JPS63257000A - Voice synthesization - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed description of the invention] A. Industrial application field The present invention relates to the synthesis of human speech or musical instrument sounds.

B1 Summary of the Invention The present invention focuses on the fact that the phenomenon of sound propagation caused by a change in the cross-sectional area of a sound tube is similar to a system transient phenomenon in an electric circuit such as a power transmission line, and the present invention is based on Sound is synthesized by varying the surge impedance in response to impedance changes.

In addition, sound synthesis by changing the length of the acoustic tube of a wind instrument, etc.
This is a sound synthesis method in which a delay circuit is provided in the impedance coupling section, and the delay constant corresponding to the change in the length of the delay circuit is changed to enable synthesis.

C. Conventional technology Electronic devices that artificially synthesize and output so-called sounds, such as voice synthesis and mycosic synthesizers (electronic musical instruments), have recently become LS for voice recognition and voice synthesis using one or several chips.
I can now be realized at low cost through audio information processing and semiconductor large-scale integrated circuit technology, and various systems have been proposed depending on the purpose of use and constraints. This speech synthesis method involves two methods: recording and editing raw human voices, combining them appropriately and editing them into sentences, and another method that does not directly use the human voice. There is a method of extracting only the parameters of a speech signal and controlling those parameters during the speech synthesis process to artificially create a speech signal.

There is a widely used PARCOR method that allows high-quality synthesized sounds to be obtained using this parameter method.

When handling audio using an electronic computer, the audio waveform is sampled at certain intervals, the audio signal value at each sampling point is converted from analog to digital, and the resulting values are displayed as codes of 0 and 1. To record faithfully to analog signals, it is necessary to increase the number of bits, but voice synthesis signals require a large amount of memory.

Therefore, various highly efficient encoding methods are being researched and developed in order to reduce the amount of information as much as possible.

One such method is the delta modulation method, which uses a minimum of 1 bit for information of one audio signal = 4-. This method uses one bit to determine whether the next audio signal value is higher or lower than the current value, and if it is high, it gives the code "l" and if it is low, it gives the code "0" to encode the audio signal. In the actual system configuration, a constant amplitude step amount (delta) is determined, and the audio value obtained by conventional encoding is compared with the input audio signal to prevent errors from accumulating. The residual signal is encoded.

This kind of configuration is called predictive coding, and is based on the linear prediction method (
There are two methods: the Percoll method (which uses a partial autocorrelation function called a Percoll coefficient instead of the prediction coefficient of the linear prediction method).

D1 Problems to be Solved by the Invention When predictive coding is used as in Moekei, there is a problem in that it is difficult to make articulatory connections, which correspond to joints between sounds.

In other words, in Figure 10, the horizontal axis represents the time E of sound generation, and the vertical axis represents the Percoll coefficient k.For example, in the utterance from a vowel to a consonant to a vowel, from the stationary state of the vowel to the transient state to the consonant. In addition, in the process of transitioning through vowels and reaching steady vowel sounds, the sound at the joint between vowels is cut off, which does not give a natural sound to humans.

Furthermore, in the case of musical instrument sound synthesis, the joints between scales are important, but since the synthesis method differs from the principle of sound generation in an actual musical instrument, it still lacks a natural sound, especially in reverberant sounds. In order to approximate natural sounds in both of these systems, there are problems such as the need for a large number of electronic components such as memory and arithmetic units, making the device expensive.

Means for Solving Problem E1 In view of the above points, the present invention proposes that human sounds and musical sounds of musical instruments are produced by changes in the shape of the human oral cavity and the length and cross-sectional area of the acoustic tube. Therefore, we analyzed the traveling wave phenomenon that represents the transmission of sound waves in acoustic tubes using acoustic tube equivalent circuits, and focused on the fact that the cross-sectional area of the acoustic tube is inversely proportional to the surge impedance, and simulated the cross-sectional area by changing the surge impedance. By continuously changing the surge impedance, it is possible to perform articulatory combination smoothly, making it easier to synthesize sounds similar to human speech, and improving the naturalness of speech. be.

In addition, changes in the length of the instrument were simulated by the number of stages in the delay circuit, and together with changes in cross-sectional area, it was possible to easily create a more natural instrument sound.

F1 effect Humans produce sounds by not moving their mouths, and wind instruments produce musical sounds by changing the length and shape of their acoustic tubes. In the present invention, since the cross-sectional area of the acoustic tube (the human vocal tract from the vocal cords to the lips can be considered as one acoustic tube) corresponds to the surge impedance of the equivalent circuit on a l to l basis, this surge impedance Changing this is the same as changing the cross-sectional area of the acoustic tube. Changing this surge impedance is extremely easy electrically, so it is possible to synthesize sounds exactly like human sounds. By continuously changing the impedance, it is possible to produce sounds that are close to natural.

Also, changing the length of the acoustic tube delays the traveling wave of the sound wave, so in terms of electrical circuits it is called a delay circuit (memory).
This corresponds to changing the number of stages, and can be simulated extremely easily by adjusting the constants of the delay circuit. Therefore, in combination with the cross-sectional area change, a more natural musical instrument sound can be easily achieved.

G. Example In order for sound to be radiated out of the mouth, a sound source is required, and this sound source is produced by the vocal cords. On the other hand, the vocal cords have the function of intermittently stopping exhalation by opening and closing two folds, and this intermittent action generates airflow called puffs, and when the vocal cords are tensed, tension is added to the folds and the frequency of the opening and closing of the folds increases. becomes high, producing a high-frequency puffing sound. When the exhalation flow is increased, the sound becomes louder.

When this sound source wave passes through a cylindrical acoustic tube like the vocal tract, a resonance phenomenon causes certain components of the sound waves from the open end to be emphasized and certain components to be attenuated, creating a complex vowel waveform.

Even if the sound source has the same waveform, it will be affected by the shape of the vocal tract that it passes through before being emitted from the lips. That is, if the shape of the vocal tract is constant, the spectral envelope will not change much even if the pitch or intensity of the sound source is changed. The vocal tract has an extremely complex shape depending on the vowel, but it can be divided into parts that do not change much and parts that change significantly. For example, as shown in FIG. 1, it can be assumed that two acoustic tubes having different lengths and cross-sectional areas, A, A2, are connected.

Figure 1 is an acoustic tube model diagram, Figure 2 is its equivalent circuit diagram,
The cross-sectional area is A1. This is a case where two different acoustic tubes are connected to A2.

Focusing on the connecting surfaces of the acoustic tubes, when the flow of sound waves has different cross-sectional areas, a phenomenon occurs in which a portion of the sound waves is reflected from the different surfaces. This phenomenon is the same as the transient phenomenon that occurs when impulse currents are passed through lines with different impedances in an electrical circuit.

As mentioned above, sound generation is performed by intermittent sound sources by the vocal cords, which is electrically equivalent to the intermittent application of impulses.

Sound is a type of vibration that can be transmitted through any gas, liquid, or solid, but in terms of an electrical circuit, it can be made to correspond to a lossless LC distribution circuit with no resistance.

And the electrical impedance (V/I) of this equivalent circuit
is JL/C, so when converted to the case of a sound wave, it becomes ρC', which is the speed of the sound wave, the air density ρ multiplied by the sound speed C', and the impedance in the sound field, that is, the acoustic impedance, depends only on the mass of the gas and the speed of sound. .

When acoustic tubes with different cross-sectional areas are connected, reflection occurs at the interface between them. This can be simulated by electrical surge impedance. In other words, the equivalent circuit in which blocks of acoustic tubes with different cross-sectional areas as shown in Fig. 1 are connected is the second one.
Here, the air density is ρ and the speed of sound is C′
Then, the acoustic admittances Y and Yt of each acoustic tube are given as follows.

However, Zl and Z2 are the acoustic impedance, and then the propagation current source from the adjacent block is It, 12,
If the current distribution determined by this is a I + a 2 and II+ 12 and the voltage at junction A is e, then a + ""i, + It, a2=i, +1゜j+=a
+-11+ j2=a2-1, and the propagation current source 11Z 12' of the adjacent block for the next step is +1' = il+a+, 12' -! 2+at.

-''In one set! Substituting 1''a+-11 and 12-a2+2, I,'=2a,-1.,12'=2a2-1.
You can close it.

FIG. 7 shows the interpolation situation of the cross-sectional area A with respect to time in the method described in 2. Here, linear interpolation, which is the simplest to calculate, is shown.

Figure 3 is an equivalent model diagram of the electrical circuit of an acoustic tube. Figure 9 (A) is a model diagram of the acoustic tube in which the vocal tract from the vocal cords to the lips is regarded as one acoustic tube, and Figure (B) is its electrical circuit. model diagram,
(1) Figure shows a traveling wave equivalent model diagram.

Before explaining Figure 3, I will explain how human vowels are made.

Figure 4 simulates the change in the cross-sectional area of the vocal tract when a sound is produced. Figure (A) shows that when a sound is uttered, the back of the throat is narrow and the lips are open and the sound is pushed out of the lungs. The vocal cords open and close intermittently during exhalation, and the sound waves that are reflected repeatedly in the vocal tract (acoustic tube) emerge as the sound waveform of "a". "I" is (A) If the throat is wide and the tip of the lips is narrow as shown in the figure, the sound waveform of "I" will be output.

In this way, the frequency is determined by the shape of the mouth, and if the shape of the mouth is simulated, "a" or "i" will be uttered.

The shape of the mouth can be simulated by the cross-sectional area of the sound tube, and changes in the cross-sectional area of the sound tube can be simulated by changes in surge admittance, and changes in surge admittance can be very easily varied using an electrical circuit. Figure 3 (a) shows the cross-sectional area AI, A2.
...An acoustic tube with a different cross-sectional area is connected to simulate the vocal tract. In the same figure (A), the acoustic impedance is replaced with an LC circuit of an electric circuit, and each sound tube is made into one LC line, and the whole is a concentrated line of n-1.
This is an electric circuit. In addition, Figure 3 (1) is a traveling wave equivalent model diagram, in which the acoustic impedance of each acoustic tube is 2,
．． 22.2. is inversely proportional to the cross-sectional area of the acoustic tube (acoustic admittance is proportional) and proportional to the speed of the sound wave. In addition, Z in the same figure. is the sound source impedance.

ZL indicates radiation impedance, and arrows between blocks indicate forward waves and backward waves.

Now, if you want to make the sound "A", the cross-sectional area A1 corresponding to the cross-sectional area of the tip of the lips in Figure 4 is = 16-, give the shape of the mouth of "A", and pulse P intermittently. By applying
In order to produce the sound "i" from the "i" sound, "i" can be obtained by narrowing the cross-sectional area to A + ' to give the appearance of the mouth of "i" as shown in the same figure (a).

When the impulse P is applied continuously and intermittently to change the entire cross-sectional area to look like the mouth of "i", the vocal tract is simulated by n acoustic tubes shown in Figure 3, so these By moving each cross-sectional area from "A" to "A", the shape of the mouth will change continuously to "A". Changing the cross-sectional area of the acoustic tube is done by gradually changing the surge impedance.

Therefore, as shown in Fig. 5, the cross-sectional area is continuously changed from A to A1', so of course the steady state sounds of "a" and "i" can be obtained, but the impedance also changes. Since it can be varied continuously, it is possible to obtain intermediate sounds, that is, sounds between two sounds. Therefore, as shown in FIG. 6, articulatory combinations similar to human pronunciation should be performed smoothly without any sound breaks.

Next, considering the propagation speed of the sound wave, it has length Q and L C
This is similar to the transient phenomenon that occurs when an impulse is applied to a wire with .

That is, the line having the LC as shown in FIG. 7 is equivalently represented as shown in FIG. 8. Here is the surge impedance Z seen from both ends. II ZO2 is ZOl-JL/C
, Zo, =&L/C.

Here, if we consider the traveling wave arriving from the other party as an equivalent current source, 1p = i + (t-τ) + -V + (-τ) o
l, and if there are n delay circuit blocks Z in the middle, the current is output after n hours. In other words, what occurred in the left circuit reached the right circuit one hour later.

(2) is the current i++ V+ (t-τ) on the feed pipe side. However, in the dinotal calculation, voltage or current is subdivided, so V, , V represents the voltage at measurement time t, and τ represents the elapsed time.

In Figure 8, if an impulse is applied to the L and C circuits, τ
After some time, it will come out to the output tube side. This equivalently represents that what has reached the τ time plane has also reached the other party. Setting the line length ρ to 1 can be easily calculated by normalizing the delay block n to 1. ρ
When dividing into 3 cm, n of the delay block should be set to 3 blocks.

The human vocal tract in Figure 3 (A) is approximately 17c1N for a male, so if we simulate it with 17 acoustic tubes in lam increments, the waveform that enters from A will be obtained by dividing the half-cycle current into 10, and then dividing the current at Δt. If it is 10μsec, it takes 170μsec to come out from the An side. If you think of the musical instrument trompon, the musical tone of the trompon changes depending on the length of the acoustic tube. According to the present invention, it is only necessary to have two parameters from the "a" sound of trompon to the "two" sound of trompon. All you need is two parameters: the upper tone. The intermediate tones can be freely articulated by changing the delay blocks of the delay circuit.

That is, the delay circuit ZI-+Zn, Zn→Z1 in FIG.
By varying , it is possible to respond to changes in the length of acoustic tubes that have the same area.

Next, the arithmetic processing of the traveling wave equivalent model shown in FIG. 3 will be explained based on the flowchart shown in FIG.

When an impulse is input to the acoustic tube A1, the arithmetic processing device consisting of a computer performs step S.

From memory at A, a OAI i QA, I
Take out OAI E. Based on the extracted value, in step S2, aoA'-f(E, IaA) i OA'-21OA' I . Perform the calculation of A. This calculated value a. A'+ioA' and step S
Values a+B, a+ of tube A2 introduced from memory in step 3
A, l IB, i+A, I IB+ I
In step S4 using IA, a IB' -9IB(I IB+ I IA)j
IB'-fL+B'I+.

! IA'-2LIA'IlB11B'
-1°A'-aoA' is calculated. In step S5, it was determined in S4!
Using IB' l aha', calculate I oA' = i + a + a + a. On the other hand, 4tl2 (direct j+A'
+2L+A' and the value of tube A3 introduced from memo 1) in step S6 i12B+ 12A+ 12B
The following calculation is performed in step S7 using +l 2A+T 2B+ 12-&.

a 2B' -92B (12B+ 12A)i2B'
-a2B' -12B j 2A'-12A' I 2A 12B' -i IA+ a IA' In step S8, 12B'11.
Using A', the calculation 11A'-1tB'+a2A' is performed. Thereafter, calculations corresponding to the cross-sectional areas A and ~An of the simulated acoustic tube are performed in the same manner, and in step Sn-1, anB' -f (InB) l nB'-and' I nB I nB ``'in-IA + a n-IA is calculated. Using the result, in step Sn, 11□A'-1nB' + a nB' is calculated. In other words, the Al-An of the acoustic tube is The output of the calculation result at the final stage (n stage) of the corresponding equivalent circuit is D/A converted and output to a speaker (not shown), and is output as audio from the speaker.

Therefore, since the arithmetic processing unit performs calculations corresponding to the acoustic tubes A, -A, this arithmetic processing unit calculates the current value of each part flowing through the equivalent circuit of each of the acoustic tubes A, -An, and the function f, SIB , 5iA (j=1°2...n-1)
a memory having as a table; a first calculation means for calculating the current value of each part of the equivalent circuit; and second calculation means for calculating.

H1 Effects of the Invention As described above, the present invention focuses on the fact that the cross-sectional area of the sound tube corresponds to the surge impedance of the equivalent circuit in a one-to-l ratio, and calculates the change in this cross-sectional area by the change in the surge impedance. Since there is a simulated grid, you only need to have the cross-sectional area as a parameter, so it can be done extremely easily.
Unlike conventional methods, it does not require a large amount of memory for speech synthesis.

Furthermore, articulatory coupling is also performed smoothly due to continuous changes in surge impedance, resulting in vocalizations that are close to natural sounds.

Furthermore, musical tones can be easily obtained by changing the number of memory stages in the delay circuit, and together with the change in cross-sectional area, more natural musical instrument sounds can be easily realized.

[Brief explanation of the drawing]

1 to 9 are diagrams for explaining the present invention, in which FIG. 1 is an acoustic tube model diagram, FIG. 2 is an acoustic tube equivalent circuit diagram, and FIG. 3 is an electric circuit equivalent model diagram of the acoustic tube. Figure 4 is an explanatory diagram of changes in the vocal tract, Figure 5 is an illustration of interpolation of acoustic tube cross-sectional area over time, Figure 6 is a detailed diagram of the present invention, and Figure 7 is an electrical simulation of sound propagation. An electric circuit diagram, FIG. 8 is an equivalent circuit diagram of FIG. 7, FIG. 9 is a flowchart showing an example of a computer processing program for sound synthesis of the present invention, and FIG. 1O is an explanatory diagram of articulatory combination by conventional Percoll synthesis. shows. A1. A2...Ao...Cross-sectional area of the acoustic tube, Yl, Y+
Acoustic admittance of acoustic tube, C,, C,...On capacitance of electric circuit model, L, 1. ··bow,. ...
ditto actance, 2. ．． 2. ...Zo Surge impedance.

Claims (3)

[Claims] (1) In sound synthesis simulated using acoustic tubes, two adjacent acoustic tubes are treated as a parallel circuit with a propagating current source and a surge impedance inversely proportional to the acoustic cross-sectional area, and the change in cross-sectional area due to the change in the shape of the acoustic tube is Correspondingly, Ai/(Ai+A_i_+_1), A_i_+_1/(
Ai+A_i_+_1) (Ai and A_i_+_1 are the cross-sectional areas of two adjacent acoustic tubes, and i is 1, 2, 3...n)
A sound synthesis method characterized by having coefficients of. (2) In sound synthesis simulated using acoustic tubes, two adjacent acoustic tubes are treated as parallel circuits with a propagating current source and a surge impedance inversely proportional to the acoustic cross-sectional area, and changes in cross-sectional area due to changes in the shape of the acoustic tube are Correspondingly, Ai/(Ai+A_i_+_1), A_i_+_1/(
Ai+A_i_+_1) (Ai and A_i_+_1 are the cross-sectional areas of two adjacent acoustic tubes, and i is 1, 2, 3...n)
The sum of the current flowing into the surge impedance and the current flowing into the sound tube, or the value twice the current flowing into the surge impedance and the current source of the sound tube, which has a coefficient of A sound synthesis method characterized by using differences. (3) In sound synthesis simulated using acoustic tubes, two adjacent acoustic tubes are treated as a parallel circuit of a propagating current source and a surge impedance inversely proportional to the cross-sectional area of the acoustic tube, and a delay circuit is added to the propagation of the propagating current source. and by making this delay circuit variable, a sound synthesis method is characterized in that it corresponds to changing the length of an acoustic tube having the same area.