changing the amplitude
exactly on the upper-side frequencies, producing a beat frequency function to that shown in Fig. 15, and with the following
or tremulant of 1 Hz. The realism can be further improved by parameters, a drum-like sound can be produced.
making the function controlling the index the same as the
amplitude function only through the attack and steady state and, P3 = .2
thereafter, remaining constant. If Fig. 12 is the shape of the P5 = 200 Hz
amplitude function, then Fig. 13 would be the shape of the index P6= 280 Hz
function. The evolution of the spectrum during the attack is P7 = 0
apparently not always reversed during the decay. P8 = 2.
1.00 The principal difference from the bell sound, in addition to the
0 short duration, is the vastly reduced initial bandwidth of the
spectrum.
0.50
1.00
0
P3 = 2 The above examples are intended to give some feeling for the
P5 = 80 Hz power and economy of means in FM synthesis, although they by
P6 = 55 Hz no means exhaust the potential of this instrument. With an
P7 = 0 additional five unit generators, as shown in Fig. 17, further
P8 = 25. control can be gained over the spectrum. U.g. 10 provides another
carrier wave, but uses the same modulating oscillator. The
1.00 frequency deviation (proportional to the index) can be scaled up
0 or down by the multiplier, u.g. 8. Since the second-carrier
frequency, Pl2, is independent, it can be set to be a multiple of the
0.50 first-carrier frequency and therefore add components in another
region of the spectrum. The proportion of the two modulated
carriers is determined by the multiplier, u.g. 7, which scales the
amplitude before it is applied to the second carrier. The outputs
are mixed by the adder, u.g. 11. With the parameters
REFERENCES
THE AUTHOR
John M. Chowning was born in Salem, N.J., in 1934. He
1 received the B.Mus. degree in composition from Wittenberg
Jean-Claude Risset and Max V. Mathews, "Analysis of Musical University in l9S9 after military service. He studied
Instrument Tones," Physics Today, vol. 22, no. 2, pp. 23-30 composition in Paris for three years with Nadia Boulanger. In
(1969). 1966, he received the doctorate in composition from Stanford
2
Frederick E. Terman, Radio Engineering, pp. 483-489 University, where he studied with Leland Smith. With the help
(McGraw-Hill, New York, 1947). of Max Mathews of Bell Telephone Laboratories, in 1964 he
3
Murlan S. Corrington, "Variation of Bandwidth with set up a computer music program at Stanford. Dr. Chowning
Modulation Index in Frequency Modulation," Selected Papers on currently teaches music theory and computer-sound synthesis
Frequency Modulation, edited by Klapper (Dover Publications, and composition at Stanford’s Department of Music.
1970). [Based upon the work presented in this paper, Stanford
4
Max V. Mathews, The Technology of Computer Music (The University began in 1972 a relationship with Yamaha in
MIT Press, Boston, 1969). Japan, which led to the most successful synthesizer series in
5 the history of electronic musical instruments. The author
Jean-Claude Risset, op. cit.
retired from Stanford University in 1996. He can be reached
at [email protected] digital version 10/30/2001].