JPS587198A - singing voice generator - 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 The present invention relates to a singing voice generating device that can synthesize and generate a singing voice with accompaniment by electronic means.

Now, there is a method of editing a singing voice by storing syllables of natural voices in advance for each of the 50 tones and scales, specifying the syllable and scale, and reading out the syllables of the desired pitch.

FIG. 1 shows a block diagram of a conventional singing voice generator using this method. In FIG. 1, 1 is a syllable storage section, 2 is a syllable selection section, 6 is a syllable selection control terminal, 4 is an I)/A converter, and 5 is a speaker.

The operation will be explained below. Syllable memory part 1 contains the sound of a,
All the syllables used for the i sound are falsified by scale. That is, several dozen audio signals are stored for each of the sounds A and B. Therefore, in order to sing the song "Haretaru Aozorata Dayou Kumoyo" as shown in Figure 2, the syllable selector 12 selects the scale E of C, E of shi, F of evening, G of ru, G of A, F. Noo, E, ZoD, Nou, C
evening, C da, D yo, E no tsu, E nori, E no mo, D 3. The syllables may be selected and read out in chronological order from the syllable storage unit 1 and passed through the D/A converter 4. 1. You must prepare syllables for each scale. ``Since there is no such thing, it is necessary to have a large storage capacity in the storage section 1.

Furthermore, conventional singing voice generators do not take into consideration the generation of singing voices with accompaniment that generates melodies along with lyrics.
), but it could not be said that σ) was sufficient.

The present invention has been made in view of the above-mentioned conventional technical circumstances, and therefore, an object of the present invention is to provide a singing voice generating device that requires a small storage capacity and is capable of generating a singing voice with an accompaniment example. Our goal is to provide the following.

In the singing voice generating device of the present invention, in order to reduce the storage capacity to 1.1'', the singing voice generating device of the present invention separates syllables into spectrum envelope information and pitch information, and stores only the spectrum envelope information. When generating a singing voice, the output signal output from the musical tone generator is used as pitch information, and the output signal of the musical tone generator is passed through a musical tone filter to create a melody, and the melody and singing voice are It is possible to generate a singing voice with accompaniment by synthesizing it and outputting it from a speaker.

Next, the present invention will be explained in detail, but before that, in order to make the present invention easier to understand, the speech synthesis technology used in the present invention will be briefly explained.

As a speech synthesis method, P uses a partial autocorrelation coefficient (hereinafter referred to as pA R ('OR coefficient)), which is a type of linear prediction coefficient.
There is an ARCOR synthesis method. This method is already well known in the speech research field, so a detailed description will be omitted.
Here, we will briefly explain the speech spectrum and pARcOR speech synthesis. Figure 3 shows a frequency analysis of the sound "o". (a) represents the entire spectrum.

This consists of a spectral envelope (A) that changes slowly with frequency and a spectral fine structure (c) that changes rapidly with frequency.
) can be decomposed into the product of The spectral envelope mainly reflects the resonance of the vocal tract and includes information on phonology, that is, whether it is "a" or "o". The spectral fine structure contains information about the period (pitch) of the sound, that is, the pitch of the sound. The PARCOR coefficient is physically a characteristic parameter representing vocal tract transmission characteristics. Therefore, sounds can be synthesized by reproducing the filter characteristics representing the phoneme using the pARcOR coefficient.

FIG. 4 shows the configuration of the PARCOR speech synthesizer.

In Fig. 4, the same symbols as in Fig. 1 indicate the same parts, and
In the figure, 6 is a noise generator, 7 is a pulse train generator, 8 is a voiced/unvoiced switch, 9 is a multiplier, 10 is a digital filter, 11 is pitch information, 12 is a voiced/unvoiced switching signal, 13 is volume information 14 is PARCOR coefficient 15 is an audio data storage section. The voice data storage unit 15 stores voiced and
Silent discrimination information, pitch information, volume information, PARCOR
Coefficients are stored in chronological order. Therefore, when synthesizing speech, digital filter 0 reproduces the filter characteristics representing the phoneme using the pARcOR coefficient, and if it is a voiced sound, it adds a pulse train to the filter at the period given by the pitch information, and if it is an unvoiced sound, it adds noise to the filter. Synthesize sounds by doing this.

FIG. 5 shows the configuration of digital filter 0.

In FIG. 5, 14-1 is a first-order PARCOR coefficient input section, and 14-2 is a second-order pARcOR coefficient input section 14.
-P is the second-order 1) A R (,' OR coefficient human power part,
14A.

14B is a multiplier, 14C and 1sD are adders, 14E is a delay memory, and 4RCOR coefficients are added to each multiplier. 16 is a pulse input terminal, 1
7 is a synthesized sound output terminal. When a pulse or noise is applied to the input terminal 16 of this filter, the response output from the output terminal 17 has the same spectral envelope characteristics as voice.

From the above, it seems that the gist of the speech synthesis device has been understood, so one embodiment of the present invention will be described below.

FIG. 6 is a block diagram illustrating an embodiment of the present invention.

In FIG. 6, the IIIJ control section 18 controls the tone generation section 2 according to the tone length and pitch data stored in the storage section 19.
Controls 0. That's naive.Similarly, the control section 18 also controls the storage section 19.
The lyrics data button stored in the button and the audio generating section 21 are controlled accordingly.

The musical tone generating section 20 determines the frequency frozen pitch and the gate signal, that is, the tone length, based on the frequency division value data from the control section 18, and outputs them as pitch information.

The sound generation section 21 uses the pitch information from the musical tone generation section 20 to determine the pitch of the sound to be generated.

The sound generation section 21 determines spectrum information of the sound to be generated based on the control signal from the control section 18 . In this way, the voice generating section 21 generates a singing voice according to the data stored in the storage section 19.

On the other hand, the musical tone filter 22 receives pitch information from the musical tone generator 20 and can output a melody. Therefore, the melody from the musical tone filter 22 and the voice generator 21
By mixing the singing voice with the singing voice, amplifying it with the amplifier 26, and outputting it from the speaker 24, it is possible to generate the singing voice with accompaniment.

FIG. 7 is a detailed block diagram of the sound generation section 21 in FIG. 6. In the figure, the digital filter 10 is
With reference to FIG. 4, it is possible to synthesize and output a singing voice using the pitch information 11 provided by the musical sound generation section 20 and the PARCO coefficient provided from the storage section via the control section 18. It will be clear from the above description of speech synthesis techniques.

FIG. 8 is a block diagram showing an example of the musical tone generating section 2001 in FIG. 6.

In the figure, the programmable frequency divider 25 divides the oscillation output signal from the main oscillator 26 into a frequency set according to the frequency division value data from the control section, and outputs it as a rectangular wave via the flip-flop 27.

This output signal is input to the NAN'l) circuit 28 and is turned ON/OFF by the gate signal/ON/'OFF signal from the control section.

The main oscillator 26 is a %CA, 1'OS inverter circuit 29
~31, capacitor 32. This is an oscillator composed of a variable resistor 33. The oscillation frequency can be arbitrarily varied by adjusting the resistance value of the variable resistor 63.

Tables 1 and 2 show that the oscillation station wave number of the main oscillator 26 is
529.76 KHz No. 9 C' 4- B
s'! ! The frequency division value for the pitch of T, the output frequency obtained at that time, the true frequency corresponding to the pitch, and the error of the output frequency with respect to the true frequency are shown.

It can be seen that the frequency error shown in the above table is within =20.1%, and has sufficient accuracy for the human's specific frequency discrimination limit (about 0.2%).

Margin 12 Tables 1 and 2 show the frequencies after being divided into 1/2 by the flip-flop circuit 27 in FIG. Here, the oscillation frequency of the main oscillator 26 is set to 12
．． If you adjust the frequency to a 25% higher frequency, you can transpose the pitch by a whole tone difference. In other words, if the key is C major, it can be transposed to D major.

In FIG. 6, the pitch information, that is, the information regarding the pitch of the sound, is not provided from the audio data storage section 19, but directly from the musical tone generation section 20.

Therefore, if the circuit shown in FIG. 6 is used, only the type of sound needs to be stored in the audio data storage section 19, so the storage capacity of the audio data storage section 19σ) is significantly reduced compared to the conventional system. be able to.

As described above, according to the present invention, singing voices can be generated with a small storage capacity, and a singing voice generating device can be constructed at low cost. Furthermore, since the melody and singing voice are output at the same time, the singing voice is accompanied by accompaniment, making it possible to create a sense of realism.

By adjusting the oscillation frequency of the main oscillator in the musical tone generator, it is possible to arbitrarily transpose the key.

Of course, it goes without saying that it is possible to output only the melody or only the singing voice.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an example of a conventional singing voice generator,
Figure 2 is an explanatory diagram showing a musical score, Figure 6 is a frequency analysis waveform diagram of a sound, Figure 4 is a block diagram showing the configuration of a PARCOR speech synthesizer, and Figure 5 shows the configuration of a digital filter in Figure 4. A circuit diagram, FIG. 6 is a block diagram showing an embodiment of the present invention, FIG. 7 is a detailed block diagram of the sound generation section in FIG. 6, FIG. 8 is a detailed block diagram of the musical tone generation section in FIG. 6, It is. Symbol Explanation 1...Syllable storage section 2...
...Syllable selection section 6...Syllable selection control terminal 4...D/A converter 5...
...Subika 6...Noise generator 7.
......Pulse train generator 8...
Selector switch 9... Multiplier 10...
...Digital filter 11...Pinch information 12...Switching signal 15...Volume information 14...p
, ARcOR coefficient 15...Audio data storage unit 16...Pulse input terminal 17...Synthesized sound output terminal 18...Control unit 19... ...Storage section 20...Music sound generation section 21...Speech generation section 22...Musical sound filter 25...
,・Amplifier 24...Speaker 25...Programmable frequency divider 26...
- King oscillator 27...Flip-flop 28...NAND gates 29-31...Inverter 32...Capacitor 63...Variable resistor brush 1 Figure 1 ') Figure 2 Haretaru Ao-no-ratagyoukutunyo-ro 3 Figure 1 + notification eyelid 4 Figure 5 Figure 5 O

Claims (1)

[Scope of Claims] 1) A musical tone generating section that creates and outputs a musical tone σ) pitch information based on given data, and a voice σ) that generates voice by synthesizing a tone using spectrum envelope information. , a musical tone filter, a storage section that stores data necessary for creating voice data and pitch information as the envelope information, and a control section, and the musical tone generation section is under the control of the control section. The necessary data is given from the storage unit to create the pitch information of the musical tone and output to the musical tone filter and the audio generation unit, and the audio generation unit generates the audio output from the storage unit under the control of the control unit. The lyrics of a musical tone are synthesized and output using the data and the pitch information provided by the musical tone generating section, and the musical tone filter receives the pitch information from the musical tone generating section and outputs the melody of the musical tone. A singing voice generating device characterized in that it synthesizes and outputs a 'C accompaniment singing voice. 2. In the singing voice generating device according to claim 1, the musical tone generating section comprises means for dividing the frequency of an oscillation output output from one main oscillator to create arbitrary pitch information. A singing voice generator featuring: 5) The singing voice generating device according to claim 2, wherein the singing voice generating device transposes by adjusting the oscillation frequency of the main oscillator.