JPS58211797A - Band split type vocoder - 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 band-splitting vocoder, and more particularly to a band-splitting linear predictor vocoder.

In recent years, practical linear predictive vocoders have become widespread due to the establishment of linear predictive speech analysis methods (LPC analysis methods).

This LPC analysis method approximates the sectrum envelope of speech using an all-pole model, but this has traditionally involved underestimating the formant bandwidth, which has been known empirically, and using the It is said that there are two drawbacks: the approximation of ormant is poor.

The above-mentioned first drawback is thought to occur because the poles of the frequency spectrum where the energy of the first formant is concentrated are excessively concentrated.

Recently, in order to prevent such concentration of poles at specific frequencies, the audio band is divided into multiple subbands, and LPG analysis of an appropriate order is performed on each subband to measure the appropriate dispersion of the poles. A band-splitting linear predictive analysis method has been proposed.

By appropriately selecting the band division, this method alleviates not only the first drawback but also the second drawback, and has the potential to become an effective means of contributing to improving the sound quality of linear predictive vocoders. It's hidden.

Conventionally, Shikajina-1lX et al., even when band division is performed in this way, the special characteristics of band division are not particularly taken into account for the excitation sound source used when producing synthetic sounds,
The conventional method is followed as is. Therefore, it has the disadvantage that it is not necessarily sufficient to bring the synthesized sound closer to the structure of the real voice and improve its naturalness.

The object of the invention is to obviate the above-mentioned drawbacks of the prior art.

The band division type vocoder of the invention (1) divides the audio transmission band into multiple bands by 11 degrees, and uses a waveform train with a controlled repetition frequency as a sound source for a predetermined low group side band. For the predetermined high group side band, there is provided means for performing speech synthesis using a waveform train having a controlled repetition frequency and noise as sound sources.

In addition, the band division type vocoder of the second invention divides the audio transmission band into a plurality of bands, and for voiced sounds whose voicing degree is in a predetermined medium range, a predetermined low group side band is used. For this, a waveform train with a controlled repetition frequency is used as a sound source, and for a predetermined high group side band, a waveform train with a controlled repetition frequency and noise are used as sound sources to synthesize speech. For voiced sounds whose frequency is 4 higher than the medium range, speech synthesis is performed using the waveform train having the controlled repetition frequency as the sound source for both the low group side band and the high group side band. For voiced sounds whose voicing degree is lower than the medium range, the waveform sequence with the controlled repetition frequency and the noise are used for both the low group side band and the high group side band. It has means for performing speech synthesis using as a sound source.

The present invention will now be described in detail with reference to one drawing.

1, 2, 3, and 4 are block diagrams showing one embodiment of the present invention. FIG. 1 is a block diagram showing the overall configuration, and FIG. Terotsuk diagram showing,
FIG. 3 is a block diagram showing the configuration of the receiving side of an embodiment of the invention (1), and FIG. 4 is a block diagram showing the configuration of the receiving side of the embodiment of the invention (2).

In this embodiment, as shown in FIG. It consists of a receiving side 2/group and a transmission line 3.

The transmitting side 1 further includes a low-pass filter and an A/D converter 101. as shown in FIG. Window processor 102.
7-Rie transformer 103. power spectrum memory 10
4. Low frequency side autocorrelation coefficient measuring device 105. Low band side linear prediction coefficient analyzer 106. High band side autocorrelation coefficient measuring device 107.
High-frequency side linear prediction coefficient analyzer 108. Voicing degree measuring device 109
．． It has a vivchi extractor 110 and an encoder 111.

Further, the receiving side 2 of the embodiment of the invention (1) includes a decoder 201. as shown in FIG. Low-pass LPC filter 202.
Low-frequency interpolator 203. Low side band filter 204. High-frequency side LPC filter 205. High frequency side interpolator 2069 frequency converter 207. High band filter 208. Low and high frequency synthesizer 209. D/A converter and low pass filter 210.
Pitch generator 211. Pocket waveform generator 212. Noise generator 213, variable synthesizer 214. It has a low band side variable gain amplifier 215 and a high band side variable gain amplifier @216t-.

Further, as shown in FIG. 4, the receiving side 2 of the embodiment of the invention (2) includes all the components shown in FIG. Contains.

Now, the audio waveform to be transmitted is input from the audio input terminal 1000 of the transmitting side 1 shown in FIG.
333[1z or higher high frequency components are cut off, the signal is sampled at a sampling frequency of, for example, 8000h, quantized into a 12-bit digital signal by an A/D converter, and supplied to the window processor 102. .

The window processor 102 temporarily stores the input quantized audio signal in an internal memory. This memory stores 1, for example, 30m5Ec (240 samples) of the quantized audio signal, and performs window processing to multiply this by a window function such as a Hamming window or a rectangular window. For example l
This is repeated at a QmsEc period, and this becomes the basic analysis period (hereinafter referred to as basic frame period).

Now, what about the windowed audio waveform data?

For each basic frame period, the 7-lier transformer 103 is supplied to the voicing degree measuring device 109 and the pitch extractor 110.

The Fourier transformer 103 uses the input audio waveform data that has been subjected to the window processing described above, performs a 7-lier transform on this to obtain the spectrum components of each frequency, and further calculates the spectrum components of each frequency by taking the square of this absolute value. It is converted into a power spectrum component in frequency and stored in the power spectrum memory 104. Thus memory 1
Each data representing a power spectrum component stored in 04 is freely read out by a low-frequency autocorrelation coefficient measuring device 105 and a high-frequency autocorrelation coefficient measuring device 107, and the autocorrelation coefficient it used for measurement.

Now, the low-frequency side autocorrelation coefficient measuring device 105 reads out the power spectrum components on the low-frequency side 1, for example, from O to 1333 of the power spectrum from the memory 104, and performs an inverse Fourier transform on this to determine the required range. The autocorrelation coefficient at each delay time is measured and supplied to the low-band linear prediction coefficient analyzer 106. At the same time, the measuring device 105 outputs the measured autocorrelation coefficient with a delay time of 0 as the short-time average power on the low frequency side in this basic frame period via the output line 1050.

The signal is supplied to the encoder 111.

The analyzer 106 converts the parameters from the supplied set of autocorrelation coefficient data to a predetermined order, for example, by performing autocorrelation (AUTOCOI'LRELATI ON).
) method and other linear predictive analysis methods.

The extracted low frequency side parameters are supplied to encoding@111.

On the other hand, the high-frequency side autocorrelation coefficient measuring device 107 measures the high-frequency side of the power spectrum, 1333, , H in the above example.
By reading out the power spectrum components from z to 3333 Hz from the memory 104 and subjecting them to inverse Fourier transform, the autocorrelation coefficient at each delay time within the necessary range is measured, and this is used as the high-frequency side linear prediction coefficient. Analyzer 10
Supply to 8. However, in calculating the above-mentioned 7-lier inverse transform, the power spectrum component t- from 1333Hz to 3333Bz.

Shift the frequency by 1333Hz to the lower frequency, and
The inverse Fourier transform is performed on the power spectrum of 2000 tlz, and the autocorrelation coefficient is measured. In addition, the measuring device 107 uses the autocorrelation coefficient of the measured delay time O as the high-frequency side short-time average power in this basic frame period.

It is supplied to encoder 111 via output line 1070.

Note that the inverse Fourier transform performed by the above-described low-frequency autocorrelation coefficient measuring device 105 and high-frequency autocorrelation coefficient measuring device 107 is performed when the power spectrum is a scalar quantity (a quantity containing only real components and no imaginary components). Therefore, it is not necessarily limited to the 7-lier inverse transform, and either the Fourier transform or the cosine transform may be used.

From the supplied data set of autocorrelation coefficients, the analyzer 108 modulates the parameters to a predetermined order.

For example, it is extracted using a linear predictive analysis method such as an autocorrelation method, and the parameters are encoded on the extracted high frequency side.
11.

For details of the above-mentioned autocorrelation method, see, for example, John Makhoul.
): “Linear Prediction, y (Li-near
Prediction ) Near Tute Riff kV View (A Tutorial Review
)”, Proceedin-gs o
f the IEEE, vol, 63. /16
4. pp, 561-580. April, 19
See 75.

Furthermore, when analyzing the linear prediction coefficients on the low-frequency side and the high-frequency side by the single analyzers 106 and 108.

As mentioned above, the frequency band of the audio signal input to these is from 0 to 1333[1z] on the low side, and from O to 2000[]z on the high side, and the highest frequency is also the original sample. The maximum frequency of 4,000 ports2 determined by the conversion period is limited to 1/3 and 1/2, respectively, so the sampling period when performing all linear prediction coefficients is three times the original sampling period. and double decimate (dec
imate ) l, which is equivalent to using .

Now, the voicing degree measuring device 109 receives the window-processed audio waveform data, measures the voicing degree of the audio signal in this frame every 7 basic frame periods, and uses the measurement results as a useful tool. Output line 1090 as degree display signal
ft to the encoder 111. Koya voicing degree measuring device 109 is.

For example, it is possible to apply the voiced/unvoiced discrimination device disclosed in Japanese Patent Application Laid-Open No. 54-151303, and use the value of the discriminant obtained there as the usefulness display signal.

The pitch extractor 110 extracts pitch frequency data in each basic frame period from the supplied window-processed audio waveform data.

This is supplied to all encoders 111.

The encoder 11.1 encodes the various data supplied in this way to create a transmission frame, and transmits one transmission frame for each basic frame to the receiving side 2/'l via the transmission path 3.
Send to.

Now, on the receiving side 2/z, the above-mentioned transmission frame transmitted via the transmission path 3 is received, and audio is synthesized according to it as described below.

First, the receiving side 2 of an embodiment of the (1) invention shown in FIG. 3 will be described.

On the receiving side 2, as shown in FIG. 3, the transmission frame transmitted via the transmission path 3 is received.

The signals are sequentially supplied to the decoders 20,1.

The decoder 201 reproduces each data on the transmitting side by decoding these transmission frames.

Each of these data is supplied as follows.

First, the reproduced low frequency side parameters are supplied to the low frequency side LPC filter 202, and similarly, the reproduced high frequency side parameters are supplied to the high frequency side LPC filter 205. Output the reproduced usefulness display signal 2 in 2010
'Th is supplied as a variable synthesis control signal to the variable synthesizer 214. Furthermore, the reproduced pitch frequency data is supplied to the pitch generator 211 as a pitch frequency control signal.

The decoder 201 further supplies the reproduced low-frequency short-time average power to the low-frequency variable gain amplifier 215 as low-frequency gain control information via an output line 2011. Similarly, the reproduced high-frequency short-time average power is supplied to the high-frequency variable gain increaser 216 as high-frequency gain control information via the output fins +'- and 2012t-.

Now, the pitch generator 211 generates impulse data having a repetition frequency specified by the pitch frequency data, and supplies this to the onomatopoeic distortion waveform generator 212.

The onomatopoeic distortion waveform generator 212 generates an onomatopoeia waveform having a predetermined waveform at a time position specified by the supplied impulse data.

That is, as shown in FIG. 5(5), the output data from the pitch generator 211 is generated at the pitch pulse generation time point 1.
o, 1, 1.1, 2. . . ., the onomatopoeic distortion waveform generator 212 . -Ha.

As shown in Fig. 5, a predetermined pseudo vocal fold waveform as shown in Fig. 5 (q) is generated based on the time point of each specified pitch pulse. The voice distortion waveform generator 212 is a ROM (read-only memory)
This can be easily realized by writing all of the predetermined pseudovocal waveform data into the ROM and sequentially reading out the contents of this ROM at the time each pitch pulse is generated. The purpose of using the pseudophonic vocal cord waveform generator 212 is to make the sound source excitation data described later closer to the real voice.

In some cases, this can be omitted (see Figure 5 f).
The pitch pulse train shown in FIG. 5 (5) can also be used as is in place of the onomatopoeia waveform movement shown in B).

Now, the output of the pseudophonic waveform generator 212 is fed as one input signal of the variable synthesizer 214 via the line 2120 t- on the one hand and the 2-in 2121' on the other hand.
It is supplied to the input of the low-band side variable gain amplifier 215 via the input signal I.

The variable synthesizer 214 receives, on the one hand, the aforementioned onomatoid waveform motion supplied via line 2120 and, on the other hand, the noise data supplied from the noise generator 213 via line 2130e via line 2010. variable synthesis is performed in response to control of the variable synthesis control signal by the voicing level indicating signal.

In other words, the 1 synthesizer 214 multiplies the output P of the pseudovocal waveform generator 212 by the weighting coefficient WP, multiplies the output N of the noise generator 213 by the weighting coefficient WN, adds the two, and calculates the output E by E = WP P + WN The loading coefficients Wp and Wti are
This is a predetermined function of the voicing level indicating signal.

That is, the weighting coefficient WP is a specific function that increases as the voicing degree V displayed by the voicing degree display signal increases,
In addition, the weighting coefficient WN is the voicing degree displayed by the voicing degree display signal.
Let it be a specific function that increases as the value decreases. As a result, in the output 2140 of the variable synthesizer 214, when the voicing degree ■ indicated by the voicing degree display signal is extremely high, the onomatopoeia waveform movement becomes overwhelmingly larger than the noise, and as a result, the noise is 2 By ignoring @, and conversely, when the voicing degree V is extremely low, the noise becomes overwhelmingly larger than the onomatopoeia waveform movement. 1 As a result, the output 2140 can be regarded as only noise. is synthesized into The combined output of the variable combiner 214 thus combined is supplied to the input of the high-frequency side variable gain amplifier 216 via an output line 2140.

Now, the low frequency side variable gain amplifier 215 is connected to the line 2121.
The pseudovocal waveform motion supplied through the t-ray y201
1k is loaded with the low-frequency side short-time average power supplied through the excitation signal line 215.
0 ft to the low-frequency side LPG filter 202 as sound source excitation data.

Further, the high-frequency side variable gain amplifier 216 is made variable by loading the composite output of the variable combiner 214, which is supplied via the 2-in 2140 gold, with the high-frequency side short-time average power supplied via the line 2012. The signal is amplified and sent to the high-frequency side LPC filter 20 via the excitation signal line 2160.
5 as sound source excitation data.

Now, the low-pass LPG filter 202 that receives the regenerated low-pass parameters converts the supplied parameters into α parameters, and uses the α parameters as filter coefficients of the t-LPC filter. and synthesizes the sound source excitation data supplied via the line 2150, this filter coefficient, and the audio waveform data on the lower frequency side.

This is supplied to the low frequency side interpolator 203.

Due to the aforementioned decimation on the transmitting side, the sampling period of the low-frequency audio waveform data synthesized from the low-frequency parameters is three times the normal sampling period. The low-band interpolator 203 interpolates the supplied audio waveform data 11c1333[1z by passing it through a low-pass filter to create audio waveform data with a normal sampling period, and passes this to the low-band filter 204. supply

The low band filter 204 removes frequency components in unnecessary bands by passing the supplied data through a band filter having a band from 300 flz to 1333 flz, for example, to generate low band audio waveform data. It is supplied to one input of the high frequency synthesizer 209.

On the other hand, the high-frequency side LPC filter 205 that has received the parameter on the reproduced high-frequency side converts the supplied parameter into an α parameter, and converts the α parameter (filter coefficient of the i-LPG filter) into an α parameter. The sound source excitation data supplied via the 2-in 2160e, this filter coefficient, and higher-frequency audio waveform data are synthesized, and this is supplied to the higher-frequency interpolator 206.

Due to the processing on the transmitting side 1 described above, the parameters on the high frequency side are 1333[(z to 3333B) of the original audio signal.
0 to 2000 by frequency shifting the component of z
This is a parameter for a voice waveform transmitted in the band up to Hz and decinotated to a sampling period twice the normal rate. Therefore, the audio waveform data synthesized from this parameter has a sampling period that is twice the normal sampling period.
times t).

Furthermore, the waveform is shifted to a lower frequency by 1333EIz.

Therefore, the high-frequency interpolator 206 interpolates the supplied audio waveform data by passing it through a low-pass filter of t'2000[1z to create audio waveform data with a normal sampling period. 207.

The frequency converter 207 shifts the frequency of the audio waveform by 1333 tlz by adding a sine wave total loss of 1333 Elz to the supplied audio waveform data, and supplies this to the high-frequency sideband filter 208 .

The high side band filter 208 filters the supplied data to 13
By passing the signal through a bandpass filter having a band from 33Hz to 3333Elzt, frequency components in unnecessary bands are removed to generate high-frequency audio waveform data.

It is supplied to the other input side of the low and high frequency synthesizer 209.

The low frequency high frequency synthesizer 209 adds and synthesizes the supplied low frequency audio waveform data and high frequency audio waveform data. Thus, synthesized speech data of the band-splitting linear predictive vocoder is generated at its output, and is supplied to the D/A converter and low-pass filter 210.

The D/A converter and low-pass filter 210 converts the supplied synthesized voice data into an analog voice signal using a D/A converter, and further blocks components of 4000 EIz or more using a low-pass filter, thereby converting the synthesized voice data into one synthesized voice signal. It is output from the output terminal 2000 as .

As is clear from the above explanation, the vocoder of this embodiment divides the audio band into a high frequency side and a low frequency side, and sets a repetition frequency controlled by the extracted pitch frequency for the low frequency side. pseudovocalic waveform movement (or pitch pulse train) with
is used as a sound source.

Furthermore, for the high frequency side, speech synthesis is performed using a variable synthesis of the onomatopoeia waveform movement (or pitch pulse train) and noise as a sound source.

This is true when modeling the sound generation mechanism.

In a relatively low frequency range, air vibration becomes a volume flow and the repetitive structure due to pitch pulses becomes clear, but in a relatively high frequency range, the air vibration does not become a volume flow and the pitch structure becomes unclear. 1 The synthesized voice is made closer to the real voice and its naturalness is improved.

In the above embodiments, different sound source configurations are used for the low frequency band and the high frequency band.

This is a more faithful 11-bit version of the voice generator, but the one that goes even further is the one shown below (
This invention is 2).

Next, an embodiment of the second invention will be described in detail with reference to the drawings.

The transmitting side 1 of this embodiment is completely the same as the transmitting side embodiment of the invention (1) shown in FIG.

On the other hand, the receiving side 2 is as shown in FIG.

In addition to all the components of the receiving side embodiment of the invention (1) shown in FIG. 3, it also includes a sound source switching/control section 217.

This sound source switching/control unit 217 has an output line 2120 of the onomatoid waveform generator 212 that supplies the onomatoid waveform motion as its first input, and a synthetic output of the onomatoid waveform motion and noise as its second input. output line 2140 of variable combiner 214 that supplies
an output line 2170 for supplying the decoded voicing level indicating signal of 01; Output line 21 supplies high-frequency side sound source information to the high-frequency side variable gain amplifier 216 as the output of 2.
71 't- has. This sound source switching/control unit 21
7 features line 2010'? In response to the optically indicated voicing level V of the voicing level indicating signal supplied through the voicing level display signal, the output line 2170 and the output line 2171 are selectively reconnected to the input lines 2120 and 2140 as shown in FIG. It is to be.

That is, for the angular V displayed by the angular display signal supplied via the line 2010t, two predetermined threshold values vh and Vl are selected such that vh>vz, and the angular V is vh''. When the range is 2v>vl, this is considered to be in the middle angular region, and in this case, the line 21 input to the low frequency side variable gain amplifier 215
70 is connected to a line 2120 that supplies an onomatopoeic waveform train, and a 2-in 2171 that is input to the high-frequency side variable gain amplifier 216 is connected to a line 2140 that supplies a combined output of an onomatopoeic waveform train and noise. As a result, in the medium angular region, similar to the embodiment of the first aspect of the invention, the onomatoid waveform array is used as the excitation sound source on the low-frequency side, and as the excitation sound source on the real high-frequency side. The synthesized output of the onomatopoeia waveform sequence and noise is used.

On the other hand, when the angular V is in the range v>vh, this is defined as being in a high angular region, and in this case, each input to the low frequency side and the high frequency side 2-in 2170 and line 2171 are both connected to line 2120 that supplies the pseudophonic waveform train. As a result, in the high-level angular region, the onomatopoeic band waveform sequence is used as the same excitation sound source on both the low-frequency side and the high-frequency side.

Furthermore, when the angular V is in the range of V<Vl, this is defined as being in a low angular region, and in this case, each line 21 input to the low frequency side and the high frequency side
70 and line 2171 are both connected to line 2140 which supplies a composite output of the onomatopoeia waveform sequence and noise. As a result, in the low angular region, a synthesized output in which the onomatoid waveform sequence and noise are synthesized as the same excitation sound source is used on both the low-frequency side and the high-frequency side.

As described above, when this embodiment is used, it is possible to respond in an angular manner and supply an even more appropriate excitation sound source to the low and high frequency sides than the embodiment of the first invention. .

In addition, in the embodiments of the inventions (1) and (2), the transmission band is only divided into two parts, the high-band side and the low-band side. The frequency band is divided into a band belonging to the low group side and a band belonging to the high group side, and for the band belonging to the low group side, the sound source is the same as the corresponding sound source for the low frequency side in the above embodiment. Using the sound source.

Furthermore, for the band belonging to the high group side, a sound source having the same configuration as the sound source for the high band side in the corresponding embodiment described above may be used.

Also, the 133 used to separate the high-frequency side and the low-frequency side
The division frequency of 3[1z is merely an example and is not limited to this.

Similarly, the decimate ratio is just an example.

Furthermore, as the sound source waveform having a controlled repetition frequency, an onomatopoeia waveform sequence or a pitch pulse sequence can be used, as already explained.

Furthermore, in each of the embodiments described above, when performing band division on the transmitting side, the power spectrum of the entire band is first obtained and divided into each band, but instead of this, it is handled on the time axis. It is also possible to adopt a configuration in which the input waveform is divided using a bandpass filter and then converted into a base band by frequency shifting, and this waveform is decimated according to the bandwidth before linear predictive analysis is performed.

Furthermore, on the synthesis side of each of the above embodiments, output waveforms on the time axis were obtained for discrimination on the low-frequency side and high-frequency side, and these were synthesized on the time axis, but instead of this, as shown below. It is also possible to generate waveforms on the time axis after synthesizing them in the frequency domain.

That is, the spectral envelopes of the low and high frequency sides are calculated using the parameters of the low and high frequency sides reproduced by the decoder 201, and the short-term averages of the reproduced low and high frequency sides are calculated. Power information (Line 2011 and Line 2
012) to calculate the total and band autocorrelation coefficients. α parameters for all bands are calculated from the all-band autocorrelation coefficients by linear prediction analysis, and all band LPC filters are constructed using this α parameter as all filter coefficients. Also.

A full-band predicted residual power is calculated from the short-time average power information on the low-band side and high-band side and spectral envelope information on the low-band side and high-band side, and this is applied to the full-band LPC filter described above. Used as a gain control signal for the full-band variable gain amplifier that supplies the excitation signal.

On the other hand, sound source information to be used as an input to this full-band variable gain amplifier is generated as follows. That is, the signal i″ to be supplied to the low-frequency side variable gain amplifier 215 of the above-described embodiment
That is, line 2121 or line 2170 )t
's output once passed through a low-frequency sound source band filter having a constant low-frequency band characteristic of 5%, and a signal to be supplied to the high-frequency variable gain amplifier 216 of the above-mentioned embodiment (i.e., 2-in 214゜ or line 2171) is combined with the output passed through a high-frequency sound source band filter having specific high-frequency side band characteristics by addition, and this is supplied as input sound source information to the above-mentioned full-band variable gain amplifier. do.

By employing the method described above, an embodiment can be obtained in which a processing format is employed in which parameter information for each band is synthesized in the frequency domain. In addition, the band characteristics of the above-mentioned low-side sound source band filter, illumination, and high-side sound source band filter are made variable, and the voicing degree display signal (line 201
0)'if: can also be used to control the optimal band characteristics in response to the degree of voicing.

As described above, the present invention has the effect of making the synthesized sound of the band-splitting vocoder closer to the structure of the real voice and improving its naturalness.

[BRIEF DESCRIPTION OF THE DRAWINGS] FIG. 1 is a block diagram showing the overall configuration of an embodiment of the invention (1) and (2), and FIG. 2 is a block diagram showing details of the transmitting side of the embodiment. FIG. 3 is a proof diagram showing details of the receiving side of the embodiment of the invention (1), and FIG. 4 is a block diagram showing details of the receiving side of the embodiment of the invention (2). Figure 5 (5) shows the 〈Tschipa, Rus sequence, Figure 5 (
Bl is a time chart for explaining the onomatopoeia waveform movement, and FIG. 5 is a time chart for explaining the onomatopoeia waveform. FIG. In the figure, 1... the transmitting side, 2... the receiving side of the embodiment of the (1)th invention, 2... the (2nd)
) Receiving side of the embodiment of the invention, 3... Transmission line, 1
01...Low pass filter and A/D converter, 1
02...Window processor, toa...
・Fourier transformer, 104...Power spectrum memory, 105...Low band side autocorrelation coefficient measuring device, 106...Low band side linear prediction coefficient analyzer,
107...High frequency side autocorrelation coefficient measuring device, 108
...High frequency side linear prediction coefficient analyzer, 109...
...voicedness measuring instrument, 110...pitch extractor, 111...signature device, 201...
Decoder, 202...Low band side LPC filter,
203...Low frequency side interpolator. 204...Low band filter% 205...
...High-frequency side LPC filter, 206...High-frequency side interpolator. 207... Frequency converter, 208...
High band side band filter, 209...Low frequency synthesizer, 210...D/A converter and low pass filter, 211...Pitch generator, 212... ...
...Vocal fold waveform generator. 213...Noise generator, 214...Variable synthesizer. 215...Lower side variable gain amplifier, 216...
. . . High-frequency side variable gain amplifier, 217 . . . Sound source switching/control unit. Purification of drawings! ) (There is no sad history in the content) Drawings 8 (There is no sad history in the content ＼ □□□ Handshake, continuation of the amendment.) Director of the Japan Patent Office 1, Indication of the case 1982 Patent Application No. 0959
No. 29 No. 2, Title of the invention Bandwidth-splitting vocoder 3, Relationship to the amended case Applicant 5-33-1 Shiba, Minato-ku, Tokyo (423) NEC Corporation Representative Tadahiro Sekimoto 4, Agent Address: 108, 37-8 Shiba 5-chome, Minato-ku, Tokyo Resident Mita Building; Rokomansori 4kn2 Ozakino 0) Drawings subject to C correction Plane Z Contents of correction Mainly correct drawings 3 and 4 01 Agent Patent Attorney Susumu Uchihara -Kuchi+-

Claims (2)

[Claims] (1) The audio transmission band is divided into multiple bands, and a waveform train with a controlled repetition frequency is used as the sound source for the predetermined low group side band, and for the predetermined high group side band. A band division type vocoder comprising means for performing speech synthesis using a waveform sequence having a controlled repetition frequency and noise as sound sources. (2) The audio transmission band is divided into multiple bands, and for voiced sounds whose degree of voicing is in a predetermined medium range, a predetermined low group side band has a controlled repetition frequency. Speech synthesis is performed using a waveform train as a sound source and, for a predetermined high group side band, a waveform train with a controlled repetition frequency and noise as sound sources. For voiced sounds whose voicing degree is higher than the medium range, the waveform train having the controlled repetition frequency is used as a sound source for both the low group side band and the high group side band. Perform synthesis. For voiced sounds in a range whose voicing degree is lower than the medium range, the waveform sequence and noise having the controlled repetition frequency are used as sound sources for both the low group side band and the high group side band. What is claimed is: 1. A band-splitting vocoder, characterized in that it has means for performing speech synthesis.