JPH0833748B2 - Sound synthesis method - Google Patents

Description

The present invention relates to a method for synthesizing human voice or musical instrument sound.

B. SUMMARY OF THE INVENTION The present invention is a method of synthesizing human voice or musical instrument sound, in which a radius of an acoustic tube is calculated, and a phenomenon of acoustic propagation in the calculation result is a system transient phenomenon in an electric circuit such as a transmission line. Focusing on the approximation, calculate the radius of the acoustic tube,
By varying the surge impedance in response to changes in the surge impedance of the calculated output, it is possible to obtain natural sound quality synthesis.

In addition, for the synthesis of sound by changing the length of the acoustic tube of a wind instrument, etc., a delay circuit is provided in the impedance coupling section and the delay constant corresponding to the change in length can be changed to synthesize the sound. Is.

C. Conventional technology Speech synthesis and music synthesizer (electronic musical instrument)
Recently, electronic devices that artificially synthesize and output so-called sounds such as LS have been used for LS for voice recognition and voice synthesis of one or several chips.
I has come to be realized at a low price by means of voice information processing and semiconductor large-scale integrated circuit technology, and various methods have been proposed depending on the purpose of use and constraints. In this speech synthesis, a raw voice generated by a human being is recorded, and the recording and editing method in which the raw voice is appropriately combined and edited into a sentence, and the human voice is not directly used. There is a method of extracting only the parameter of, and controlling the parameter in the voice synthesis process to artificially generate a voice signal.

There is a percall (PARCOR) method which is widely used because a high quality synthesized sound can be obtained by this parameter method.

When a voice is handled by an electronic computer, the voice waveform is sampled at a certain cycle, the value of the voice signal at each sampling point is converted from analog to digital, and the value is displayed with a code of 0 and 1. It is necessary to increase the number of bits in order to record faithfully to the analog signal, but the voice synthesis signal requires a very large amount of memory.

Therefore, various highly efficient coding methods have been researched and developed in order to reduce the amount of this information as much as possible.

As one of the methods, there is a delta modulation method in which the information of one audio signal is at least 1 bit. This method uses one bit to judge whether the next audio signal value is higher or lower than the current value, and if the value is higher, the code "1" is given, and if the value is lower, the code "0" is given. In the actual system configuration, a fixed amplitude step amount (delta) is set, and the audio value obtained by the encoding up to now and the input audio signal are set so that errors are not accumulated. The residual signal of is encoded.

Such a configuration is called predictive coding, and uses a linear prediction method (predicting from several previous sample values) and a Percoll method (a partial autocorrelation function called a Percoll coefficient k instead of the prediction coefficient of the linear prediction method). ).

D. Problems to be Solved by the Invention As described above, the method using predictive coding has a problem that articulatory coupling corresponding to a joint between sounds is difficult. That is, in FIG. 10, the horizontal axis represents the sound generation time t, and the vertical axis represents the Percoll coefficient k. Also, in the process of passing a vowel transition to a stationary vowel sound, the seam of the vowel and the vowel sound is cut off, which does not give a natural feeling to humans.

Also, in the case of musical instrument sound synthesis, the seam of the scale is important, but the synthesis method is different from the principle of sound generation of the actual instrument,
After all, there is no natural feeling, and it appears remarkably especially in reverberation sound. In order to bring the sound closer to a natural sound in both cases, there is a problem that a large amount of memory and an electronic component such as an arithmetic unit are required and the apparatus becomes expensive.

E. Means for Solving the Problems In view of the above points, in the present invention, the generation of human sound or the musical sound of musical instrument is created by the shape change of the human oral cavity or the acoustic tube, such as the length and cross-sectional area. Therefore, the traveling wave phenomenon that represents the transmission of sound waves in these acoustic tubes is analyzed by an acoustic tube equivalent circuit, the radius of the acoustic tube is calculated, and then the cross-sectional area is calculated by multiplying the pi and the square of the radius. Focusing on the fact that the cross-sectional area is inversely proportional to the surge impedance, the cross-sectional area calculated by calculating the radius of the acoustic tube is changed by changing the surge impedance, and the surge impedance is continuously changed to articulate coupling. This facilitates smooth synthesis of sounds similar to human speech and improves the naturalness of speech.

Also, the change in the length of the musical instrument is simulated by the number of stages of the delay circuit,
Together with the change in cross-sectional area, we made it easier to achieve more natural instrument sounds.

F. Action Humans produce sounds by moving their mouths, and wind instruments make musical sounds by changing the length and shape of acoustic tubes. The present invention calculates a radius r of an acoustic tube (a vocal tract from a human vocal cord to a lip can also be regarded as one acoustic tube), then obtains a cross-sectional area from πr ² and calculates the cross-sectional area of an equivalent circuit. Since the surge impedance is made to correspond one-to-one, changing this surge impedance is the same as changing the cross-sectional area of the acoustic tube. Since this surge impedance can be changed very easily in terms of electrotechniques, it is possible to synthesize sounds that are exactly the same as the generation of human sounds. In particular, the articulation coupling that is a problem between sounds and sounds, which is a problem in the past, is also surged. Good performance is achieved by continuously changing the impedance, and it is possible to produce a sound that is close to natural.

Further, changing the length of the acoustic tube is to delay the traveling wave of the sound wave, and therefore, in terms of an electric circuit, it corresponds to changing the number of stages of the delay circuit (memory), and is extremely adjusted by adjusting the constant of the delay circuit. It can be easily simulated. Therefore, a more natural musical instrument sound can be easily realized in combination with the cross-sectional area change obtained by calculating the radius.

G. Example First, before describing the example of the present invention, the background of the present invention will be described.

A sound source is required to radiate sound out of the mouth, which is produced by the vocal cords. On the other hand, the vocal cords have the function of intermittently stopping the exhalation by opening and closing two folds, and the intermittent flow creates an air flow called a puff. Becomes higher and a high frequency puff sound is generated. And when the expiratory flow is increased, a loud sound is produced.

When this sound source wave passes through a cylindrical acoustic tube such as the vocal tract, a certain component of the sound wave is emphasized by the resonance phenomenon from the open end, and a certain component is attenuated to form a complicated vowel waveform. Even if the sound source has the same waveform, it is affected by the shape of the vocal tract that passes through before it is emitted from the lips. That is, if the shape of the vocal tract is constant, the spectrum envelope does not change much even if the pitch or intensity of the sound source is changed. The vocal tract shows an extremely complicated shape due to vowels, but it can be divided into a part where the vocal tract does not change much and a part where the vocal tract changes greatly. For example, as shown in FIG. ₁ , it can be assumed that two acoustic tubes having different lengths and sectional areas A ₁ and A ₂ are connected.

FIG. 1 is an acoustic tube model diagram, and FIG. 2 is an equivalent circuit diagram thereof, in which two acoustic tubes having different sectional areas A ₁ and A ₂ are connected.

Focusing on the surface to which the acoustic tube is connected, when the sound waves have different cross-sectional areas, a phenomenon occurs in which a part of the sound wave is reflected on the different surface. This phenomenon is the same as a transient phenomenon when an impulse current is passed through lines having different impedances in an electric circuit.

As described above, the sound is generated by the intermittent sound source by the vocal cords, which is electrically equivalent to the intermittent application of impulses.

Sound is a kind of vibration that can be transmitted by gas, liquid, or solid, but in terms of electrical circuits, it can be applied to a lossless LC distribution circuit with no resistance. And the electrical impedance (V / I) of this equivalent circuit is Therefore, when the sound wave is replaced, the sound wave velocity, the air density ρ, and the sound velocity c are multiplied by ρc, and the impedance in the sound field, that is, the acoustic impedance depends only on the mass and the sound velocity of the gas.

When acoustic tubes with different cross-sectional areas are connected in series, reflection occurs at the boundary surface. This can be simulated as an electrical surge impedance. That is, an equivalent circuit in which blocks having different cross-sectional areas of the acoustic tube as shown in FIG. 1 are connected is replaced with FIG.

Here, assuming that the air density is ρ and the sound velocity is c, the acoustic admittances Y ₁ and Y ₂ of the respective acoustic tubes are given as follows.

However, Z ₁ and Z ₂ are acoustic impedances.

Next, let the propagation current sources from the adjacent blocks be I ₁ and I _2, and let the current distributions a ₁ and a ₂ and i ₁ and i ₂ and the voltage at the junction point be e, which are determined by these. Also, a ₁ = i ₁ + I ₁ , a ₂ = i ₂ + I ₂ i ₁ = a ₁ -I ₁ , i ₂ = a ₂ -I, and the propagation current source I ₁ ′, of the adjacent block for the next step is obtained. I ₂ ′ becomes I ₁ ′ = i ₁ + a ₁ and I ₂ ′ = i ₂ + a ₂ .

It is also possible to substitute i ₁ = a ₁ -I ₁ and i ₂ = a ₂ -I ₂ in the above formula to obtain I ₁ ′ = 2a ₁ −I ₁ and I ₂ ′ = 2a ₂ −I ₁ .

FIG. 5 shows the interpolation situation with respect to time of the sectional area A in the above method. This FIG. 5 shows the linear interpolation which is the simplest to calculate.

Fig. 3 is an electrical circuit equivalent model diagram of the acoustic tube. Fig. 3 (a) is an acoustic tube model diagram in which the vocal tract from the vocal cords to the lips is regarded as one acoustic tube, and Fig. 3 (a) is its electrical circuit model diagram. , (C) shows a traveling wave equivalent model diagram.

Before explaining FIG. 3, it will be explained how human vowels are created.

Fig. 4 simulates the cross-sectional area change of the vocal tract when a voice is generated. Fig. 4 (a) shows the case of "A" when it is pushed out of the lungs with a narrow throat and open lips. The vocal cords open and close during breathing, and the sound waves that are repeatedly reflected in the vocal tract (acoustic tube) emerge as a voice waveform of "A". As shown in (a), “i” has a wide throat and a narrow lip, and the audio waveform of “i” is output.

In this way, the frequency is determined by the mouth preference, and if the mouth preference is simulated, "a" or "a" is uttered. The appearance of the mouth can be simulated by the cross-sectional area of the acoustic tube, and the change of the cross-sectional area of the acoustic tube can be simulated by the change of the surge admittance, and the change of the surge admittance can be changed very easily on the electric circuit. Figure 3 (A) is obtained by simulating the vocal tract by connecting the sound tube having different cross-sectional area as the cross-sectional area _{A 1, A 2 ... A n} . The same figure (a) shows the acoustic impedance of the electric circuit.
It is replaced with an LC circuit, and each acoustic tube is one LC line, and the whole is an n-1 electric circuit of a concentrated line. Further, FIG. 3C is a traveling wave equivalent model diagram, in which acoustic impedances Z ₁ , Z ₂ ... Z _n of each acoustic tube are inversely proportional to the cross-sectional area of the acoustic tube (acoustic admittance is proportional) and Because they are proportional Becomes In the figure, Z _O represents the sound source impedance, Z _L represents the radiation impedance, and the arrows between the blocks represent the traveling wave and the reflected wave.

When uttering the voice "A", "A" is placed at the cross-sectional area A ₁ corresponding to the cross-sectional area of the tip of the lip in Fig. 4.
When the sound of "A" is obtained by applying the impulse P intermittently by giving the shape of the mouth of "A" and the sound of "A" is produced from "A", As shown in the figure, the cross-sectional area is narrowed to A ₁ ′ to give the mouth of “a” the appearance of “a”.
Is obtained.

If impulses P are given continuously and intermittently and the entire cross-sectional area is changed to the shape of the mouth of "a", the vocal tract is simulated by n acoustic tubes shown in FIG. Move each cross section from "A" to change the mouth appearance to "A"
Will be changed continuously. Changing the cross-sectional area of this acoustic tube is performed by gradually changing the surge impedance.

An example of linear interpolation in which the cross-sectional area is continuously changed from A ₁ to A ₁ 'as described above is shown in FIG. Even in the case of the linear interpolation in FIG. 5, the composition is considerably better, but the movement of the human oral cavity does not correspond to the linear movement of the cross-sectional area. Since they correspond to the relationship, the synthesized sound may be somewhat unnatural in the linear interpolation shown in FIG.

Therefore, in the present invention, the radius r of the acoustic tube is calculated as an interpolation target, and the cross-sectional area A is calculated by A = πr ² from the calculation result, and the calculation result is expressed by the above formulas (1) and (2). By substituting, the area ratio of the equations (1) and (2), that is, the coefficient becomes the following equation.

Note that r ₁ and r ₂ are the radii of the adjacent acoustic tubes. As described above, since the radius r of the acoustic tube is calculated and the cross-sectional area is obtained from the result, it is of course possible to obtain the steady-state sounds of "a" and "a", but in between Sound, that is,
A sound between sounds can be obtained with a quality and quality close to natural. FIG. 6 is an explanatory diagram of the result of calculation with the radius r of the acoustic tube according to the present invention as an interpolation target, and the curve indicated by the alternate long and short dash line in the figure is a change characteristic curve of the cross-sectional area when the radius r is calculated. From this curve, a sound between sounds can be obtained with a sound quality close to natural.

The above is for two adjacent acoustic tubes,
The calculation of adjacent multi-branch pipe parts is performed using the following general formula.

(However, r _k is the radius of each acoustic tube of the k = 1, 2 ... n, n branch tube part, and r _i is the radius of the i-th branch tube.) Next, considering the propagation velocity of the sound wave, this is It is similar to the transient phenomenon when an impulse is applied to an electric line with a length of 1 and LC.

That is, a line having LC as shown in FIG. 7 is equivalently expressed as shown in FIG. Here, the surge impedances Z ₀₁ and Z ₀₂ seen from both ends are Becomes

Considering the traveling wave that arrives from the other party as an equivalent current source, If there are n delay circuit blocks Z in the middle, the next current is output after n hours. That is, what has occurred in the circuit on the left has arrived on the right after τ.

I ₂ is the current on the feed tube side Becomes However, in the digital calculation, since the voltage or current is subdivided, V ₁ and V ₂ indicate the voltage at the measurement time t, and τ indicates the elapsed time.

In Fig. 8, if impulses are applied to the L and C circuits, τ
It goes out to the output tube side after time. And, it is equivalently expressed that what has arrived τ time ago has also reached the other party. Setting the line length 1 to 1 facilitates calculation by normalizing the delay block n to 1. l
In case of dividing into 3 cm, the delay block n may be set to 3 blocks.

Figure 3 (a) shows that the human vocal tract is about 17 cm for men, so 1c
If you simulate with 17 acoustic tubes in m steps, the waveform input from A ₁ will come out from the An side in 170 μsec if you divide the half-cycle current into 10 and Δt is 10 μsec. Considering the musical instrument thrombone, the thrombone changes the musical sound depending on the length of the acoustic tube. According to the present invention, it suffices to have two parameters from the sound of "A" of thrombon to the sound of "D" of thrombon. The sound of "A" of thrombone is the sound of "Do" of thrombone There are two parameters. The intermediate tone can be freely articulated by changing the delay block of the delay circuit.

That is, by varying the delay circuits Z ₁ → Z _n and Z _n → Z ₁ in FIG. 8, it is possible to deal with changing the length of the acoustic tube having the same area.

Next, the calculation processing of the traveling wave equivalent model of FIG. 3C will be described based on the flowchart of FIG.

When an impulse is input to the acoustic tube A ₁ (radius r _{1 of the} tube), the arithmetic processing unit including the computer executes step S
_{At 1} , a _OA , i _OA , I _OA , E of A ₁ is taken out from the memory. Based on the retrieved values, in step _{S 2, a OA '= f} (E, I aA) i OA' calculation of = a _OA '-I _OA performed. This calculated value a _OA ′, i _OA ′ and the value a _{1B of} the pipe A ₂ (radius r ₂ ) introduced from the memory in step S ₃
In step S ₄ , using a _1A , i _1B , i _1A , I _1B , I _1A , a _1B ′ = S _1B (I _1B + I _1A ) a _1A ′ = S _1A (I _1B + I _1A ). i _1B ′ = a _1B ′ −I _1B i _1A ′ = a _iA ′ −I _1B I _1B ′ = i _0A ′ −a _0A ′ is calculated. In step S ₅ , I _0A ′ = i _1B + a _1B is calculated using i _1B ′, a _1B ′ obtained in S ₄ . On the other hand, the value i _1A ′, a obtained in S ₅
Next a _1A ', the value a _2B tube A ₃ introduced from the memory in step S ₆ (radius _{r 3), a 2A, i} 2B, i 2A, I 2B, at step S ₇ by using the I _2A Is calculated.

a _2B ′ = S _2B (I _2B + I _2A ) a _2A ′ = S _2A (I _2B + I _2A ) i _2B ′ = a _2B ′ −I _2B i _2A ′ = a _2A ′ −I _2A I _2B ′ = i _1A + a _iA ′ In step S ₈ , i _2B ′, i _2A ′ obtained in S ₇ is used. The calculation of I _1A ′ = i _2B ′ + a _2A ′ is performed. Similarly, the radius r of the simulated acoustic tube is calculated, and πr ² is calculated from the calculation result to obtain the cross-sectional area. In step S _n-1 , a _nB ′ = f (I _nB ) _inB The calculation of ′ = a _nB ′ −I _nB I _nB = i _n−1A + a _n−1A is performed. Using the result, in step S _n , I _n-1A ′ = i _nB ′ + a _nB ′ is calculated. In other words, calculates the radius r ₁ ~r _n of the acoustic tube, to calculate the πr ₁ ² ~πr _n ² from the operation result, the operation result in the final stage (n stages) of the equivalent circuit corresponding to the cross-sectional area obtained Output is D / A converted and output to a speaker (not shown), and is output as sound from the speaker.

Therefore, since the arithmetic processing unit performs the arithmetic operation corresponding to the acoustic tubes A _{1 to} A _n , this arithmetic processing unit uses the electric current value and the function f of each part flowing through each equivalent circuit of A _{1 to} A _n of the acoustic tube. A memory having S _iB , S _iA (i = 1, 2 ... N-1) as a table, a first calculating means for calculating the current value of each part of the equivalent circuit, and the equivalent circuit are Second arithmetic means for calculating the current value of the equivalent circuit by using the current value of the adjacent equivalent circuit.

H. Effects of the Invention As described above, the present invention calculates the radius r of the acoustic tube, calculates πr ² from the calculation result, and the obtained cross-sectional area corresponds to the surge impedance of the equivalent circuit one to one. In consideration of the fact that the change in the cross-sectional area obtained by calculating the radius of the acoustic tube is simulated by the change in surge impedance, the parameters are:
Since it only needs to calculate the radius of the acoustic tube, it can be made extremely simple and does not require a lot of memory for speech synthesis as in the conventional case. Further, articulatory coupling is smoothly performed by the continuous change of surge impedance, and utterance close to natural sound can be obtained.

In addition, a musical sound can be easily obtained by changing the number of stages of the memory of the delay circuit, and a more natural musical instrument sound can be easily realized in combination with a change in the calculated value of the radius.

[Brief description of drawings]

1 to 9 are diagrams for explaining the present invention. FIG. 1 is an acoustic tube model diagram, FIG. 2 is an acoustic tube equivalent circuit diagram, and FIG. 3 is an electrical circuit equivalent model diagram of the acoustic tube. 4 is an explanatory view of changes in the vocal tract, FIG. 5 is an explanatory view of interpolation of the acoustic tube cross-sectional area with respect to time, FIG. 6 is an explanatory view of interpolation of the radius r of the acoustic tube according to the present invention with respect to time, and FIG. An electric circuit diagram electrically simulating propagation, FIG. 8 is an equivalent circuit diagram of FIG. 7, FIG. 9 is a flow chart diagram showing an example of a program for computer processing the sound synthesis of the present invention, and FIG. The articulatory coupling explanatory view by Percoll synthesis is shown. A ₁ , A ₂・ ・ ・ A _n …… Sound tube, r ₁ ,, r ₂・ ・ ・ r _n ………… Sound tube radius, Y ₁ , Y ₂ …… Sound admittance of sound tube, C ₁ , C ₂
… C _n …… Capacitance of electric circuit model, L ₁ , L ₂ … L _n ……
Same reactance, Z ₁ , Z ₂・ ・ ・ Z _n・ ・ ・ Surge impedance.

Claims (3)

[Claims] 1. In sound synthesis that simulates using an acoustic tube,
Two adjacent acoustic tubes are treated as a parallel circuit of a propagating current source and a surge impedance inversely proportional to the acoustic cross section, and the radius of the acoustic tube is calculated according to the shape change of the acoustic tube. When calculating the cross-sectional area by multiplying the (However, r _i and r _{i + 1} are radii of two adjacent acoustic tubes, and i is 1,
2,3 ... n) is used. 2. In the sound synthesis simulated by using an acoustic tube,
Two adjacent acoustic tubes are treated as a parallel circuit of a propagating current source and a surge impedance inversely proportional to the acoustic cross section, and the radius of the acoustic tube is calculated according to the shape change of the acoustic tube. When calculating the cross-sectional area by multiplying the (However, r _i and r _{i + 1} are radii of two adjacent acoustic tubes, and i is 1,
(2,3 ... n) and the current flowing into the surge impedance of the current source to be newly propagated to the next, and the sum of the current flowing into the acoustic tube, or the value of twice the current flowing into the surge impedance and the acoustic A sound synthesis method characterized by using a difference between a tube and a current source. 3. In sound synthesis that simulates using an acoustic tube,
Two adjacent acoustic tubes are treated as a parallel circuit of a propagating current source and a surge impedance inversely proportional to the acoustic cross section, and the radius of the acoustic tube is calculated according to the shape change of the acoustic tube. When calculating the cross-sectional area by multiplying the (However, r _i and r _{i + 1} are radii of two adjacent acoustic tubes, and i is 1,
(2,3 ... n) and has a delay circuit for the propagation of the propagation current source, and the length of an acoustic tube having the same area is changed by making the delay circuit variable. A sound synthesis method characterized by that.