JPH0248915B2 - - Google Patents

Description

[Detailed description of the invention]

[Industrial Field of Application] The present invention relates to a musical tone signal forming device capable of forming musical tone signals having moving formant characteristics in a simple manner. [Prior Art] A typical method for forming a musical tone signal with desired formant characteristics is to use a filter having characteristics that realize the desired formant, and pass a sound source signal having the frequency of the musical tone to be generated through the filter. There is a method in which a musical tone signal having the desired formant characteristics is outputted from the filter. However, this method has the disadvantage that the configuration is complicated due to the use of a filter. Moreover, in order to realize a moving formant, the filter characteristics must be moved in relation to the frequency of the musical tone signal input to the filter, which makes the process even more complicated. On the other hand, as a method of forming a musical tone signal with desired formant characteristics without using a filter,
There is a method shown in Publication No. 48-53606. Here, by controlling the amplitude of a burst signal oscillated at a constant frequency using a time window waveform with a period corresponding to the pitch of the musical tone, the frequency of the burst signal is set as the center frequency and the frequency is an integer multiple of the pitch of the musical tone. A musical tone signal having formant characteristics as a spectral component is formed. [Problems to be Solved by the Invention] The method using a burst signal as described above has the advantage of a simpler configuration than the method using a filter, but the frequency of the burst signal is independent of the pitch of the musical tone. Since it is constant, there was a problem that only a fixed formant could be realized. This invention was made in view of the above points,
It is an object of the present invention to provide a musical tone signal forming device which can easily form a musical tone signal having moving formant characteristics with a simple configuration. [Means for Solving the Problems] A musical tone signal forming device according to the present invention includes pitch specifying means for specifying the pitch of a musical tone to be generated;
address signal generating means for generating an address signal that repeatedly changes at a rate corresponding to the pitch frequency of the pitch specified by the pitch specifying means; address signal conversion means for converting into an address signal that repeatedly changes at a rate corresponding to a frequency N times (however, N is greater than 1) the pitch frequency of the pitch specified by the high specification means; It has a waveform storage means that stores waveform data,
Vibration waveform generation means for generating a vibration waveform having a frequency N times the pitch frequency by repeatedly reading out stored waveform data from the waveform storage means in accordance with the address signal converted by the address signal conversion means. a control waveform generating means for repeatedly generating a control waveform at a period corresponding to the pitch specified by the pitch specifying means; and controlling the amplitude of the vibration waveform generated by the vibration waveform generating means using the control waveform. and an amplitude control means for controlling the musical tone signal, wherein a musical tone signal having a moving formant characteristic whose formant characteristic changes in accordance with a change in pitch specified by the pitch specifying means is obtained from the amplitude control means. It is characterized by: [Operation] By generating a vibration waveform with its amplitude controlled by a periodic control waveform, the resulting musical tone signal, that is, the formant of the amplitude-controlled vibration waveform, changes the frequency of the vibration waveform to the center frequency. and has a line spectrum harmonized with the frequency corresponding to the repetition period of the control waveform. The vibration waveform has a frequency that is N times the pitch frequency depending on the pitch of the musical tone to be generated, where N is greater than 1. Further, the control waveform has a period related to the pitch frequency. Therefore, the formant characteristics of the obtained musical tone signal have a center frequency that is N times the pitch frequency of the tone signal, and
It has a line spectrum that is in harmony with the pitch. The center frequency changes as the pitch changes, and is always N times the pitch frequency at that time. In this way, the formant characteristic of the musical tone signal obtained becomes a moving formant. [Embodiments] Examples of the present invention will be described below with reference to the drawings. FIG. 1 is a waveform diagram showing an example of a musical sound waveform generated in an embodiment of the present invention, in which the horizontal axis represents time t, the vertical axis represents each waveform, TW is the musical sound waveform, and W ₁ , W ₂ , W ₃ , W ₄ , and W ₅ are control waveforms, for example, rectangular waves.
The period of the musical tone to be generated is T ₀ , and the musical sound waveform TW is Asin4(2π/T ₀ )t from t=0 to t=T ₀ /2.
Between t=T ₀ /2 and t=T ₀ /2+T ₀ /4, the second vibration waveform consists of Asin8(2π/T ₀ )t, and t=T ₀ /4. 2+T ₀ /
Between 4 and t=T ₀ /2+T ₀ /4+T ₀ /8
It consists of the third vibration waveform Asin16(2π/T ₀ )t, from t=T ₀ /2+T ₀ /4+T ₀ /8 to t=T ₀ /
Asin32 between 2+T ₀ /4+T ₀ /8+T ₀ /16
Consists of the fourth vibration waveform (2π/T ₀ )t, where t=
From T ₀ /2+T ₀ /4+T ₀ /8+T ₀ /16, t=T ₀ /
Between 2+T ₀ /4+T ₀ /8+T ₀ /16+T ₀ /32
The fifth vibration waveform is Asin64(2π/T ₀ )t. That is, a vibration waveform consisting of a sine wave represented by Asink(2π/T ₀ )t is generated during a time window with a width of 2T ₀ /K. Here k=4, 8, 16, 32, 64. This is equivalent to passing the sine wave Asink (2π/T ₀ )t, which continues for an infinitely long time, for a time of 2T ₀ /k every T ₀ time, and blocking it at other times, and its meaning. The first one showing the time width 2T ₀ /k
The rectangular waves W ₁ , W ₂ , W ₃ , W ₄ , and W ₅ in the figure are respectively called a time window or a time window function. Now, it is easy to perform Fourier analysis on the musical sound waveform TW shown in Fig. 1, but in order to clarify the general concept, each part of the musical sound waveform TW is oscillated by a sine wave Asink(2π/T ₀ )t. It is decomposed as the product of the waveform and a corresponding time window function with width 2T ₀ /k. Figure 2 is a spectrum diagram explaining the spectral structure of the waveform shown in Figure 1, where the horizontal axis shows the frequency f, the vertical axis shows the intensity of each spectrum, and S ₁ k is the time with a width of 2T ₀ /k. Window spectrum, S ₂ k
is the spectrum of the sine wave Asink(2π/T ₀ )t, S ₃ k
is the spectrum of the product of the above sine wave and the corresponding time window, that is, the spectrum indicating the frequency component of the convolution of the frequency component of the spectrum S ₂ k and each frequency component of the spectrum S ₁ k. The spectrum of a rectangular time window with a width of 2T ₀ /k has _an envelope of (sin
It is well known that it takes the form x/x). When the time window function is repeated with a period of T ₀ , the time window function becomes a line spectrum, and the line spectrum starts from the direct current at f = 0 and becomes f ₀ = 1/
Arranged in frequency intervals of T ₀ . Also, if the time width of the time window function does not change, its repetition period
Even if T ₀ changes, the shape of its spectral envelope does not change. For example, if k=8, the array points of the line spectrum are the points circled on the horizontal axis of FIG. 2 S ₁ k. The frequency components of the spectral envelope and the sine wave Asink as denoted by S ₁ k
(2π/T ₀ ) The convolution with the frequency component of t is the second
This results in a spectrum in the form of an envelope as shown in Figure _s3k . Here, the center frequency k/T ₀ of the formant is the frequency of the vibration waveform, which is k times the frequency 1/T ₀ of the musical tone to be generated. If the value of T ₀ changes, the center frequency of the formant k/
T ₀ also changes and exhibits moving formant characteristics. Figure 3 is a spectrum diagram showing the spectral envelope of the musical sound waveform TW shown in Figure 1, where k=4,
For example, calculate the spectral envelopes S ₄ , S ₈ , S ₁₆ , S 32 , and S ₆₄ for 8, 16, ₃₂ , and 64 in the same manner as described for FIG. The horizontal axis represents the frequency of . To summarize the above, in one embodiment of the present invention, a musical sound waveform shown as TW in FIG. 1 is generated,
The waveform has a spectral envelope as shown in FIG. The musical sound waveform TW in Fig. 1 can be easily generated using a simple digital circuit, and the generated spectral distribution is A/k as shown in Fig. 2, the spectrum S ₃ k.
_As shown _{in FIG .}
A desired spectral distribution can be easily obtained by controlling the amplitudes of the sine waves in the W ₅ portion independently and in a time-division manner using a single control circuit, and the frequency of the vibration waveform that further extends the time window. It has many advantages as a musical sound waveform, such as the ability to realize moving formant characteristics by making it k times the pitch frequency, that is, a predetermined times, in relation to the pitch of the musical tone. Note that, as will be described later, k is not limited to an integer multiple as described above, but may be a non-integer multiple, in short, it is sufficient if it is a number larger than 1. Furthermore, as shown in the spectral envelope S ₁ k in Fig. 2, the time width 2T ₀ /k of the time window and the bandwidth of the spectral envelope (for example, from f=0 to f=
Since it is inversely proportional to the width of the time window (up to k/2T ₀ ), by changing the time width of the time window, the frequency bandwidth corresponding to the time window can be changed. As explained above, for example, the amplitudes of the time windows W ₁ , W ₂ , ₃ , W ₄ , W ₅ in FIG. 1 can be controlled independently and in a time-division manner by a single control circuit. Therefore, changing the time width of the time window means changing the frequency bandwidths that can be independently controlled in a time-division manner. That is, by changing the time width of the time window, the width of the frequency bandwidth that can be independently controlled, in other words, the frequency resolution in controlling the component frequencies can be changed. In the present invention, by setting the time width of the time window to an appropriate value according to design requirements, the independently controllable frequency bandwidth can be adjusted to suit the requirements. FIG. 4 is a waveform diagram showing an example of a musical sound waveform generated in another embodiment of the present invention;
The same symbols as in the figure have the same meaning, and in Figure 1, the sine wave represented by Asink(2π/T ₀ )t is 2T ₀ /
The sine wave represented by Asink(2π/T ₀ )t in Fig. 4 passes through a time window of width k.
It passes through a time window of width T ₀ /k. Further, in FIG. 4, k=2, 4, 8, 16, 32, 64, and the time windows W ₁ , W ₂ , W ₃ , W ₄ , W ₅ in FIG. 8, 16, and 32, and W ₆ is the time window corresponding to k=64.
As is clear from comparing FIG. 4 with FIG. 1, in FIG. 4, the ratio of the time width of the time window to the period of the passing sine wave is halved compared to the case of FIG. In FIG. 1, two periods of the sine wave pass through each time window, whereas in FIG. 4, one period of the sine wave passes through each time window. Therefore, the spectral envelope of the waveform passing through each time window in FIG.
It has twice the frequency bandwidth compared to the case shown in the figure. FIG _. 5 is a spectral diagram showing the shape of the spectral envelope of _the musical sound waveform shown in FIG.
The spectral envelope of the waveform passing through W ₁ to _{W 6} is shown. However, in the same way as shown in Fig. 3, in Fig. 5, if k doubles, the intensity of the spectrum will be halved, and the intensity of the spectrum will increase in each time window from S(W ₁ ) to S(W ₆ ). For example, the spectral strength of S(W ₆ ) is 1/64 of that of S(W ₁ ), but the 5th
In the figure, this decrease in the intensity of the spectrum is corrected to make the drawing easier to read. Next, an example of a circuit for generating the musical tone waveform TW shown in FIG. 4 will be explained. FIG. 6 is a block diagram showing an embodiment of the present invention, in which 11 is a keyboard section, 12 is a frequency information memory, 13 is an accumulator, 14 is a coefficient generator, and 2 is a sine wave memory having a capacity of 512 words. , 3 is a decoder, 4 is an address switching device, 5 is a selector, 6 is a multiplication circuit, 7
8 is a digital-to-analog converter (hereinafter abbreviated as DAC), and 8 is a sound system. A digital value _F0 corresponding to the pitch of the key pressed on the keyboard section 11 is output from the frequency information memory 12, and is accumulated in the accumulator 13 for each clock pulse φ.
Outputs overflow pulse CA from the MSB (most significant bit) carry terminal. The frequency of this overflow pulse CA is designed to be f ₀ .
Here, f ₀ =1/T ₀ is the repetition frequency of the tone waveform TW shown in FIG. 4, and each value is determined corresponding to the pressed key. The upper 10 bits of this accumulator 13 are used for reading out the sine wave memory 2 and for generating gate waveforms in the decoder 3. In FIG. 6, the conductive wires are shaded with short diagonal lines, and numbers enclosed in small circles are appended to indicate the number of bits in one word constituting the digital signal transmitted by the conductive wire. (The same applies to the following drawings.) From the upper 10 bits of the output of the accumulator 13
If the sine wave memory 2 is read using 9 bits excluding the MSB 1 bit as an address, a 2f ₀ sine wave will be obtained as the output. That is, the time window W ₁ shown in FIG.
This is the part corresponding to . As shown in Figure 4, while the time window changes from W ₁ to W ₆ , the sine wave memory 2
A decoder 3 and an address switching device 4 are provided to change the frequency of the sine wave read out in stages. FIG. 7 is a circuit diagram showing an example of the decoder 3. This decoder 3 outputs c ₉ , c ₈ , c 7 , c ₆ , c ₅ , c ₄ among the outputs c ₉ to c ₀ of the upper 10 bits of the accumulator ₁₃ . Input the upper 6 bits shown as W ₁ , W ₂ ,
These are decoders that output time windows shown as W ₃ , W ₄ , W ₅ , and W ₆ , and 301 to 306 are inverters 311 to 316 are AND gates, and small circles are attached to the intersections of the vertical and horizontal lines in the figure. The dotted portion indicates the connection to the input terminal of the corresponding AND gate (similar representations will be used in the following drawings). It is clear that the connections shown in FIG. 7 make it possible to generate the gating waveforms shown in FIG. 4 as time windows W ₁ -W ₆ . FIG. 8 is a circuit diagram showing another example of the decoder 3,
In Fig. 8, 32 is a 7-stage ring counter;
21 to 326 are AND gates, 33 is an OR gate, and c ₉ to _{c 4} are the same as c ₉ to c ₄ in Fig. 7, and
CA is the overflow pulse of the accumulator 13,
KON is a signal indicating the state of a key operated on the keyboard section 11. The ring counter 32 is initialized by the key-on signal KON and the OR gate 3
It is clear that each stage of the ring counter 32 can generate gate waveforms corresponding to the time windows W ₁ to W ₆ similar to the case of FIG. 7. FIG. 9 is a connection diagram showing the internal connections of the address switching device 4, whose input is the upper level of the accumulator 13.
Of the 10 bits (c ₉ to _{c 0} ) of the output, 9 bits (C ₈ to C ₀ ) excluding the MSB (c ₉ ) are used as control inputs in the time window W ₁ to _{W 6} output from the decoder 3. be.
Also, the output of the address switching device 4 is the sine wave memory 2.
The 9 bits from a ₈ to a ₀ are the address of c ₈ to _{c 0.}
The connections _a8 to _a0 are changed according to the time windows _W1 to _W6 as shown in FIG. That is, while W ₁ , c ₈ is
a ₈ and c ₇ are connected to a ₇ respectively, and the sine wave memory 2
A sine wave of 2f ₀ is read out from, for example, W ₂
During this period, C ₇ is connected to A ₈ , C ₆ is connected to A ₇ , and A ₀ is always kept at logic “0”, so sine wave memory 2 is connected to 512
Only every other word, 256 words, will be read out, so a 4f ₀ sine wave will be read out. In this way, the output of sine wave memory 2 has the waveform shown in Figure 4.
It becomes as shown in TW. FIG. 10 is a block diagram showing an example of the coefficient generator 14, in which 140 is a coefficient memory, 141 is a coefficient memory changeover switch, 142 is a counter, 143 is a clock generator that generates relatively low frequency pulses, and 144 is a NAND gate. , 145 is an AND gate. As shown in FIG. 6, when the time windows W ₁ to W ₆ are sequentially switched in the output of the decoder 3, the frequency of the sine wave read out from the sine wave memory 2 via the address switching device 4 is sequentially switched. In synchronization with the switching, the coefficients are sequentially switched and output in the selector 5, and in the multiplier circuit 6, the amplitude of the sine wave of the frequency determined by the time window is controlled and output by the corresponding coefficient. . In the embodiment shown in FIG. 6, six types of coefficients b ₁ to b ₆ respectively corresponding to six types of time windows W ₁ to W ₆ are used.
is input from the coefficient generator 14 to the selector 5. In order to generate a musical tone with the desired timbre, it is necessary to set these coefficients to appropriate values.
In addition, in order to achieve temporal changes in timbre like in natural instruments, it is necessary to change each coefficient as an appropriate time function.
Furthermore, in order to change the type of timbre that is generated, it is necessary to change the form of the time function of each coefficient. In response to such requirements, the coefficient memory 140 shown in FIG. 10 includes a plurality of sets of coefficients b ₁ to b ₆ that output values that change in response to changes in address.
It is assumed that the performer determines in advance which set of coefficients b ₁ to _{b 6} to select by switching the coefficient memory changeover switch 141. Coefficient memory 140 is addressed by the count value of counter 142, which receives pulses of appropriate frequency from clock generator 143 to increase the count value. It is assumed that clock generator 143 can adjust its output frequency. Further, in the embodiment shown in FIG. 10, the maximum value of the address of the coefficient memory 140 is designed to match the maximum value of the count of the counter 142,
When all bits of the output of the counter 142 become logic "1", the AND gate 145 is inactivated via the NAND gate 144, and pulse input to the counter 142 is blocked. The counter 142 is cleared by the key-on signal KON indicating that a new key has been pressed on the keyboard section 11. Therefore, the coefficients b ₁ to _{b 6} stored in the coefficient memory 140 are signals.
From KON to next signal KON or signal
From KON, six coefficients b ₁ to b ₆ are read out in parallel and input to the selector 5 until the pulse input is blocked by the AND gate 145 . Since the coefficients b ₁ to b ₆ control the amplitude of the sine wave within the time windows W ₁ to _{W 6} shown in FIG. 4, respectively, regarding the spectrum diagram in FIG. 5, the coefficients b ₁ to b ₆ correspond to Spectral envelope S (W ₁ ) ~
The amplitude of S(W ₆ ) will be controlled. That is, in the embodiment of the present invention, the single multiplication circuit 6 calculates the spectral envelopes S(W ₁ ) to S(W ₆ ).
can be controlled independently of each other. The output of the multiplier circuit 6 is input to the DAC 7, converted to an analog voltage, and produced as a musical tone by the sound system 8. Since the DAC 7 and the sound system 8 are well known,
Detailed explanation will be omitted. As explained earlier in the comparison between Fig. 1 and Fig. 4, the time width of the time window in Fig. 4 is set to be relatively narrow with respect to the period of the sine wave passing through the time window. Therefore, as shown in FIG. 5, the frequency bandwidth that can be independently controlled is relatively wide, and adjacent spectral envelopes overlap. In other words, it can be said that the frequency resolution in spectrum control is relatively low. In order to improve this resolution, it is sufficient to increase the time width of the time window. For example, TW
If a musical sound waveform as shown in FIG. 6 is generated, the frequency resolution in control can be improved more than in the circuit shown in FIG. FIG. 11 is a block diagram showing another embodiment of the present invention, in which the same reference numerals as in FIG. 6 indicate the same or corresponding parts, and redundant explanation will be omitted. In FIG. 11, 20 is a sine wave memory with a capacity of 1024 words;
1 is a sine wave memory with a capacity of 512 words, 22 is a sine wave memory with a capacity of 512 words,
Sine wave memory with a capacity of 256 words, 61, 62,
63 is a multiplication circuit, and 64 is an addition circuit. The upper 10 bits of the output of the accumulator 13 (c ₉
~ c ₀ ) to read out the sine wave memory 20 and continuously output a sine wave of frequency f ₀ , and MSB from the upper 10 bits (c ₉ ~ c ₀ ) of the output of the accumulator 13.
The 9 bits (c ₈ to c ₀ ) excluding (c ₉ ) are used to read out the sine wave memory 21 and continuously output a sine wave with a frequency of 2f ₀ . The 8 bits (c ₇ to c ₀ ) obtained by removing the most significant bit c ₈ from the above 9 bits (c ₈ to _{c 0} ) are input to the address switching device 4 , and the output of the address switching device 4 is used to select the sine wave memory. 22 and outputs the waveform shown in the musical tone waveform TW in FIG. The circuit of the decoder 3 shown in FIG. 11 is similar to the circuit shown in FIG. 7 or 8, but since there are five types of time windows W ₁ to W ₅ in the waveform of FIG. The circuit related to the time window _W6 shown in FIG. 8, that is, the inverter 306 and the AND gate 316 in FIG. 7 are missing, and the ring counter 32 in FIG.
has six stages and lacks the AND gate 326. Therefore, the input to the decoder 3 is the upper five bits ( _c9 to _c5 ) of the output of the accumulator 13, and its output is the time window _W1 to _W5 . Further, the connection of the address switching device 4 is similar to that shown in FIG. 9, but since the number of words of the sine wave memory 22 is 256 words, the input C ₈ bits and the output A ₈ bits are omitted from FIG. 9, and the time window is omitted. This is a connection that omits the connection corresponding to W ₆ . Furthermore, the coefficient generator 14 is similar _{to the} circuit shown in _{FIG .}
Coefficients b ₀ and b ₁ corresponding to the respective outputs of 0 and 21
occurs. Coefficients b ₂ to b ₆ are input to the selector 5, and coefficients b ₀ and b ₁ are input to multiplication circuits 61 and 62, respectively. Therefore, an operation similar to the reading of the sine wave memory 2 of FIG. 6 is performed for reading of the sine wave memory 22 of FIG. 11, except that, unlike the case of FIG. 6, each time window W ₁ to W ₅ in sine wave memory 22
is applied twice and read out repeatedly, and its output waveform is the first one.
The result will be as shown in waveform TW in the figure. Therefore _, the spectral envelopes corresponding to the time windows W ₁ to _{W 5} are shown as S ₄ to _{S 64 in} FIG _. The amplitude is independently controlled by ~ _b6 and output. Figure 3 to 5
As is clear from a comparison with the figures, the frequency bandwidth that can be independently controlled can be made narrower in the embodiment shown in FIG. 11 than in the embodiment shown in FIG. 6. However, in the embodiment shown in FIG. 11, the waveform read from the sine wave memory 22 (musical sound waveform TW in FIG. 1) contains almost no frequency components of f ₀ and 2f ₀ as shown in FIG. 3. , f ₀ , 2f ₀ components are provided, and the amplitude of the output thereof is controlled by coefficients b ₀ and b ₁ in multiplication circuits 61 and 62, respectively.
The outputs of 62 and 63 are combined by an adder circuit 64.
Input to DAC7. In the above embodiment, the three sine wave memories 20, 21, and 22 are individually provided, but it is also possible to provide one sine wave memory and use the memory in a time-sharing manner. Of course. In the embodiments shown in FIGS. 6 and 11, an accumulator 13 is used, and a sine wave memory 2 is used to generate the musical sound waveform TW shown in FIG.
or 22, but the present invention is not limited to such a specific type of circuit,
Musical sound waveform TW using any conventionally known circuit
Needless to say, this may occur. Further, in the embodiments shown in FIGS. 6 and 11, the time windows are both rectangular time windows. Although the rectangular time window can effectively utilize the time width, the spectrum S ₁ k in Fig. 2
As shown in Figure 3, the spread of the spectrum is large, and the so-called side-lobe spectrum is particularly strong, that is, even in the region where the frequency f is considerably large compared to k/T ₀ , the amplitude is still not small enough, and as a result, as shown in Figures 3 and 3. As can be seen in Figure 5, when trying to control the intensity of the spectrum at a desired frequency point, the entire spread of the spectrum is controlled, making it difficult to control each band independently. . In order to eliminate this drawback, a time window with a shape other than a rectangle and a time window with small side lobes in the frequency spectrum may be used. However, in this case, the spread of the main rope in the spectrum is 1
In order to suppress it within an octave, a window longer than a rectangular time window is required, which has the disadvantage that the utilization rate of the time width of the time window decreases. FIG. 12 is a waveform diagram showing an example of musical sound waveforms generated in still another embodiment of the present invention, in which the horizontal axis represents time t, waveforms TW ₁ ,
TW ₂ are sine waves and waveforms with frequencies f ₀ and 2f ₀ , respectively.
TW ₄ shows the waveform when four sine waves with a frequency of 4f ₀ are passed through a time window other than a rectangle with a time width equal to the period T ₀ , such as a Hanning window, and TW ₈ shows the waveform when the time width is T ₀ / 2, T ₀ /4, T ₀ /8, T ₀ /16
Hanning windows with frequencies of 8f ₀ , 16f ₀ , and
The waveform is shown when passing sine waves of 32f ₀ and 64f ₀ . Waveforms TW ₄ and TW ₈ in Figure 12 are the waveforms in Figure 1.
Compared to TW, the time width of the time window is twice as large, and the shape of the time window has changed from a rectangular window to a Hanning window. As is clear from Fourier analysis of the Hanning window waveform shown in Figure 12 waveform TW ₄ , strong side lobes (with frequencies above k/2T ₀₎ as seen in the rectangular window spectrum shown in Figure 2 as S ₁ k (spectrum in the frequency domain) is sufficiently attenuated in the spectrum of the Hanning window, and since the time width of the Hanning window in Figure 12 is twice the time width of the rectangular window in Figure 1, the width of the main lobe is also the same as that of the rectangular window. The width is about the same as that of . In order to improve the frequency resolution in spectrum control, the time width of the time window should be increased, and in order to attenuate the side-lobe spectral distribution that is harmful to improving the frequency resolution in spectrum control, the time window should be changed from a rectangular window to a Hanning window. You can change it to a window. In the embodiments shown in FIGS. 6 and 11 above, the time width of each time window is an integer fraction of the period T ₀ of the musical tone.
The ratio of the period of the sine wave, that is, the vibration waveform passing through the time window, and the time width of the time window is also an integer value, and as a result, the frequency of the vibration waveform whose amplitude is controlled by the time window waveform is equal to that of the musical tone. It is made to be an integral multiple of the pitch frequency. However, the invention is not limited to this, and the width of each time window and the period of a vibration waveform passing through that time window, that is, the period of a sine wave, for example, can be calculated as the period of a musical tone.
It is also possible to set T ₀ to 1 by a non-integer, so that the frequency of the vibration waveform becomes a non-integer multiple of the pitch frequency of the musical tone. For example, set the width of each time window to 2Tw/k (however, Tw
is a numerical value that satisfies the condition of Tw<T ₀ , and therefore changes according to the pitch of the musical tone, that is, T ₀ ), and the frequency of the sine wave, that is, the vibration waveform, passing through the time window is k/Tw, and kw =4,8,16,32,
64, _{T 0 ≧} (2Tw/4+
It is clear that if the relationship 2Tw/8+2Tw/16+2Tw/32+2Tw/64) is maintained, a waveform similar to the waveform shown in waveform TW in Figure ₁ can be generated. However, in this case, the above Tw is The relationship is a non-integer ratio. In this case as well, since the above waveform is repeated exactly at the tone period T ₀ , the line spectrum obtained by analyzing this waveform becomes a line spectrum with a frequency that is an integral multiple of f ₀ =1/T ₀ . However, k/
Since Tw is a non-integer multiple of the frequency 1/T ₀ of the musical tone, the center frequency k/Tw of the resulting moving formant is shifted from the position of the overtone frequency of the musical tone (see FIG. 13). This allows the timbre to be controlled. FIG. 13 is a spectrum diagram showing an example of a spectrum in still another embodiment of the present invention.
For To, sine wave for 2Tw/4 every T ₀ time
The relationship between the spectrum envelope and the line spectrum when Asin4(2π/Tw)t is passed is shown.
In Fig. 13, Sw is a spectral envelope whose apex is the point 4/Tw, as is clear from the explanation regarding Fig. 2, whereas the actually generated frequency is an integer multiple of f ₀ = 1/T ₀ . There are 3f ₀ , 4f ₀ ,
This is a line spectrum of frequencies 5f ₀ , 6f ₀ .... In FIG. 13, an example of the spectral envelope corresponding to spectral envelope S ₄ among the spectral envelopes shown in FIG. 3 has been explained, but other spectral envelopes shown in FIG.
Similarly, for each spectrum envelope corresponding to S ₈ to _{S 64} , the shape of each spectrum envelope is determined by the above Tw, whereas the actually generated frequency is an integer multiple of f ₀ = 1/T ₀ . it is obvious. FIG. 14 is a block diagram showing yet another embodiment of the present invention, and FIG. 15 is a waveform diagram showing an example of the waveform read out from the sine wave memory 22 of FIG. 14. In the circuit shown in FIG. 14, the waveform read from the sine wave memory 22 is kept constant for musical tone frequencies within the same octave, and the waveform read from the sine wave memory 22 when the octave of the musical tone frequency changes. As shown in FIG. 15, switching is performed. Further, the repetition frequency of the waveform is controlled to always match the musical tone frequency, and the starting point of the waveform is controlled to be synchronized with a predetermined phase point of the musical tone frequency. The circuit in Figure 14 has four types of octaves, namely, the first octave (OC1), the second octave (OC2), the third octave (OC3), and the fourth octave (OC4), and each octave has a sine wave memory. The waveform read from 22 is the 15th waveform.
The figure shows TW (OC1), TW (OC2), TW (OC3), TW
(OC4). This waveform TW
The periods of musical tones corresponding to (OC1) to TW (OC4) are shown as T ₀₁ to _{T 04} in Fig. 15, and the waveform WT is
Although the time range from (OC1) to TW (OC4) is not necessary,
The lengths of T ₀₁ to _{T 04} vary. As can be seen from the waveform in FIG. 15, the highest frequency of the sine wave read from the sine wave memory 22 in FIG. 14 is always constant even if the octave changes. Therefore, even if the waveform changes from TW (OC1) to TW (OC4), the upper limit frequency of the spectrum corresponding to that waveform does not change. In other words, the spectral envelope of waveform TW (OC1) corresponds to the spectral envelopes S ₈ to S ₆₄ in Fig. 3, and waveform TW (OC2),
The spectral envelopes of TW(OC3) and TW(OC4) correspond to the spectral envelopes S ₈ to _{S 64} , S ₁₆ to _{S 64} , and S ₃₂ to _{S 64} in FIG. 3, respectively. Third
Since the frequency range higher than the spectral envelope _S64 in the figure falls into the inaudible frequency range and has little effect on the timbre, the generated waveform is switched as shown in Figure 15 according to the type of octave. The waveform generation circuit is simplified. In FIG. 14, the same symbols as in FIG. 11 indicate the same or equivalent parts, and 15 is an encoder that encodes information corresponding to the pitch of the key pressed on the keyboard section 11 into an octave code OCC and a pitch name code NTC. Convert and output. The octave code OCC is 2-bit data representing the four types of octaves OC1 to OC4 mentioned above, and the note name code NTC is 4 bits representing the 12 note names.
This is bit data. 121 and 122 are first and second frequency information memories, respectively, which correspond to the frequency information memory 12 in FIG. memory 1
22 is the octave code OCC and note name code NTC
is input and outputs the numerical value F ₀ as in the frequency information memory 12 of FIG. 16 and 17 are multiplication circuits, and the operations of these multiplication circuits will be explained later, but if there is no multiplication input shown as WOW and VIB in FIG.
Output F ₀ as is. 131 is a first accumulator, and 132 is a second accumulator, each of which corresponds to the accumulator 13 in FIG _{. c0} )
Reading the sine wave memory 22 via the address switching device 43 by reading the sine wave memory 22 via the address switching device 4 is performed by the output of the upper 8 bits (c ₇ to c ₀ ) of the accumulator 13 shown in FIG. This is the same operation as reading 22. 18 is a flip-flop. The accumulator 132, like the accumulator 13 in FIG. Therefore, the first address from which the sine wave memory 22 is read is synchronized with the overflow pulse CA,
As shown in Figure 15, the waveform TW (OC1) ~ TW
The starting point of (OC4) is synchronized with the starting points of musical tone cycles T ₀₁ to _{T 04} , respectively. 30 is a decoder, which corresponds to decoder 3 in FIG. 11, and 43 is an address switching device, which corresponds to address switching device 4 in FIG. 11. Furthermore, in the circuit of FIG. 14, like the circuit of FIG. 11, a coefficient generator 14, sine wave memories 20, 21,
It is equipped with multiplication circuits 61, 62, addition circuit 64, DAC7, and sound system 8, but these are
Since it is the same as in FIG. 1, it is not shown in FIG.
In FIG. 14, sine wave memories 20, 21 (not shown in the drawing) are addressed by the output of accumulator 132. In FIG. FIG. 16 is a circuit diagram showing an example of the decoder 30 in FIG. 14, in which the same symbols as in FIG. 7 indicate the same or corresponding parts, 34, 35, and 36 are AND gates, and 37 and 38 are OR gates, respectively. ,
According to FIG. 7, the upper five bits ( _c9 to _c5 ) of the accumulator 131 are input and the AND gates 311 to
315 outputs time windows W ₁ to W ₅ , but the 2-bit octave type of the octave code OCC
It is decoded into a signal representing OC1 to OC3 and is decoded according to the octave OC1, OC2, OC3, OC4 by OR gates 37, 38 and AND gates 34, 35, 36.
The time windows of W ₁ to _{W 5} , W ₁ to W ₄ , W ₁ to W ₃ , and W ₁ and W ₂ are output, respectively. As the octave increases, the output frequency of the accumulator 131 increases,
Therefore, the time width of the time window becomes narrower; for example, the width of time window W ₁ in octave OC4 becomes equal to the width of time window W ₄ in octave OC1. The connection of the address switching device 43 is the same as the connection of the address switching device 4 in FIG. 11, but depending on the type of octave, the time window may be changed as shown in FIG.
Since only a portion of W ₁ to _{W 5} is output, the address switching device 43 switches only the type of time window input as a control input. As described above, the sine wave memory 22 outputs TW (OC1) in FIG. 15 according to the octave code OCC.
The waveform shown in ~TW (OC4) is output. The selector 5 is similar to the selector 5 in FIG. Only the coefficients corresponding to are output, and the multiplier circuit 63 controls the amplitudes of the sine waves corresponding to the waveforms TW(OC1) to TW(OC4) shown in FIG. Therefore, the spectrum envelope of the waveform output from the sine wave memory 22 in FIG _{. 14} is as shown in FIG. ₃ in the _{octave OC1} , and On the other hand, the fundamental frequency of a musical tone changes within a range of one octave with the upper limit f ₀ in Figure 3, and in the octave OC2, the spectral envelope S ₈ , S ₁₆ , S ₃₂ ,
S ₆₄ , whereas the fundamental frequency of musical tones changes within the range of 2f ₀ - f ₀ in Figure 3, and the octave OC3
In Fig. 3, the spectral envelopes S ₁₆ , S ₃₂ ,
S ₆₄ , whereas the fundamental frequency of musical tones varies within the range of 4f ₀ - 2f ₀ in Figure 3, and varies by octave.
In OC4, Fig. 3 Spectral envelope S ₃₂ , S ₆₄
In contrast, the fundamental frequency of musical tones is 8f ₀
-4f Varies within the range of ₀ . In the multiplier circuit 63, the amplitude of each sine wave is independently time-divisionally controlled by the coefficients b ₂ to b _6. As a result, in the octave OC1, the intensities of the portions corresponding to the spectral envelopes S ₄ to _{S 64} in FIG. 3 are controlled by the coefficients. b ₂ to b ₆ , and in octave OC2, the strength of the part corresponding to the spectral envelope S ₈ to _{S 64} is controlled by coefficients b ₃ to b ₆ ,
Spectral envelope S ₁₆ in Octave OC3
The strength of the part corresponding to ~S ₆₄ is controlled by coefficients b ₄ to b ₆ , and in octave OC4, the strength of the part corresponding to spectral envelopes S ₃₂ and S ₆₄ is controlled by coefficient b ₅ .
Controlled by b ₆ . Also, the spectral envelope shown in Figure 13
As can be easily understood from the relationship between Sw and each line spectrum, a vibrato effect accompanied by spectral fluctuation can be obtained by giving a minute fluctuation to the frequency relationship between fw = 1/Tw and fo = 1/T ₀ , or A wah-wah effect accompanied by spectral fluctuations can be obtained. The multiplication circuit 16 in FIG. 14 has a coefficient (WOW
(coefficients shown as Since the position of each line spectrum on the horizontal axis does not change, the intensity of each line spectrum changes as the spectrum envelope Sw changes, producing a wah-wah effect. In addition, the multiplier circuit 17 multiplies the numerical value Fo by a coefficient (coefficient shown as VIB in the drawing) that slightly fluctuates around the numerical value 1. At this time, if the output numerical value of the multiplier circuit 16 is constant, the spectral envelope shown in FIG. While Sw is constant, the position on the horizontal axis of each line spectrum moves to produce a vibrato effect. By the way, in the above explanation, the first frequency information memory 121 stores numerical values corresponding to each octave.
I explained that it was memorizing Fw, but memory 1
21 is the value for the lowest octave (OC1)
Alternatively, only Fw may be stored, and the numerical value Fw for the other upper octaves (OC2 to OC4) may be shunted one bit at a time to obtain the numerical value Fw for the upper octave. In this case, if the numerical value Fw to be stored in the memory 121 is Fw=1.0, then the memory 121
The configuration is very simple. Furthermore, the memory 121
Including the multiplication circuit 16, (1+Δ)·2 ^(oct-1) , where Δ is a coefficient for WOW, and oct is a numerical value representing each octave. In all of the embodiments described above, cases have been described in which a sine wave is passed through a time window, but it goes without saying that the present invention is not limited to sine waves and any vibration waveform may be passed through a time window. do not have. Further, in the embodiments described above, the frequency of the generated sine wave is changed stepwise, but the same effect can be obtained even if the frequency is changed continuously. FIG. 17 is a block diagram showing still another embodiment of the present invention, in which the same reference numerals as in FIG. 6 indicate the same or corresponding parts and operate in the same way, so a description thereof will be omitted. 28 is a waveform generator, and the accumulator 1
A sine wave whose frequency changes continuously is generated starting from the overflow pulse CA of No. 3, and this sine wave is terminated before the next overflow pulse CA arrives. 19 is a coefficient (coefficient b ₁ in the circuit of FIG. 6) whose value changes in accordance with the frequency change of the output sine wave of the waveform generator 28, and whose value changes as a function of time starting from the key-on signal KON.
This is a coefficient generator that outputs a coefficient corresponding to ~ _b6 ). Circuits that generate sine waves whose frequency changes continuously are conventionally well known as frequency modulation circuits in wireless communication devices, frequency sweep circuits in various measuring instruments, etc.
Any conventionally known circuit may be used. FIG. 18 is a block diagram showing an example of the waveform generator 28 in FIG. 17, in which 2 is a sine wave memory similar to the sine wave memory 2 in FIG. 6, 123 is a frequency information memory, 133 is an accumulator, which corresponds to the first accumulator 131 in FIG.
281 and 282 are AND gates, 283 is an OR gate, 284 is an inverter, 285 is a shift register, 286 is a multiplication circuit, and 287 is a flip-flop. A numerical value R corresponding to the pitch of a key pressed on the keyboard section 11 is read out from the frequency information memory 123 and accumulated in the accumulator 133 via an AND gate 281 and an OR gate 283. However, the numerical value R
is input to the accumulator 133 at the 17th
Overflow pulse from accumulator 13 in the figure
Each time CA is output, at other times the output of the OR gate 283 is sent to the multiplier circuit 286, the shift register 285, and the AND gate 2.
82 and the value multiplied by the constant k is accumulated in the accumulator 133 at every clock point. Therefore, the numerical value R' added to the accumulator 133 changes exponentially, and as a result, the output frequency of the sine wave memory 2, which is read out using the upper 9 bits of the accumulator 133 as an address, changes exponentially. When the carry pulse is output from the multiplier circuit 286, the flip-flop 287 is set and the overflow pulse CA from the accumulator 13 is output.
This resets the flip-flop 287, synchronizes the starting point of the waveform read from the sine wave memory 2 with a predetermined phase point of the musical tone frequency, and causes the waveform to end within the musical tone period. FIG. 19 is a block diagram showing an example of the coefficient generating device 19 of FIG. 17. The same reference numerals as in FIG. 10 indicate the same parts and perform the same operations, so redundant explanation will be omitted. 146 is a coefficient memory corresponding to the coefficient memory 140 in FIG. Further, 191 is an envelope information memory, 192 is an accumulator, 1
93 is an AND gate, and 194 is a flip-flop. From the coefficient generator ₁₄ shown in FIG.
- b ₆ are read out in parallel, but the coefficients corresponding to b ₁ - _{b 6} are read out from the coefficient generator 19 shown in FIG. 23 in synchronization with the frequency change of the sine wave read out from the waveform generator 28. It is read out serially in time so that it changes. Therefore, the coefficient memory 146 is
Like the coefficient memory 140 shown in the figure, it is addressed by the output of the counter 142 and at the same time by the output of the accumulator 192, which changes in synchronization with changes in the output frequency of the waveform generator 28. The AND gate 193 sets the flip-flop 194 when the output logic of the accumulator 192 becomes "1" for all bits, and since the flip-flop 194 is reset by the overflow pulse CA of the accumulator 13, the tone waveform generator 28 The coefficient memory 146 outputs a coefficient that changes in synchronization with changes in the output frequency of the key-on signal KON and changes as a function of time starting from the key-on signal KON. The envelope information memory 191 outputs a digital numerical value corresponding to the pitch of the key pressed on the keyboard section 11 and inputs it to the accumulator 192, and this numerical value and the frequency information memory 12 of FIG.
3 maintains a predetermined relationship between the change in the output frequency of the waveform generator 28 and the change in the coefficient output from the coefficient generator 19. The frequency spectrum of a sine wave whose frequency changes continuously has been well analyzed as the spectrum of a frequency modulated wave, so its explanation will be omitted. It is clear that the generation of uncontrolled harmonics at the discontinuity points of each sine wave can be prevented and advantageous control can be achieved. Various embodiments of the present invention have been described above with reference to the specific circuits shown in the drawings, but the invention is not limited by these embodiments or by the specific circuits shown in the drawings. Needless to say, this is not something that can be done. Finally, the correspondence between each component of the present invention and the above-described embodiments is summarized as follows. Pitch specification means: Keyboard section 11 Address signal generation means: Frequency information memory 1
2, 121, 122, 123 and accumulator 1
3, 131, 132, 133 parts Address signal conversion means: address switching device 4,
43, as well as the multiplication circuit 28 in the waveform generator 28
6. Shift register 285, AND gate 28
2. Loop of OR gate 283 Vibration waveform generation means: Sine wave memory 2, 21, 2
2 Control waveform generation means: Decoders 3, 30, selector 5, coefficient generation device 14, and coefficient generation device 19 Amplitude control means: Multipliers 6, 62, 63 In an embodiment, the control waveform generation means includes: A control waveform is generated in accordance with the address signal generated by the address signal generating means. In the example, this means that the accumulator 1
3 is decoded by the decoder 3, the selector 5 selects the output of the coefficient generator 14 with a rectangular waveform corresponding to the decoded output, and as a result, a control waveform consisting of a rectangular waveform of the selected coefficient value is generated. It corresponds to that. In another embodiment, the address signal converting means temporally changes the N within one period of a musical tone. The N times in this address signal converting means may be an integer multiple or a non-integer multiple. In another embodiment, the address signal generating means modulates information representing the pitch specified by the pitch specifying means and generates an address signal that repeatedly changes according to the modulated information. It is. This means that in the embodiment of FIG.
The outputs of the frequency information memories 121 and 122 are used as coefficients.
It supports modulation with WOW and VIB. In another embodiment, the control waveform generating means generates a time window waveform having a predetermined time width within one cycle of the pitch frequency as the control waveform. This corresponds to the fact that in the embodiment, the control waveform output from the selector 5 is a time window waveform having a predetermined time width according to the output of the decoder 3. [Effects of the Invention] As is clear from the above description, the present invention provides an excellent effect in that a musical tone signal having moving formant characteristics can be generated with a simple configuration.

[Brief explanation of drawings]

FIG. 1 is a waveform diagram showing an example of a musical sound waveform generated in an embodiment of the present invention, FIG. 2 is a spectrum diagram explaining the spectral structure of the waveform shown in FIG. 1, and FIG. 4 is a waveform diagram showing an example of the musical sound waveform generated in another embodiment of the present invention. FIG. 5 is a spectrum diagram showing the spectral envelope of the musical sound waveform shown in FIG. 4. 6 is a block diagram showing an embodiment of the present invention, FIG. 7 is a circuit diagram showing an example of the decoder of FIG. 6, and FIG. 8 is another example of the decoder of FIG. 6. 9 is a connection diagram showing the internal connections of the address switching device of FIG. 6, FIG. 10 is a block diagram showing an example of the coefficient generating device of FIG. 6, and FIG. 11 is a circuit diagram showing an example of the coefficient generating device of FIG. 12 is a waveform diagram showing an example of a musical sound waveform generated in another embodiment of the present invention. FIG. 13 is a block diagram showing yet another embodiment of the present invention. FIG. 14 is a waveform diagram showing an example of a musical sound waveform generated in still another embodiment of the present invention, FIG. 15 is a spectrum diagram showing the spectral envelope of the waveform shown in FIG. 14, and FIG. FIG. 13 is a spectrum diagram showing an example of a spectrum in still another embodiment of the present invention; FIG. 14 is a block diagram showing still another embodiment of the present invention; FIG. 15 is a waveform diagram showing an example of the waveform generated in the circuit of FIG. 14, and FIG.
FIG. 4 is a circuit diagram showing an example of the decoder, FIG. 17 is a block diagram showing still another embodiment of the present invention, FIG. 18 is a block diagram showing an example of the waveform generator shown in FIG. 17, and FIG. 18 is a block diagram showing an example of the coefficient generator of FIG. 17; FIG. 11...Keyboard section, 12, 121, 122, 123
...Frequency information memory, 13, 131, 132, 1
33...Accumulator, 14,19...Coefficient generator, 15...Encoder, 2,20,21,22...
Sine wave memory, 3, 30...decoder, 4...address switching device, 5...selector, 6, 61, 63...multiplication circuit, 64...addition circuit, 7...DAC, 8...sound system.

Claims (1)

[Scope of Claims] 1. Pitch specifying means for specifying the pitch of a musical tone to be generated; and an address signal that repeatedly changes at a rate corresponding to the pitch frequency of the pitch specified by the pitch specifying means. and an address signal generating means for generating the address signal, the address signal generated from the address signal generating means being N times the pitch frequency of the pitch specified by the pitch specifying means (however, N is greater than 1). an address signal conversion means for converting the address signal into an address signal that repeatedly changes at a rate corresponding to the frequency of the address signal, and a waveform storage means for storing waveform data of a predetermined waveform, the address signal converted by the address signal conversion means Accordingly, by repeatedly reading out the stored waveform data from the waveform storage means,
vibration waveform generation means for generating a vibration waveform having a frequency N times the pitch frequency; control waveform generation means for repeatedly generating a control waveform at a period corresponding to the pitch designated by the pitch designation means; amplitude control means for controlling the amplitude of the vibration waveform generated by the vibration waveform generation means by the control waveform, and the amplitude control means follows changes in pitch specified by the pitch specification means. A musical tone signal forming device characterized in that a musical tone signal having moving formant characteristics in which the formant characteristics change is obtained. 2. The musical tone signal forming apparatus according to claim 1, wherein said control waveform generating means generates a control waveform in accordance with an address signal generated by said address signal generating means. 3. The musical tone signal forming apparatus according to claim 1, wherein the address signal converting means temporally changes the N within one period of a musical tone. 4. The address signal generating means modulates information representing the pitch specified by the pitch specifying means and generates an address signal that repeatedly changes according to the modulated information. 1
The musical tone signal forming device as described in . 5. The musical tone signal forming apparatus according to claim 1, wherein the control waveform generating means generates, as the control waveform, a time window waveform having a predetermined time width within one cycle of the pitch frequency.