Tone Synthesis Method Using Modulation Operation for an Electronic Musical Instrument

This invention relates to a tone synthesis method in which a musical tone signal is synthesized by a frequency modulation operation or an amplitude modulation operation and, more particularly, to a tone synthesis method capable of controlling a relatively large number of frequency components by a simple operation.

Chowning, U.S. Pat. No. 4,018,121 discloses a fundamental technique for synthesizing a tone signal having a desired harmonic composition by a frequency modulation operation in which the carrier and modulating frequencies are both in the audio frequency range. For synthesizing a tone of a satisfactory tone color having sufficient harmonic components by employing the technique disclosed in this United States Patent however, a frequency modulation (abbreviated as FM hereunder) operation using a simple monomial expression is insufficient and it requires an FM operation of a multiplex or polynomial expression. This inevitably necessitates a synthesizing circuit with a complicated and large-scale construction and, in a system which performs the synthesis operation for each term on a time shared basis, a control clock of a high rate must be employed with a resulting high manufacturing cost. A tone synthesis technique using this type of multiplex or polynomial FM operation is disclosed in U.S. Pat. No. 4,253,367.

It has also been conceived, as a technique for synthesizing a tone containing abundant harmonic components by a relatively simple operation, to prestore a waveshape containing abundant frequency components in a waveshape memory and use its output as a modulation wave or a carrier wave. Since a waveshape usable for such operation is fixed to the one waveshape which has been stored in a waveshape memory, there is limitation in a tone color which can be synthesized by this technique. The same problem exists not only in the tone synthesis technique of the FM operation type but also in that of an amplitude modulation operation (abbreviated as AM hereunder) type.

It is an object of the present invention to provide, a technique of synthesizing a tone signal by a predetermined modulation operation, a method capable of synthesizing a tone signal having abundant frequency components with a relatively simple construction.

The tone synthesis technique according to the invention is characterized in that waveshape data is stored in a logarithmic form in a waveshape table used for generation of a modulation wave function or a carrier wave function, a multiplier means is provided for multiplying the logarithmic waveshape data read out from this table with any desired coefficient, changing a function of waveshape data which is antilogarithm of this logarithm by this coefficient multiplication and utilizing the changed function for the modulation operation. If a modulation wave or a carrier wave prepared in a waveshape table f is expressed by f(.omega.t), its logarithmic expression is log f(.omega.t). If this log f(.omega.t) is multiplied with a pre-established coefficient k, i.e., k.multidot.log f(.omega.t)=log {f(.omega.t)}.sup.k, then the antilogarithm of k.multidot.log f(wt) becomes {f(.omega.t)}.sup.k which is a function having a waveshape that is quite different from the original function f(.omega.t), containing more frequency components than the original function. The waveshape of the function {f(.omega.t)}.sup.k obtained which can also be expressed as f(wt).multidot.[f(wt)].sup.k-1 is not limited to a single waveform having fixed harmonic content but may be varied by simply changing the value of the supplied waveshape changing coefficient k.

An advantageous result derived by practicing the above operation is that a tone signal containing abundant frequency components can be synthesized with a simple modulation operation (changing the supplied k), a predetermined waveshape of a modulation wave or a carrier wave prestored in a waveshape table is changed to a new waveshape containing more frequency components by a very simple operation and this changed waveshape is used for the modulation operation. Further, since the waveshape itself of the modulation wave or carrier wave can be changed by merely changing the value of the supplied coefficient k, a tone synthesis control for generating various tone colors can be realized with a very simple construction.

In the accompanying drawings:

FIG. 1 is an electrical block diagram showing an embodiment of the present invention applied to the FM operation type tone synthesis technique;

FIG. 2 is an electrical block diagram showing an example of a circuit for supplying operation parameters used in the circuit of FIG. 1;

FIG. 3 is a block diagram showing a specific example of a carrier wave generation section in FIG. 1;

FIGS. 4a-4d are diagrams showing examples of waveshapes of output data from respective portions of FIG. 3;

FIGS. 5a and 5b are graphs showing an example of a function obtained finally in the circuit of FIG. 3 with respect to different shift amounts (i.e., coefficients);

FIG. 6 is a block diagram showing an example of a modulation function generation section in FIG. 1 which has been modified by applying the present invention; and

FIG. 7 is a block diagram showing an embodiment of the invention applied to the AM operation type tone synthesis technique.

FIG. 1 shows an embodiment of the invention applied to the FM modulation type tone synthesis method. This embodiment is adapted to execute a monomial FM operation equation. The circuit generally comprises a modulation wave function generation section 10, a carrier wave function generation section 11 and an adder 12 for phase-modulating a carrier wave. In the modulation wave function generation section 10, sine waveshape data sin .omega..sub.m t is read out from a sine wave table 13 in response to a modulation wave phase angle data .omega..sub.m t and thereafter is multiplied with modulation index data I(t) in a multiplier 14. In the adder 12, modulation wave data I(t) sin .omega..sub.m t provided by the multiplier 14 is added to carrier wave phase angle data .omega..sub.c t for performing phase-modulation of the carrier wave. The carrier wave function generation section 11 generates a predetermined carrier wave function in accordance with phase angle data of the phase-modulated carrier wave .omega..sub.c t+I(t) sin .omega..sub.m t provided by the adder 12 and, as a result, produces a frequency modulated signal.

In the embodiment of FIG. 1, the present invention is applied to the carrier wave function generation section 11. Waveshape data of a sine wave is prestored, in logarithm, in a waveshape table 15 and this logarithmic waveshape data is read out in response to phase angle data provided by the adder 12. A shift circuit 16 constitutes multiplication means for multiplying the logarithmic waveshape data read out from the waveshape table 15 with a coefficient k. The shift circuit 16 performs multiplication of k=2.sup.i (where i is any integer representing a shift amount) by shifting bits of the logarithmic waveshape data of binary digital data to the right (i.e., to less significant bit) or to the left (i.e., to more significant bit). The shift amount of the shift circuit 16 is designated by shift data SFT. If the phase angle data .omega..sub.c t+I(t) sin .omega..sub.m t provided from the adder 12 to the waveshape table 15 is represented by .theta., the read out output of the table 15 is log sin .theta. and the output of the shift circuit 16 is 2.sup.i log sin .theta.=log (sin .theta.).sup.2.spsp.i whereby the carrier wave function which corresponds to antilogarithm of the logarithmic data has substantially been changed from sin .theta. to (sin .theta.).sup.2.spsp.i.

The logarithmic waveshape data provided by the shift circuit 16 is applied to an adder 17 where amplitude weighting of the data is made using amplitude coefficient A(t). More specifically, amplitude coefficient data log A(t) expressed in logarithm is applied to the adder 17 and, by addition of logarithms, multiplication of antilogarithms is substantially effected, i.e., log (sin .theta.).sup.2.spsp.i +log A(t)=log A(t).multidot.(sin .theta.).sup.2.spsp.i. The logarithmic waveshape data thus having been weighted in amplitude is applied to a logarithm-linear converter 18 where it is converted to waveshape data in the linear form (i.e., antilogarithm).

The various parameters .omega..sub.m t, .omega..sub.c t, I(t), log A(t) and SFT used for the FM operation are provided by a circuit as shown in FIG. 2. A keyboard circuit 19 detects a key depressed in a keyboard of an electronic musical instrument and thereupon produces depressed key data. A phase data generation circuit 20 generates, in response to the depressed key data provided by the keyboard circuit 19, modulation wave phase angle data .omega..sub.m t and carrier wave phase angle data .omega..sub.c t at a period corresponding to the tone pitch of the depressed key. An envelope generator 21 generates, in response to depression of the key, modulation index data I(t) and amplitude coefficient data log A(t) as functions of time. Tone color selection data is supplied from a tone color selection device (not shown) to the phase data generation circuit 20 and the envelope generator 21 and the frequency ratio between .omega..sub.c t and .omega..sub.m t and time functions of I(t) and A(t) are thereby controlled in accordance with the tone color. A shift data generation circuit 22 generates shift data SFT by operating a selection switch 23. The shift data generation circuit 23 may be so constructed that it will generate predetermined shift data SFT in response to tone color selection data.

FIG. 3 shows a specific example of the carrier function generation section 11 in FIG. 1. In the waveshape table 15, a quarter period waveshape of a sine wave (FIG. 4a) corresponding to an angular range between 0 and (.pi./2) is stored in logarithm. The most significant bit MSB of the phase angle data .theta. is used as a sign bit indicating polarity of the waveshape data and the second bit MSB-1 counting from the MSB is used for switching reading direction of the waveshape table 15 upon elapse of each quarter period. Exclusive OR gates 24, 25, . . . , 26 are provided for respective bits excluding the two bits MSB and MSB-1 of the phase angle data .theta. and these respective bits are applied to one inputs of these exclusive OR gates 24 through 26. To other inputs of these exclusive OR gates 24 through 26 is commonly applied the second most significant bit MSB-1. This bit MSB-1 is "0" when the first and third quarter periods of the sine waveshape are read out, causing the less significant bits of data .theta. applied to the exclusive OR gates 24-26 to pass through these gates 24-26 and to be applied to an address input of the waveshape table 15. The bit MSB-1 is "1" when the second and last quarter periods are read out, causing the less significant bits of data .theta. to be inverted by the exclusive OR gates 24-26 and thereafter to be applied to the address input of the waveshape table 15. Accordingly, the reading direction of the waveshape table 15 is inverted each quarter period and waveshape data as shown in FIG. 4b is read out in logarithm. This logarithmic waveshape data is shifted to the right or left by a suitable amount in the shift circuit 16, as was described previously, and supplied to the logarithm-linear converter 18 after being weighted by the amplitude coefficient A in the adder 17. By the shifting in the shift circuit 16, the function of waveshape data which is antilogarithm of the logarithmic data is changed to (sin .theta.).sup.2.spsp.i as described above (.theta. is 0.ltoreq..theta..ltoreq..pi. because the control is directed to a half period of a positive polarity in a sine wave).

The respective bits of the waveshape data provided by the logarithm-linear converter 18 are applied to one inputs of exclusive OR gates 27, 28, . . . , 29 provided for these bits. To other inputs of these exclusive OR gates 27, 28 and 29 is commonly applied a sign bit (i.e. MSB of the data .theta.). When the sign bit is "0" (representing the positive polarity), waveshape data is passed as it is whereas the sign bit is "1" (representing the negative polarity), the respective bits of the waveshape data are inverted to produce 1's complement. In this manner, the waveshape data is converted in the exclusive OR gates 27-29 to a data form for which the polarity has been taken into account. If the linear waveshape data produced from the logarithm-linear converter 18 assumes a form as shown in FIG. 4c, the waveshape data for which the polarity has been taken into account assumes a form as shown in FIG. 4d.

An example of a function obtained in accordance with the shift amount in the shift circuit 16 will now be described.

In a case where the read out output from the waveshape table 15 is shifted by one bit to the right, the coefficient 2.sup.i is 2.sup.-1 =1/2 so that the logarithmic waveshape data produced by the shift circuit 16 becomes 1/2 log sin .theta.=log sin 1/2.theta. and the carrier function obtained as antilogarithm of this logarithmic waveshape data can be expressed by employing .sqroot.sin .theta. (.theta. being 0.ltoreq..theta.<.pi.). Since the finally obtained function in the range of .pi..ltoreq..theta.<2.pi. is one obtained by inverting .sqroot.sin .theta. where .theta.is 0.ltoreq..theta.<.pi. to the negative polarity, a waveshape (solid line) which is distorted to the outside of the sine wave (dotted line) as shown in FIG. 5a is obtained.

In a case where the read out output from the waveshape table 15 is shifted by one bit to the left, the coefficient 2.sup.i is 2.sup.1 =2 so that the logarithmic waveshape data produced from the shift circuit 16 becomes 2 log sin .theta.=log sin.sup.2 .theta. and the function obtained as its antilogarithm can be expressed by employing sin.sup.2 .theta.=(1-cos 2.theta./2) (where 0.ltoreq..theta.<.pi.). In the same manner as was previously described, since the finally obtained function in the range of .pi..ltoreq..theta.<2.pi. is one obtained by inverting sin.sup.2 .theta. where .theta. is 0.ltoreq..theta.<.pi. to the negative polarity, a waveshape (solid line) which is distorted to the inside of the sine wave (dotted line) as shown in FIG. 5(b).

Further increase in the shift amount will result in generation of a waveshape which is more distorted than the waveshapes shown in FIGS. 5a and 5b. The function obtained can be generally expressed by (sin .theta.).sup.2.spsp.i in the range of 0.ltoreq..theta.<.pi. and by the inverted form of (sin .theta.).sup.2.spsp.i of 0.ltoreq..theta.<.pi., i.e., -sin {(.theta.-.pi.)}.sup.2.spsp.i in the range of 0.ltoreq..theta.<2.pi..

Since in the carrier function generation section 11, the phase angle data .theta. is .omega..sub.c t+I(t) sin .omega..sub.m t which has already been phase-modulated, the waveshape data provided actually by the logarithm-linear converter 18 assumes a more complicated waveshape than those shown in FIGS. 5a and 5b (waveshapes obtained by FM modulating the waveshapes of FIGS. 5a and 5b with a sine wave).

It is of course possible to apply this invention to the modulation wave function generation section 10 in FIG. 1. In this case, the circuit is constructed in the same manner as the carrier wave function generation section 11 as shown in FIG. 6. More specifically, waveshape data of a sine wave is stored in logarithm in a waveshape memory 15A and this is read out in response to the modulation wave phase angle data .omega..sub.m t. Of course, it is unnecessary to provide the waveshape memory 15A if the waveshape memory 15 in FIG. 1 is on time division basis used commonly to generate the modulation wave function and the carrier wave function. Shift circuit 16A, adder 17A and logarithm-linear converter 18A are the same as those shown in FIG. 1 but an adder 17A adds modulation index data log I(t) expressed in logarithm. The output I(t) (sin .omega..sub.m t).sup.2.spsp.i of the logarithm-linear converter 18A is applied to the adder 12 to modulate the carrier phase .omega..sub.c t. In this case, this invention may be applied to the modulation wave function generation section 10 only and not to the carrier wave function generation section 11. Alternatively, the invention may be applied to both.

As described above, the fundamental FM operation equation in a monomial form originally is A(t) sin {.omega..sub.c t+I(t) sin .omega..sub.m t}, but according to the present invention, it is changed to A(t)[sin {.omega..sub.c t+I(t) sin .omega..sub.m t}].sup.2.spsp.i or A(t) sin {.omega..sub.c t+I(t) sin.sup.2.spsp.i .omega..sub.m t}, or A(t){sin .omega..sub.c t+I(t) sin.sup.2.spsp.i .omega..sub.m t}, or A(t)[sin {.omega..sub.c t+I(t) sin.sup.2.spsp.i .omega..sub.m t}].sup.2.spsp.i (where 0.ltoreq..theta.<.pi.) whereby a tone signal which has more abundant frequency components and in which control of more frequency components is possible can be synthesized.

FIG. 7 shows an embodiment in the tone synthesis method of the AM operation type. This embodiment is adapted to execute a monomial AM operation equation. The invention is adapted to a carrier function generation section 30 which includes a waveshape table 32 storing sine waveshape data in logarithm, a shift circuit 33 shifting read out data SFT from the waveshape table 32 in response to the shift data SFT, and a logarithm-linear converter 34 converting the output of the shift circuit 33 to data in the linear form. Phase angle data .omega..sub.c t of the carrier wave is applied to the waveshape table 32. For the same reason as was previously described, waveshape data (sin .omega..sub.c t).sup.2.spsp.i containing more frequency components is provided by the logarithm-linear converter 34 and this data is applied as a carrier wave signal to a multiplier 35 provided for amplitude modulation.

In a modulation wave function generation section 31, a sine wave table 36 is read in response to the phase angle data .omega..sub.m t of the modulation wave, its read out output is multiplied with modulation index Z(t) in a multiplier 37 and a cosine wave table 38 is read by an output of the multiplier 37. The waveshape data cos {Z(t) sin .omega..sub.m t} read out from the cosine wave table 38 is applied to a multiplier 35 as a modulation wave signal and thereupon the amplitude modulation operation is performed. The output of the multiplier 35 is applied to a multiplier 39 where the amplitude coefficient A(t) is multiplied.

A specific circuit of the carrier wave function generation section is constructed as shown in FIG. 3. This invention may also be applied to the sign wave table 36 or the cosine wave table 37 in the modulation wave function generation section 31. In this case; such table may be constructed of a waveshape table in a logarithmic form, a shift circuit and a logarithmic linear converter.

The waveshape stored in the waveshape tables 15, 15A and 32 is not limited to a sine wave but any desired waveshaped such as a cosine waveshape, a triangular waveshape, a square waveshape or other complicatied waveshape may be stored in logarithm. The shift circuits 16, 16A and 33 may be constructed by a general multiplication means (i.e. multiplier or divider) and a desired coefficient k may be multiplied with the logarithmic waveshape data. Further, the present invention is applicable not only to a tone synthesis method using the monomial FM or AM operation but to any suitable portion of a tone synthesis method using a polynomial, multiplex or circulating type FM or AM operation in any suitable portion.