JP3483958B2 - Broadband audio restoration apparatus, wideband audio restoration method, audio transmission system, and audio transmission method - Google Patents

Abstract

PURPOSE: To obtain the wide-band speech restoring device which restores a wide-band speech signal of high quality by estimating a stable wide-band sound source with a correct amplitude from a narrow-band speech or narrow-band speech code without being affected by a difference in speaker and a noise. CONSTITUTION: This device is equipped with a narrow-band sound source decoding means 108 which generates a narrow-band synthesized sound by using the narrow-band speech code, a spectrum decoding means 105 which estimates a wide-band spectrum parameter by using a narrow-band spectrum code separated from the narrow-band speech code, a wide-band sound source decoding means 106 which estimates a wide-band sound source signal by using a narrow-band sound source code separated from the narrow-band speech code, and a synthesizing means which generates a wide-band speech signal from the generated narrow-band synthesized sound, the estimated wide-band spectrum parameter, and the wide-band sound source signal.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a wideband voice restoration device for restoring a wideband voice signal from a narrowband voice signal whose band is limited or a narrowband voice code obtained by encoding a narrowband voice signal.

[0002]

2. Description of the Related Art An example of a narrow band voice signal is present telephone voice. Voice signal is about 300H in the telephone system
The data is transmitted in a limited band from z to 3.4 KHz, and the sound quality is poor and solid as compared with the case where there is no band limitation. To improve the quality, it is possible to construct a telephone system capable of transmitting a wideband voice signal, but it requires a lot of time and cost.

Japanese Patent Laid-Open No. 6-118995 discloses a conventional method for restoring a wideband voice signal from a narrowband voice limited to a telephone band, which is considered as a wideband voice restoration method.

Japanese Unexamined Patent Publication No. 6-118995 discloses an LPC analysis of a narrow band speech signal to calculate a spectrum parameter,
This spectral parameter is vector-quantized using a narrow band codebook. Then, the wideband spectrum parameter is decoded using the wideband codebook learned in association with the narrowband codebook. L using this spectral parameter
PC synthesis processing is performed to obtain a provisional wideband audio signal. A final wideband speech signal is generated by extracting the band components other than the narrowband speech signal from the provisional wideband speech signal and adding them to the upsampled narrowband speech signal. It should be noted that when performing wideband LPC synthesis processing, a wideband sound source signal is required, but a method for generating this sound source signal is not specifically disclosed.

As a document having the same configuration as that of Japanese Patent Laid-Open No. 6-118995 and disclosing the generation of a wideband sound source signal, reference 1 “Recovery of wideband speech from narrowband speech by codebook mapping”, IEICE, There is a study report SP93-61 (1993-08).

[0006] In this document 1, two methods are disclosed as a broadband sound source generation method. The first method uses a pitch and power obtained by analyzing a narrow band speech to generate a sound source by a method common to those skilled in the art. That is, for voiced sound, an impulse train that repeats at a pitch period,
Unvoiced sound produces white noise, whose power determines its amplitude. In Document 1, some post-processing is performed to improve sound quality. In the case of restoring the low frequency band of 300 Hz or less, the power of the low frequency band restoration sound is multiplied by a few times in order to compensate for the insufficient power of the restoration band. 3.4Hz to 7.3KH
When the high frequency band of z is restored, the cosine function is applied to collapse the pulse in order to reduce the pulse-like sound generated by using the impulse train as the sound source.

The second method vector-quantizes the spectrum parameter of the narrow band speech signal and selects a narrow band representative waveform segment and a high band representative waveform segment corresponding to the obtained code. Then, the following processing is performed on these two waveform elements. The voiced and unvoiced of the waveform segment is determined, and in the case of voiced sound, the waveform segment is superimposed in synchronization with the pitch obtained by analyzing the narrow band speech signal. In the case of unvoiced sound, a signal of a required length is cut out from a random position of the waveform segment. The power ratio between the synthesized sound and the narrowband speech synthesized from the narrowband waveform segment by using the signal generated by the above process and the narrowband spectrum parameter is calculated. Then, a synthesized sound is generated from the high-frequency waveform segment by using the signal generated by the above-mentioned processing and the wide-band spectrum parameter, and the power ratio is multiplied to obtain a high-frequency restored signal.

As another method for expanding the band of the sound source signal although the field of use is different, reference 2 "A 2.4Kbps High-Qauli
ty Speech Coder "IEEE International Conference on
Acoustics, Speech, and Signal Processing vol.1, S
Some are disclosed in 9.5, pp. 589-592 (1991.5).

[0009] Document 2 relates to a method of highly efficiently encoding and decoding telephone band voice, and in order to reduce the amount of information of a sound source at the time of encoding, 0 Hz to 3.4 KHz.
The long-term predictive analysis is performed on the sound source signal of and the long-term predictive residual signal is separated. The long-period prediction residual signal of 0 Hz to 3.4 KHz is band-limited to 0 Hz to 1 KHz for encoding. Then, a long-period prediction residual signal of a telephone band up to 3.4 KHz is generated from the band-limited long-period prediction residual signal at the time of decoding, and then a long-cycle synthesis process is performed to restore the sound source signal. Is. Restoration of the long-period prediction residual signal is performed by up-sampling a signal having a component from 0 Hz to 1 KHz to a sampling frequency of 8 KHz, leaving it at 4 sample intervals, and setting the other values to zero.

[0010]

The above-mentioned conventional methods have the following problems. Japanese Patent Laid-Open No. 6-118995,
Although it is another document, Document 1 which discloses a concrete practical example thereof is roughly divided into the following four problems, namely, sound source amplitude estimation, sound source generation method, spectrum parameter estimation method, and application to communication system. There is. First, the first sound source amplitude estimation will be described. When the first sound source generation method of Reference 1 is used, as the power used for synthesis of the restored sound, the power value obtained by analyzing the narrow band speech is used as it is or after being multiplied by a constant. Since the gain of the synthesizing filter is different between the spectrum parameter of Eq. 1 and the estimated spectrum parameter of the wide band, the amplitude of the synthetic sound obtained differs even if the same sound source amplitude is given.
Since this difference changes for each frame, there is a problem that wide-band speech having the correct amplitude cannot be restored by multiplying the sound source amplitude, that is, the power value by a constant. Further, when the second sound source generation method of Reference 1 is used, a narrow-band synthesized sound is generated, a power ratio with a narrow-band speech is calculated, and the high-frequency synthesized sound is multiplied. There is a problem that it is necessary to execute complicated processing for.

Next, the second sound source generation method will be described. When the first sound source generation method of Document 1 is used, a wide band sound source signal is generated only with a small amount of information such as pitch and power, and therefore it is not possible to sufficiently estimate an originally wide band sound source that varies. As a result, although the pulse-like sound is reduced by the cosine function, there is a problem that the pulse-like sound cannot be completely suppressed and the sound quality becomes unnatural. In addition, 1 voiced sound source whose characteristics differ greatly for each speaker
Since it is impossible to express with one fixed sound source, there is a problem that the sound quality deteriorates depending on the speaker. When the second sound source generation method of Reference 1 is used, the representative waveform segment corresponding to the code of the vector quantization result of the spectrum parameter is used, but the spectrum parameter originally depends on the shape of the vocal tract,
Since the sound source waveform depends on how the vocal cords vibrate, there is no strong correspondence between the two. The sound source waveform depends on the speaker rather. Therefore, there is a problem that an appropriate sound source is not selected.

As described in Document 1, this second
When the sound source generation method is used, unvoiced waveform segments are selected even if they are voiced, or conversely, voiced waveform segments are selected even if they are unvoiced. However, there is a problem that quality deterioration occurs when the synthesis is performed as it is. In order to avoid this, the power ratio in that part is forcibly set to 0. However, as a result, the amplitude of the restored high frequency band becomes partially 0, which causes another problem of quality deterioration. is there. Further, both of the sound source generation methods have a problem that quality deterioration due to voiced / unvoiced judgment and pitch extraction error cannot be avoided. In particular, when applied to a narrowband speech signal on which noise is superimposed, there is a problem that judgment errors and extraction errors increase, resulting in large deterioration. Further, since there are only two modes, voiced sound and unvoiced sound, a sound source having an intermediate property cannot be sufficiently expressed, and there is a problem that quality deterioration occurs at a boundary portion between voiced sound and unvoiced sound.

Next, the third spectral parameter estimation method will be described. In Japanese Patent Laid-Open No. 6-118995 and Document 1, vector quantization and inverse quantization using two codebooks are performed. However, a memory for accumulating the codebooks is required, and for the quantization processing. The problem is that a large amount of computation is required. Further, noise, unvoiced sound, and voiced sound can be easily discriminated from each other, and the correspondence between the narrowband spectrum parameter and the wideband spectrum parameter changes depending on the discrimination. However, in each case, since the spectral parameter and the power are treated independently, the information about the power is not reflected in the estimation of the broadband spectral parameter. Therefore, if the shapes of the narrow band spectra are similar, there is a problem that the same wide band spectrum is estimated regardless of the magnitude of the power.

Finally, application to the fourth communication system will be described. When applying the methods of Japanese Patent Laid-Open No. 6-118995 and Document 1 to a communication system, after decoding a narrow band synthetic sound from a received voice code, reanalyzing this narrow band synthetic sound to restore a wide band speech signal. However, when the spectrum parameter and the sound source information are separated and encoded and then transmitted, it is considered more efficient to directly use the speech code to restore the wideband speech signal. That is, JP-A-6-11
The methods of 8995 and Document 1 have a problem that they are inefficient in that reanalysis is required. Further, the parameters obtained by performing the synthesis and the re-analysis have a distortion due to the interpolation during the synthesis, the windowing during the analysis, and the like, and the quality of the wideband speech is deteriorated.

In addition, Japanese Patent Laid-Open No. 6-118995 and Document 1 do not add a signal processing process that is generally introduced to reduce the noise sensation of synthesized speech and improve intelligibility.
When the quality of the restored wideband audio signal is insufficient, there is a problem that it cannot be improved. Further, when applied to a communication system, signal processing may be applied to a narrow band synthesized sound, and in order to superimpose a processed narrow band speech signal and an unprocessed wide band speech signal, both There is a problem that the continuity of sound quality deteriorates.

According to the method of Document 2, if 0 Hz to 1 KHz is considered as a narrow band and 0 Hz to 3.4 KHz is considered as a wide band,
This means that wideband sound source signal estimation is performed, but as described above, this method takes a wideband speech signal as an input, encodes the parameters obtained by analyzing this, and decodes it to obtain a wideband synthesized sound. However, it does not disclose a method of restoring a wideband voice signal from a narrowband voice signal or a parameter extracted from the narrowband voice signal.

The present invention has been made to solve the above problems, and an object of the present invention is to realize a wideband voice restoration device which restores a wideband voice signal having a correct amplitude from a narrowband voice. Another object of the present invention is to realize a wideband speech restoration device having a relatively simple process of estimating the amplitude of a wideband sound source. Furthermore, it is an object of the present invention to realize a wideband speech restoration device which has a small dependency on a speaker and estimates a wideband sound source which is good even in the vicinity of a voiced / unvoiced boundary, and restores stable and natural sound quality wideband speech. Another object of the present invention is to realize a wide-band speech restoration device that is less susceptible to voiced / unvoiced decision errors and pitch extraction errors that tend to occur in narrow-band speech signals on which noise is superimposed. Furthermore, when it is applied to a communication system, it is another object of the present invention to realize a wideband speech restoration device that efficiently restores wideband speech without performing reanalysis. Furthermore, when the quality of the restored wideband speech signal is insufficient, it can be improved, and when the signal processing is applied to the narrowband synthesized sound, the processed wideband speech signal with good continuity in narrowband. The purpose is to realize a wideband voice restoration device that can obtain

[0018]

A wideband speech restoration apparatus according to the present invention uses an analyzing means for analyzing a narrowband speech signal to obtain a narrowband spectrum parameter and a narrowband source signal, and using the narrowband spectrum parameter. A spectrum estimating means for estimating a wideband spectrum parameter, a wideband sound source estimating means for estimating a wideband sound source signal using a narrowband sound source signal, and a synthesizing means for generating a wideband speech signal from the estimated wideband spectrum parameter and the wideband sound source signal. Equipped with.

Further, as the wide band sound source estimating means, a zero padding means for inserting a predetermined zero value into each sample interval of the input narrow band sound source signal is used.

The wideband excitation estimating means analyzes the input narrowband excitation signal to obtain a narrowband adaptive excitation code and a narrowband driving excitation signal, and wideband adaptation using the narrowband adaptive excitation code. An adaptive sound source estimating means for estimating a sound source signal, a driving sound source estimating means for estimating a wide band driving sound source signal using a narrow band driving sound source signal, and a wide band sound source signal from the estimated wide band adaptive sound source signal and wide band driving sound source signal And adding means for generating.

Alternatively, the wideband source estimation means analyzes the input narrowband source signal to obtain a narrowband long period prediction code and a narrowband long period prediction residual signal, and the narrowband long period prediction residual. A long period prediction residual estimation means for estimating a wide band long period prediction residual signal using a difference signal, and a wide band long period prediction code estimating means for estimating a wide band long period prediction code using a narrow band long period prediction code, It is composed of a long-period synthesizing means for synthesizing a wide-band source signal from the estimated wide-band long-period prediction residual signal and the wide-band long-period prediction code.

Another wideband speech restoration apparatus of the present invention uses analysis means for analyzing a narrowband speech signal to obtain a narrowband spectrum parameter and narrowband amplitude information, and the narrowband spectrum parameter and narrowband amplitude information. A spectrum estimating means for estimating at least wideband spectrum parameter or wideband amplitude information, and a synthesizing means for generating a wideband speech signal from the estimated wideband spectrum parameter and wideband amplitude information or wideband source signal.

Alternatively, there is provided a wide band estimating means for estimating a wide band speech signal using a narrow band speech signal, and a post filter means for performing post filtering on the estimated wide band speech signal.

Or, the analyzing means for analyzing the narrow band speech signal to obtain the narrow band spectrum parameter, the spectrum estimating means for outputting the wide band spectrum parameter by using the narrow band spectrum parameter as it is as the wide band spectrum parameter, and the outputting means. A synthesizing means for generating a wideband speech signal from wideband spectral parameters is provided.

Alternatively, an analyzing means for analyzing a narrow band speech signal to obtain a narrow band spectrum parameter, and converting the narrow band spectrum parameter into another domain as necessary, transforming it, and inversely transforming into the spectrum parameter domain. And a spectrum estimating means for outputting a wideband spectrum parameter, and a synthesizing means for generating a wideband speech signal from the outputted wideband spectrum parameter.

Another wideband speech restoration apparatus of the present invention comprises a spectrum decoding means for estimating a wideband spectrum parameter from a narrowband speech code, and a synthesizing means for generating a wideband speech signal from the estimated wideband spectrum parameter.

Alternatively, a spectrum decoding means for estimating a wideband spectrum parameter using a narrowband spectrum code separated from the narrowband speech code and a wideband excitation signal using the narrowband excitation code separated from the narrowband speech code. A wide band sound source decoding means for estimating and a synthesizing means for generating a wide band speech signal from the estimated wide band spectrum parameter and the wide band sound source signal are provided.

Further, as the wide band excitation decoding means, a zero padding means for inserting a predetermined zero value into each sample interval of the narrow band excitation signal restored from the narrow band excitation code is used.

Alternatively, the wideband excitation decoding means estimates the wideband adaptive excitation signal using the narrowband adaptive excitation code separated from the input narrowband speech code, and separates it from the input narrowband speech code. Wideband drive excitation decoding means for estimating a wideband drive excitation signal by using the narrowband drive excitation code described above, and addition means for generating a wideband excitation signal from these estimated wideband adaptive excitation signal and wideband drive excitation signal. did.

Alternatively, the wideband excitation decoding means estimates the wideband long-period prediction code by using the narrowband long-period prediction code separated from the input narrowband speech code, and the input narrowband. Wideband long-period prediction residual decoding means for estimating a wideband long-period prediction residual signal using a narrowband long-period prediction residual code separated from a speech code, and these estimated wideband long-cycle prediction code and wideband long-cycle prediction And a summing means for generating a wideband source signal from the residual signal.

Alternatively, a narrowband amplitude information decoding means for estimating narrowband amplitude information using a narrowband excitation code separated from the narrowband speech code, and a narrowband spectrum code and narrowband separated from the narrowband speech code. Spectrum decoding means for estimating at least wideband spectrum parameter or wideband amplitude information using amplitude information, and combining means for generating wideband speech signal from the estimated wideband spectrum parameter and wideband amplitude information or wideband source signal as necessary Equipped with.

Alternatively, there is provided wideband speech decoding means for estimating a wideband speech signal using a narrowband speech code, and postfilter means for performing postfiltering on the decoded and estimated wideband speech signal.

[0033]

In the wideband speech decompressor according to the present invention, the wideband speech signal is synthesized from the wideband spectrum parameter estimated using the narrowband spectrum parameter and the wideband sound source signal estimated using the narrowband sound source signal.

Further, a wide band source signal is generated by inserting a predetermined number of zeros between each sample of the narrow band source signal, and a wide band speech signal is synthesized using this and the estimated wide band spectrum.

In estimating the wideband excitation signal, the input narrowband excitation signal is analyzed to calculate the narrowband adaptive excitation code and the narrowband driving excitation signal, and the wideband estimated using this narrowband adaptive excitation code. The adaptive sound source signal and the wide band drive sound source signal estimated using the narrow band drive sound source were added to obtain a wide band sound source signal. A wideband speech signal is synthesized using this and the estimated wideband spectrum.

As another method of estimating the wideband excitation signal, the narrowband long-period prediction code and the narrowband long-period residual signal are calculated by analyzing the input narrowband excitation signal, and the narrowband long-period prediction code is calculated. A wideband long-period prediction code estimated using
A wideband source signal was obtained using the wideband long-period residual signal estimated using the narrowband long-period residual signal. A wideband speech signal is synthesized using this and the estimated wideband spectrum.

Further, another wideband speech decompression apparatus according to the present invention analyzes the narrowband speech signal to calculate the narrowband spectrum parameter, the narrowband amplitude information and the narrowband source signal, and outputs the narrowband spectrum parameter and the narrowband amplitude. The information is used to estimate wideband spectral parameters, wideband amplitude information, or both. Then, a wideband speech signal is combined with these signals and a wideband sound source signal estimated from the narrowband sound source signal.

Further, in another wideband speech reconstructing apparatus according to the present invention, the wideband speech signal estimated by using the narrowband speech signal is post-filtered, and the high frequency characteristic is mainly processed.

Further, in another wideband speech decompression apparatus according to the present invention, the characteristic of the narrowband spectrum parameter is expanded over the entire area and used as the wideband spectrum parameter to synthesize the wideband speech signal.

Another wideband speech decompression apparatus according to the present invention obtains a wideband spectrum parameter by inversely converting the narrowband spectrum parameter up to a specific order and converting it into a corresponding spectrum parameter, and using the wideband spectrum parameter. The audio signal is synthesized.

Further, another wideband speech decompression apparatus according to the present invention uses a narrowband speech code to generate a narrowband synthesized speech and estimates a wideband speech signal, and upsamples the narrowband synthesized speech or a narrowband speech. A wideband speech signal is synthesized by adding to the synthesized speech a signal obtained by extracting a component mainly in a high band other than the narrowband synthesized speech of the wideband speech signal.

Another wideband speech decompression apparatus according to the present invention uses a wideband spectrum parameter estimated using a narrowband spectrum code and a wideband speech signal estimated using a narrowband excitation code to generate a wideband speech signal. Is synthesized.

Furthermore, the wide band excitation decoding means
A wideband source signal is generated by inserting a predetermined number of zero values between each sample of the narrowband source decoded using the narrowband source code, and a wideband speech signal is synthesized using this and the estimated wideband spectrum. To be done.

Further, another wideband excitation decoding means in the present invention adds the wideband adaptive excitation signal estimated using the narrowband adaptive excitation code and the wideband driving excitation signal estimated from the narrowband driving excitation signal to add the wideband excitation. A signal is generated. A wideband speech signal is synthesized using this and the estimated wideband spectrum.

Further, the wideband long-range prediction signal estimated by the other wideband excitation decoding means in the present invention using the narrowband excitation code and the wideband long-cycle residual signal estimated from the narrowband long-cycle prediction residual signal. The broadband sound source signal is synthesized from and. A wideband speech signal is synthesized using this and the estimated wideband spectrum.

Further, in another wideband speech decompression apparatus according to the present invention, one or both of the wideband spectrum parameter and the wideband amplitude information are estimated using the narrowband spectrum code and the narrowband amplitude information. Then, a wideband speech signal is synthesized with these pieces of information and the wideband source signal estimated from the narrowband source signal.

Further, in another wideband speech reconstructing apparatus of the present invention, the wideband speech signal estimated by using the narrowband speech code is subjected to post-filtering, and the high frequency characteristic is mainly processed.

[0048]

【Example】

Example 1. An embodiment of the present invention will be described with reference to the drawings.
The present embodiment mainly describes the configuration and operation for restoring the generation of a wideband sound source signal in a more correct form. Figure 1
FIG. 1 is a configuration diagram of a wideband voice restoration device according to a first embodiment of the present invention. In the figure, 1 is an input narrowband speech signal, 2 is analysis means, 3 is spectrum analysis means, 4 is narrowband spectrum parameters, 5 is an inverse filter, 6 is narrowband source signal, and 7 is
Is a wideband spectrum estimation means, 8 is a vector quantization means, 9 is a narrowband spectrum codebook, 10 is a spectrum code, 11 is a dequantization means, 12 is a wideband spectrum codebook, and 13 is a wideband spectrum parameter. Reference numeral 14 is a wide band sound source estimation means which is an important new component in the present embodiment, 15 is a zero padding means as a specific example, 16 is a wide band sound source signal, 17 is a synthesis filter as a synthesis means, 1
8 is a bandpass filter, 19 is upsampling means, 2
0 is a broadband audio signal. In addition, FIG.
It is a signal explanatory view explaining the process of No. 5.

The operation of the first embodiment of the present invention will be described below with reference to FIGS. 1 and 2. First, the narrow band audio signal 1 sampled at, for example, 8 KHz and limited to the telephone band from 300 Hz to 3.4 KHz is input to the analysis unit 2 and the up sampling unit 19. The spectrum analysis means 3 in the analysis means 2 analyzes the narrow band speech signal 1 to calculate the narrow band spectrum parameter 4, and outputs it to the inverse filter 5 and the wide band spectrum estimation means 7 in the analysis means 2. As the narrow band spectrum parameter 4, various ones such as a linear prediction coefficient, LSP, PARCOR coefficient, and cepstrum can be applied. The inverse filter 5 inversely filters the narrowband speech signal 1 using the narrowband spectrum parameter 4, and outputs the obtained narrowband sound source signal 6 into the wideband sound source estimation means 14.

The vector quantizing means 8 in the wide band spectrum estimating means 7 vector-quantizes the narrow band spectrum parameter 4 using the narrow band spectrum codebook 9,
The obtained spectrum code 10 is output to the inverse quantization means 11 in the wide band spectrum estimation means 7. Inverse quantizer 1
1 dequantizes the spectrum code 10 using the wideband spectrum codebook 12 and outputs the obtained wideband spectrum parameter 13 to the synthesis filter 17. The processing in the wideband spectrum estimating means 7 is the same as that in the reference 1, and the detailed description of the method of generating the narrowband spectrum codebook 9 and the wideband spectrum codebook 12 and the method of vector quantization will be omitted.

The zero padding means 15 in the wide band sound source estimating means 14 which is an important part of this embodiment inserts zeros by M-1 samples between each sample value of the narrow band sound source signal 6 and obtains M times. The signal of the sample number of is output to the synthesis filter 17 as the wideband sound source signal 16. Here, M is a value obtained by dividing the sampling frequency of the wideband audio signal to be restored by the sampling frequency of the narrowband audio signal. In this embodiment, the case where M is 2 will be described. FIG. 2A shows N
It is the narrowband source signal 6 of the sample. When zero-filling processing is performed on this signal by the zero-filling means 15, M-1 samples, that is, zeros for each sample are inserted between the samples, and a wideband sound source of 2N samples shown in FIG. 2B is obtained. The signal 16 is obtained. When M is zero-padded to 2, a spectrum symmetrical with 0 Hz to 4 KHz with respect to a frequency half the sampling frequency of the wideband audio signal, that is, 4 KHz is restored from 4 KHz to 8 KHz.

The synthesis filter 17 uses the wideband spectrum parameter 13 to subject the wideband sound source signal 16 to synthesis filter processing to generate a temporary wideband speech signal. The band-pass filter 18 performs band-pass filter processing on this temporary wide-band speech signal to extract components other than the band in which the narrow-band speech component exists. Bandwidth of wideband audio signal is 0H
In the case of z to 7.3 KHz, the components of 0 Hz to 300 Hz and 3.4 KHz to 7.3 KHz are extracted. The upsampling means 19 upsamples the narrowband audio signal 1 by M times. The signal generated by upsampling has the same sampling frequency as that of the wideband speech signal 20 and has the same narrowband component as that of the narrowband speech signal 1. Then, the output of the bandpass filter 18 and the output of the upsampling means 19 are added to generate a wideband audio signal 20.

Originally, the narrowband source signal and the wideband source signal are
Since the characteristics of the sound source signal generated from the same vocal organ are reflected, there is a correlation in the characteristics of the sound source signal such as the strength of the harmonic component of the pitch frequency and the strength of the noise component between the harmonic components. . That is, when the narrow-band sound source signal has a regular characteristic that the harmonic component of the pitch frequency is strong,
Similarly, the broadband source signal has a regular characteristic that the harmonic components of the pitch frequency are strong, and conversely, if the narrowband source signal has a strong noise component, the broadband source signal also Similarly, the noise component has a strong characteristic. By configuring the wideband sound source estimating means as in this embodiment, it is possible to generate a wideband sound source signal of 0 to 8 KHz having the same characteristics as the narrow band sound source signal of 0 to 4 KHz in the low frequency range, and thus the speaker It has the effect of being able to restore a stable and natural wide-band voice with little dependence on. Also, unlike the conventional example, voiced / unvoiced judgment and pitch extraction are not required, and since this configuration can also express a sound source of an intermediate nature, a voiced / unvoiced judgment error that is likely to occur in a narrowband speech signal on which noise is superimposed. It is possible to estimate a good wideband sound source even in the vicinity of a voiced / unvoiced boundary without being affected by the pitch extraction error, and to restore a stable and natural wideband sound.

Example 2. FIG. 3 is a configuration diagram of the sound source estimation means 14 in the wideband speech restoration device according to the second embodiment of the present invention. In the figure, a new part is 21 sound source analysis means, 2
2 narrow band adaptive codebook, 23 distortion minimizing means, 24 narrow band driving excitation signal, 25 narrow band adaptive lag length, 26 narrow band adaptive gain, 27 wide band driving sound source estimating means, 28
Zero padding means, 29 wideband driving excitation signal, 30 wideband adaptive excitation estimating means, 31 wideband adaptive excitation codebook, 32
Wideband adaptive source signal of 33, wideband adaptive lag length of 33, 34
Is a wideband adaptive gain of. Since the overall configuration is the same as that of FIG. 1, the description of the configuration and the operation of parts other than FIG. 3 are omitted. According to this configuration, the broadband sound source signal can be restored even better.

The operation of the embodiment of the present invention will be described below with reference to FIG. The narrowband sound source signal 6 is input to the sound source analysis means 21 in the wideband sound source estimation means 14. The narrowband adaptive codebook 22 in the sound source analysis unit 21 stores the past narrowband sound source signal 6, and the lag length is an integer value according to the lag length sequentially output by the distortion minimizing unit 23 described later. In this case, the stored narrowband sound source signal 6 is repeatedly output with this lag length to output a signal. If the lag length is a non-integer value, then reference 3 “Pitch Predictor
s with High Temporal Resolution "IEEE Internatio
nal Conference on Acoustics, Speech, and Signal Pr
As described in ocessing vol.2, S12.6, pp.661-664 (1990.4), the signal is generated and output by the polyphase filter output. The length of the output signal is the same as that of the current narrowband sound source signal 6.

FIG. 4 shows an example of the past narrowband excitation signal 6 stored in the narrowband adaptive codebook 22 and the signal output according to the input lag length. In the figure, the horizontal axis indicates time, and time indicates that time elapses in the direction of the arrow. (A
1) and (B1) thus indicate the temporal length of the sound source signal, (A2) and (B2) indicate the lag length normalized to the output time, such as 20 to 128, and (A3) ),
(B3) shows an example of the sound source signal to be output. Figure 4 (a)
Indicates a case where the length of the output signal is shorter than the lag length. In this case, the sound source signal (A3) having the length of the output signal time T1 is output subsequently to the past sound source signal from the beginning of the lag length. When the lag length is shorter than the output signal length, such as T2, the same sound source signal (B3) is output multiple times as shown in FIG. 4B.
Is repeated to output the sound source signal following the past.

The distortion minimizing means 23 sequentially outputs a plurality of lag length values to the narrow band adaptive codebook 22 and applies a gain to the signal output by the narrowband adaptive codebook 22 for each lag length. The gain is determined so that the distortion between the multiplied signal and the narrow band sound source signal 6 is minimized. Then, the one that minimizes the distortion is selected from all the lag lengths, and is output to the wideband adaptive sound source estimation means 30 as the narrowband adaptive lag length 25.
Also, the gain value at that time is output to the wideband adaptive excitation estimation means 30 as the narrowband adaptive gain 26, and the narrowband adaptive signal 6 and the signal obtained by multiplying the signal output by the narrowband adaptive codebook 22 by the narrowband adaptive gain 26. The error signal is output as the narrow band driving sound source signal 24 to the wide band driving sound source estimating means 27. As a method of determining the gain in the distortion minimizing means 23, a generally known Lagrange's undetermined coefficient method can be used. That is, the distortion minimizing means 23 receives the narrowband excitation signal 6 and the output of the narrowband adaptive codebook 22 as input, and has the minimum lag length 25 and the gain of 2 which are the narrowband adaptive excitation codes.
6 and the narrow band driving sound source signal 24 of the error signal are output.

The zero padding means 28 in the wide band driving sound source estimating means 27 has M between the sample values of the narrow band driving sound source signal 24.
Zeros are inserted for each −1 sample, and the obtained signal having M times the number of samples is output as the wideband driving sound source signal 29. Here, M is a value obtained by dividing the sampling frequency of the wideband audio signal to be restored by the sampling frequency of the narrowband audio signal, and the operation of inserting zero is the same as the zero padding means 15.

In the wideband adaptive sound source estimation means 30, first, the narrowband adaptive lag length 25 is multiplied by M to obtain the wideband adaptive lag length 33.
Then, the narrow band adaptive gain 26 is multiplied by g to obtain the wide band adaptive gain 34. When g is set to 1, the pitch periodicity of the finally obtained wideband source signal 16 becomes equal to that of the narrowband source signal 6, and as the value becomes smaller from 1, the pitch periodicity becomes weaker than that of the narrowband source signal 6. Go. When observing the actual speech, it is often the case that the pitch periodicity becomes weaker at higher frequencies, so setting g to a value smaller than 1 can restore higher quality wideband speech when restoring high frequencies. . The wideband adaptive excitation codebook 31 in the wideband adaptive excitation estimator 30 stores past wideband excitation signals 16
Is stored, and this signal is stored in the wideband adaptive lag length 3
The signal obtained by repeating 3 is output. Then, in the wide band adaptive sound source estimation means 30, this signal is multiplied by the wide band adaptive gain 34 and output as a wide band adaptive sound source signal 32. Finally, the wideband driving sound source signal 29 and the wideband adaptive sound source signal 32 are added and output as the wideband sound source signal 16.

With this configuration, the characteristic of the pitch periodicity of the narrow-band excitation signal and the variation thereof are well expressed by the narrow-band adaptive lag length 25 and the narrow-band adaptive gain 26, and are converted into the wide-band excitation signal. Since this is reflected, it is possible to sufficiently estimate a sound source that changes in various ways, and to restore wideband speech with good sound quality without pulse-like sound. Further, there is an effect that an appropriate sound source can be estimated regardless of the speaker. In the wideband adaptive sound source signal 32, the fundamental frequency determined by the wideband adaptive lag length 33 and the frequencies of the harmonic components thereof are correctly aligned at positions of integral multiples. Restoration Wideband components are well connected, and there is an effect that high-quality wideband speech can be restored. Further, since the characteristic that the pitch periodicity becomes weaker as the frequency becomes higher can be introduced by the coefficient g, there is an effect that more natural sound quality can be obtained. In addition, since voiced / unvoiced judgment and pitch extraction are not required, and a sound source with an intermediate property can be expressed, there is no influence of voiced / unvoiced judgment error or pitch extraction error that tends to occur in a narrowband speech signal on which noise is superimposed, Even in the vicinity of the voiced and unvoiced boundaries, a good wideband sound source can be estimated, and there is an effect that a stable and natural wideband speech can be restored.

Example 3. FIG. 5 is a configuration diagram of the wideband driving sound source estimation means 27 in the wideband speech restoring device according to the third embodiment of the present invention. In the figure, the new parts are 35 power calculation means and 36 noise generation means. Since other configurations are the same as those in FIGS. 1 and 3, the description of the operation of the corresponding portions will be omitted.

The operation of the portion shown in the drawing of the third embodiment of the present invention will be described below with reference to FIG. The narrow band driving sound source signal 24 is input to the power calculating means 35 in the wide band driving sound source estimating means 27. The power calculation means 35 calculates the power of the narrow band drive sound source signal 24 and outputs it. The noise generating means 36 generates and outputs a power-normalized white noise signal. Then, in the wideband driving sound source estimation means 27, the white noise signal is multiplied by the power output from the power calculating means 35, and the obtained signal is wideband driving sound source signal 29.
Output as.

The pitch period and the strength of periodicity are changing every moment. Since small fluctuations in the pitch period and the strength of periodicity in the narrow band excitation signal 6 cannot be expressed by the narrow band adaptive lag length 25 and the narrow band adaptive gain 26, the error is included in the narrow band driving excitation signal 24. . When the wideband driving sound source signal 29 is generated by using the narrowband driving sound source signal 24 including the error component as in the second embodiment, unnecessary disturbance may occur in the wideband driving sound source signal 29, and the power is the same. It has been experimentally confirmed that a better restored sound may be obtained when white noise is generated and used as the wideband driving sound source signal 29. With the configuration as in the third embodiment, white noise having the same power as that of the narrow band driving sound source signal 24 is generated and used as the wide band driving sound source signal 29. Therefore, in addition to the effect of the second embodiment, the pitch period is increased. The effect is to obtain a good restored sound with less disturbance due to fluctuations in the strength of the periodicity.

When the zero padding process is performed, a symmetrical spectrum centering around 4 KHz is generated. Therefore, this 0Hz
From 0 to 300 Hz and from 3.4 KHz to 4.0 KHz, zero-filling is performed on the narrow band driving sound source signal 24, and 0 Hz to 300 Hz, 3.4 KHz to 4.6 K
Therefore, a signal having no components of Hz and 7.7 KHz to 8 KHz is obtained. On the other hand, with this configuration using white noise, the wideband driving sound source signal 29 having all the components from 0 Hz to 8 KHz can be obtained, so that there is an effect that a good restored sound having a band over the entire region can be obtained. Especially 0Hz
The effect is great when the restoration is performed from 1 to 300 Hz.

Example 4. FIG. 6 is a block diagram of the wideband driving sound source estimation means 27 in the wideband speech restoring device according to the fourth embodiment of the present invention. In the figure, 28 padding means, 35 power calculating means, and 36 noise generating means are the same as those in the second and third embodiments. Since other configurations are the same as those in FIGS. 1 and 3, the description of the operation of the parts other than those shown is omitted.

The operation of the embodiment of the present invention will be described below with reference to FIG. The narrow band driving sound source signal 24 is inputted to the zero padding means 28 and the power calculating means 35 in the wide band driving sound source estimating means 27. The zero padding means 28 in the wide band drive sound source estimation means 27 inserts zeros by M-1 samples between each sample value of the narrow band drive sound source signal 24 and outputs the obtained signal of M times as many samples. Here, M is a value obtained by dividing the sampling frequency of the wideband audio signal to be restored by the sampling frequency of the narrowband audio signal, and the operation of inserting zero is the same as the zero padding means 15. The power calculation means 35 calculates the power of the narrow band drive sound source signal 24 and outputs it. The noise generating means 36 generates and outputs a power-normalized white noise signal. Then, a signal obtained by multiplying the signal output by the zero padding unit 28 by the gain gr1 and a signal obtained by multiplying the white noise signal output by the noise generation unit 36 by the power output by the power calculation unit 35 and further by the gain gr2 are added. Then, it is output as a broadband drive sound source signal 29.

The restored sounds according to the second and third embodiments are
If each has advantages and disadvantages, configure it like this,
By properly setting gr1 and gr2, there is an effect that a wideband voice having a quality higher than those of both can be restored. It also has the same effect as the second and third embodiments.

Example 5. Another configuration capable of excellently restoring the wideband sound source signal will be described. FIG. 7 is a configuration diagram of the wideband sound source estimating means 14 in the wideband speech restoring device according to the fifth embodiment of the present invention. In the figure, a novel part is 37 narrow band long period prediction analysis means, 38 narrow band long period delay, 39 narrow band long period prediction coefficient, 40 long period inverse filter, 41
Narrow band long period prediction residual signal, 42 wide band long period prediction residual estimation means, 43 zero padding means, 44 wide band long period prediction parameter (code) estimation means, 45 wide band long period delay, 46 wide band The long-period prediction coefficient, 47 long-period synthesis filter, and 48 wide-band long-period prediction residual signal. The overall configuration is the same as in FIG.

The operation of the embodiment of the present invention will be described below with reference to FIG. The narrowband sound source signal 6 is input to the sound source analysis means 21 in the wideband sound source estimation means 14. The narrow band long cycle prediction analysis means 37 in the sound source analysis means 21 performs long cycle prediction analysis on the narrow band sound source signal 6, and narrow band long cycle delay 38 and narrow band long cycle which are narrow band long cycle prediction codes. The prediction coefficient 39 is output. Note that the long-period prediction analysis is a method often used in the CELP-based coding method, and therefore its description is omitted. Sound source analysis means 21
The long period inverse filter 40 in the above inversely filters the narrow band source signal 6 using the narrow band long period delay 38 and the narrow band long period prediction coefficient 39, and obtains the obtained signal as the narrow band long period prediction residual signal 41. As a wide band long period prediction residual estimation means 42
Output to.

The zero padding means 43 in the wide band long period prediction residual estimation means 42 inserts zeros by M-1 samples between each sample value of the narrow band long period prediction residual signal 41 and obtains M times the obtained value. A signal of the number of samples is output as a wideband long-period prediction residual signal 48. Here, M is a value obtained by dividing the sampling frequency of the wideband audio signal to be restored by the sampling frequency of the narrowband audio signal, and the operation of inserting zero is the same as the zero padding means 15.

The wide band long cycle prediction parameter (code) estimating means 44 multiplies the narrow band long cycle delay 38 by M to output a wide band long cycle delay 45 which is one of the prediction codes, and the narrow band long cycle prediction coefficient. 39 is multiplied by g to output a wideband long-period prediction coefficient 46 which is another prediction code. When g is set to 1, the pitch periodicity of the finally obtained wideband source signal 16 becomes equal to that of the narrowband source signal 6, and as the value becomes smaller from 1, the pitch periodicity becomes weaker than that of the narrowband source signal 6. Go. Similar to the second embodiment, when the high frequency band is restored, setting g to a value smaller than 1 gives higher quality. Finally, the long period synthesis filter 47 includes a wide band long period delay 45.
The wideband long-period prediction residual signal 48 is subjected to long-period synthesis filtering using the wideband long-period prediction coefficient 46, and the obtained signal is output as the wideband source signal 16.

With this configuration, the characteristics of the strength and fluctuation of the pitch periodicity of the narrow band source signal are well expressed by the narrow band long period delay 38 and the narrow band long period prediction coefficient 39, and the wide band source As it is reflected in the signal,
There is an effect that it is possible to sufficiently estimate a sound source that changes in various ways, and to recover wide-band speech with good sound quality without pulse-like sound. Further, there is an effect that an appropriate sound source can be estimated regardless of the speaker. In the wideband sound source signal 16, since the fundamental frequency determined by the wideband long-period delay 45 and the frequencies of its harmonic components are correctly aligned at positions of integral multiples, the narrowband component and restoration in the finally restored wideband speech signal 20 are performed. The wideband components are well connected, and the effect is that high-quality wideband speech can be restored. Further, since the characteristic that the pitch periodicity becomes weaker as the frequency becomes higher can be introduced by the coefficient g, there is an effect that more natural sound quality can be obtained. In addition, since voiced / unvoiced judgment and pitch extraction are not required, and a sound source with an intermediate property can be expressed, there is no influence of voiced / unvoiced judgment error or pitch extraction error that tends to occur in a narrowband speech signal on which noise is superimposed, Even in the vicinity of the voiced and unvoiced boundaries, a good wideband sound source can be estimated, and there is an effect that a stable and natural wideband speech can be restored.

Example 6. FIG. 8 is a block diagram of the wideband long-period prediction residual estimation means 42 in the wideband speech reconstructing apparatus according to the sixth embodiment of the present invention. In the figure, the power calculation means 35 and the noise generation means 36 are the same as those in the third embodiment. Since other configurations are the same as those in FIGS. 1 and 7,
The description is omitted.

The operation of the embodiment of the present invention will be described below with reference to FIG. The narrow band long period prediction residual signal 41 is input to the power calculation unit 35 in the wide band long period prediction residual estimation unit 42. The power calculation means 35 calculates the power of the narrow band long cycle prediction residual signal 41 and outputs it. The noise generating means 36 generates and outputs a power-normalized white noise signal. Then, the broadband long-period prediction residual estimation means 4
In 2, the white noise signal is multiplied by the power output by the power calculating means 35, and the obtained signal is output as a wideband long-period prediction residual signal 48.

Similar to the description in the third embodiment, since the minute fluctuations in the pitch period and the strength of periodicity in the narrow band excitation signal 6 cannot be expressed by the narrow band long period delay 38 and the narrow band long period prediction coefficient 39. The error is included in the narrow band long cycle prediction residual signal 41. When the wide band long period prediction residual signal 48 is generated using the narrow band long period prediction residual signal 41 including the error component as in the fifth embodiment, unnecessary disturbance occurs in the wide band long period prediction residual signal 48. Sometimes,
In some cases, better restored sound may be obtained by generating white noise having the same power and using it as the wideband long-period prediction residual signal 48. Since the white noise having the same power as that of the narrow band long cycle prediction residual signal 41 is generated and configured as the wide band long cycle prediction residual signal 48 by configuring as in the sixth embodiment,
In addition to the effect of the fifth embodiment, there is an effect that a good restored sound with less disturbance due to fluctuations in the pitch period and the strength of periodicity can be obtained.

Further, since the zero-filling process produces a symmetrical spectrum around 4 KHz, the narrow band long period prediction residual signal 41 having no 0 Hz to 300 Hz and 3.4 KHz to 4.0 KHz components is generated. When you do it, 0H
z to 300 Hz, 3.4 KHz to 4.6, KHz
A signal without a component of 7.7 KHz to 8 KHz is obtained. On the other hand, in this configuration using white noise,
Since the wideband long-period prediction residual signal 48 having all the components from 0 Hz to 8 KHz is obtained, there is an effect that a good restored sound with no insufficient band can be obtained. Especially from 0Hz to 3
The effect is great when the 00 Hz restoration is performed.

Example 7. FIG. 9 is a configuration diagram of the wideband long-period prediction residual estimation means 42 in the wideband speech restoring device according to the seventh embodiment of the present invention. In the figure, 43 zero-filling means, 3
The power calculation means 5 and the noise generation means 36 are the same as those in the fifth and sixth embodiments. Other configurations are shown in FIG.
Since it is the same as that of FIG. 7, description thereof will be omitted.

The operation of the embodiment of the present invention will be described below with reference to FIG. The narrow band long cycle prediction residual signal 41 is input to the zero padding means 43 and the power calculation means 35 in the wide band long cycle prediction residual estimation means 42. The zero padding means 43 in the wide band long cycle prediction residual estimation means 42 inserts zeros by M-1 samples between each sample value of the narrow band long cycle prediction residual signal 41, and the obtained M times the number of samples. The signal of is output. Here, M is a value obtained by dividing the sampling frequency of the wideband audio signal to be restored by the sampling frequency of the narrowband audio signal, and the operation of inserting zero is the zero padding means 15.
Is the same as. The power calculation means 35 calculates the power of the narrow band long cycle prediction residual signal 41 and outputs it. The noise generating means 36 generates and outputs a power-normalized white noise signal. Then, a signal obtained by multiplying the signal output by the zero padding unit 43 by the gain gr1 and a signal obtained by multiplying the white noise signal output by the noise generating unit 36 by the power output by the power calculating unit 35 and further by the gain gr2 are added. And outputs it as a wideband long-period prediction residual signal 48.

The restored sounds according to the fifth and sixth embodiments are
If each has advantages and disadvantages, configure it like this,
By properly setting gr1 and gr2, there is an effect that a wideband voice having a quality higher than those of both can be restored. It also has the same effect as the fifth and sixth embodiments.

Example 8. FIG. 10 is a block diagram of the wideband sound source estimating means 14 in the wideband speech restoring device according to the eighth embodiment of the present invention. The new parts in the figure are 49 upsampling means and 50 zeroing means. The overall configuration is
The description is omitted because it is the same as in FIG.

The operation of the embodiment of the present invention will be described below with reference to FIG. The narrowband sound source signal 6 is input to the upsampling means 49. The upsampling means 49 upsamples the narrowband excitation signal 6 to M times and outputs the obtained signal to the excitation analysis means 21.
Narrow band long period prediction analysis means 37 in the sound source analysis means 21.
Performs a long cycle prediction analysis on the output signal of the upsampling means 49 and outputs a narrow band long cycle delay 38 and a narrow band long cycle prediction coefficient 39. The delay search range in the long period prediction analysis is M times that in the case of the fifth embodiment. The long period inverse filter 40 in the sound source analysis unit 21 inversely filters the output signal of the upsampling unit 49 using the narrow band long period delay 38 and the narrow band long period prediction coefficient 39, and narrows the obtained signal. The long-term prediction residual signal 41 is output to the wideband long-term prediction residual estimation means 42.

The zeroing means 50 in the wideband long-period prediction residual estimation means 42 leaves only every M samples of the narrowband long-period prediction residual signal 41 and sets the value of the remaining signals to zero. Then, the obtained signal is output as a wideband long-period prediction residual signal 48. The wideband long-cycle prediction parameter estimation means 44 outputs the narrowband long-cycle delay 38 as it is as the wideband long-cycle delay 45, and the narrowband long-cycle prediction coefficient 39.
Is multiplied by g and output as a wideband long-period prediction coefficient 46.
g is the same as in Example 5. Finally, the long-period synthesis filter 47 uses the wide-band long-period delay 45 and the wide-band long-period prediction coefficient 46 to generate the wide-band long-period prediction residual signal 4
8 is subjected to long-period synthesis filtering, and the obtained signal is output as a broadband source signal 16.

With this configuration, since a long period analysis can be performed on a signal having a high sampling frequency, it becomes possible to analyze a delay with higher accuracy, and the strength of pitch periodicity of a narrow band source signal and It becomes possible to reflect the characteristics related to fluctuations in the wideband sound source signal in more detail,
There is an effect that it is possible to sufficiently estimate a sound source that changes in various ways and to restore wideband speech with good sound quality. It also has the same effect as the fifth embodiment.

Example 9. FIG. 11 is a block diagram of a wideband voice restoration device according to a ninth embodiment of the present invention. In the figure, the new parts are 51 narrow band power calculation means, 52 narrow band excitation power, and 53 narrow band power-included spectrum codebook. Since the others are the same as those described above, only the differences in operation will be described.

The operation of the embodiment of the present invention will be described below with reference to FIG. The narrow band power calculating means 51 in the analyzing means 2 calculates the power included in the amplitude information of the narrow band excitation signal 6 and outputs it as the narrow band excitation power 52. In addition to this, the spectrum parameter 4 and the narrow band excitation signal 6 are also output. The vector quantization means 8 in the wideband spectrum estimation means 7 vector-quantizes the narrowband spectrum parameter 4 and the narrowband excitation power 52 at once by using the narrowband power-included spectrum codebook 53, and the obtained spectrum code is obtained. 10 is output to the inverse quantization means 11 in the wide band spectrum estimation means 7.

Here, the narrowband power-included spectrum codebook 53 is the same as that of the reference 1 using the pair of narrowband spectrum parameters and narrowband excitation power obtained by analyzing many narrowband speech signals as learning data. Create by method. As a distance measure at the time of learning the narrow band power-included spectrum codebook 53 and at the vector quantizing means 8, a value obtained by multiplying the Euclidean distance of the logarithm of the power by w and the Euclidean distance of the spectrum parameter can be used. .

The narrow band power calculating means 51 may calculate the power of the narrow band sound signal 1 instead of the narrow band sound source signal 6 and use it instead of the narrow band sound source power 52. In this case, the pair of narrowband spectrum parameters and the power of the narrowband speech signal is used as learning data, and the narrowband power-included spectrum codebook 53 is learned.

With such a configuration, in addition to the effect of the first embodiment, the information about the power is reflected in the estimation of the spectrum parameter of the wide band, and there is an effect that a better spectrum can be estimated more stably.

Example 10. FIG. 12 is a tenth embodiment of the present invention.
FIG. 3 is a configuration diagram of the broadband voice restoration device of FIG. The new part in the figure is included in 54 source normalization means, 55 narrowband normalized excitation signal, 56 wideband normalized excitation signal, 57 wideband power codebook, 58 wideband source power, and wideband spectrum estimation means. 59 wideband sound source power estimation means. Others are the same as those described above,
The description is omitted.

The operation of the embodiment of the present invention will be described below with reference to FIG. The sound source normalization means 54 in the analysis means 2 calculates the power included in the amplitude information of the narrow band sound source signal 6 and outputs it as the narrow band sound source power 52 to the wide band sound source power estimation means 59, and at the same time, the narrow band sound source signal 6
The signal obtained by normalizing the power of
Is output to the broadband sound source estimation means 14.

In practice, the vector quantizing means 8 in the wideband source power estimating means 59 in the wideband spectrum estimating means 7 uses the narrowband power-included spectrum codebook 53 to set the narrowband spectrum parameter 4 and the narrowband source. The power 52 is vector-quantized collectively, and the obtained spectrum code 10 is output to the inverse quantization means 11 in the wideband excitation power estimation means 59. The inverse quantization means 11 decodes the spectrum code 10 using the wideband power codebook 57,
The obtained broadband sound source power 58 is output.

The wideband sound source estimating means 14 estimates the wideband sound source normalized signal 56 using the narrow band sound source normalized signal 54. For the estimation in the wide band spectrum estimating means 7 and the wide band sound source estimating means 14, the same method as in the first to eighth embodiments can be used. Then, the wide band source signal 16 is generated by multiplying the wide band normalized source signal 56 by the wide band source power 58.

With this configuration, in addition to the effect of the first embodiment, the difference in the spectrum parameter can be reflected in the estimation of the wideband sound source power, so that the wideband speech having a more correct amplitude can be restored. effective.

Example 11. FIG. 13 shows the eleventh embodiment of the present invention.
FIG. 3 is a configuration diagram of the broadband voice restoration device of FIG. The novel part in the figure is 60 wideband power-included spectral codebooks. Others are the same as those in FIG. 11 and FIG. 12, and therefore only the operations that are slightly different will be described.

The operation of the embodiment of the present invention will be described below with reference to FIG. The dequantization means 11 in the wideband spectrum estimation means 7 decodes the spectrum code 10 using the wideband power-included spectrum codebook 60, and outputs the obtained wideband spectrum parameter 13 and wideband excitation power 58. Here, the wideband power-included spectrum codebook 60 is created in the same manner as in Reference 1 by using a pair of wideband spectrum parameters and wideband sound source power obtained by analyzing many wideband speech signals as learning data.
As the distance measure, the narrowband power-included spectrum codebook 53
Use the same one that was used to create the.

With this structure, the effects of the ninth and tenth embodiments can be combined.

Example 12. FIG. 14 shows a twelfth embodiment of the present invention.
FIG. 3 is a configuration diagram of the broadband voice restoration device of FIG. The new part in the figure is 61 post-filter means. Others are the same as those in the first to eleventh embodiments, and the description thereof will be omitted.

The operation of the embodiment of the present invention will be described below with reference to FIG. The post filter means 61 performs post filtering processing on the temporary wide band audio signal output from the synthesis filter 17, and outputs the obtained signal to the band filter 18. The band-pass filter 18 performs band-pass filter processing on the signal output by the post-filter unit 61, and extracts components other than the band in which the narrow-band speech component is present. The post filtering process is
A signal processing process that improves perceptual quality, which emphasizes pitch periodicity and spectral poles, emphasizes high frequencies to improve intelligibility, and is a band with a lot of distortion that occurs when passing through a transmission path. To reduce the distortion. As a pitch periodicity emphasizing process, a method of multiplying a provisional wideband speech signal which is a pitch period before by a coefficient smaller than 1 and adding it to the current provisional wideband speech signal is general.

As the pole emphasis processing, the wide band spectrum parameter 13 is modified to have a large gain in the frequency band near the pole frequency of the wide band spectrum parameter 13 and small in the frequency band other than the pole vicinity of the wide band spectrum parameter 13. Various methods of calculating the filter coefficient of a pole-zero type filter with gain have been proposed,
This can be realized by applying this filter to a temporary wide band audio signal. Further, since the distortion that occurs when passing through the transmission path is large in the frequency band having a small amplitude, that is, the frequency band other than the extreme vicinity, it is possible to suppress the band having the large distortion by this pole emphasis processing. As the high-frequency emphasis processing, a method called pre-emphasis, that is, a method of multiplying a temporary wideband speech signal one point before by a coefficient of 1 or less and subtracting from the current temporary wideband speech signal is general. Further, in FIG. 14, the post filter means 61 and the band-pass filter 18 may be at opposite positions, or the post-filter means 61 may be applied to the wide band audio signal 20.

With such a configuration, in addition to the effect of the first embodiment, when the quality of the restored wideband speech signal is insufficient, the pitch periodicity of the wideband speech signal or the pole of the spectrum is emphasized. There is an effect that the high range is emphasized to improve the clarity and a band in which a lot of distortion is generated when passing through the transmission path is suppressed to reduce the sense of distortion. Note that FIG.
A configuration in which the inverse filter 5 and the broadband sound source estimation means 14 are removed in is also possible. This configuration corresponds to the one in which the present invention is applied to Document 1, and has the same effect as described above.

Example 13 The wideband spectrum estimating means 7 in the first to twelfth embodiments may be configured to output the narrowband spectrum parameter 4 as it is as the wideband spectrum parameter 13.

FIG. 15 is an explanatory view for explaining the general relationship between the narrow band spectrum and the wide band spectrum in this case. When the spectrum envelope represented by the narrowband spectrum parameter 4 is as shown in FIG. 15A, if this spectrum envelope is used as it is as the wideband spectrum parameter 13, the width thereof is expanded, and the spectrum envelope represented by the wideband spectrum parameter 13 is shown in FIG. FIG. 15B is a diagram in which (a) is stretched M times in the frequency axis direction, and when M is 2. Therefore, when the narrow band spectral envelope of 2 KHz to 3.4 KHz is high, the high band of 3.4 KHz or higher to be restored is also high, and conversely, when the low band of 2 KHz to 3.4 KHz is low, the high band is also low. As a result, the rough slope of the narrow-band spectrum envelope is directly reflected in the high range.

With this configuration, in addition to the effect of the first embodiment, there is a rough effect that the wide band spectrum can be restored very easily. As compared with the first embodiment, a memory for accumulating the codebook is unnecessary, which has an effect of reducing the calculation amount.

Example 14 In the first to twelfth embodiments, the wide band spectrum estimating means 7 may output the narrow band spectrum parameter 4 from the lowest order to a predetermined order as the wide band spectrum parameter 13. However, as the narrow-band spectrum parameter 4 output by the spectrum analysis unit 3, even if the one obtained from the lowest order to a predetermined order such as a PARCOR coefficient or an autocorrelation coefficient is used as the wide-band spectrum parameter 13, the composition is always stable. It is limited to when it is a parameter.

FIG. 16 is an explanatory view for explaining the general relationship between the narrow band spectrum and the wide band spectrum in this case. If the spectral envelope represented by the narrowband spectral parameter 4 is FIG. 16 (a), the the use of from this lowest order to a predetermined degree as a broadband spectral parameter 13, the spectral envelope represented by the wideband spectral parameters 13 Figure 16 (a) further becomes a form to smooth the electrode structure is stretched to M times the frequency axis direction, M is shown in FIG. 16 (b) when two. As a result, the rough slope of the narrow-band spectrum envelope is directly reflected in the high frequency range, and a strong pole that does not exist is generated in the high frequency range, and it is possible to suppress the generation of an unnatural restoration sound.

With such a configuration, in addition to the effect of the thirteenth embodiment, a strong pole that does not exist is generated in the high range, which is rare in the case of the thirteenth embodiment, and an unnatural restoration sound is generated. There is an effect that can be suppressed.

Example 15. FIG. 17 shows a fifteenth embodiment of the present invention.
It is a block diagram of the wideband spectrum estimation means 7 of the wideband speech restoration device of FIG. The new parts in the figure are 62 spectral parameter converting means, 63 order reducing means, and 64 spectral parameter inverse converting means. Others are the same as those in the first to twelfth embodiments, and the description thereof will be omitted.

The operation of the embodiment of the present invention will be described below with reference to FIG. The spectrum parameter conversion means 62 in the wide band spectrum estimation means 7 converts the narrow band spectrum parameter 4 into a parameter whose composition is always stable when the lowest order to a predetermined order such as PARCOR coefficient and autocorrelation coefficient are taken out. To do. Order reduction means 6
3 outputs the parameters extracted from the lowest order to the predetermined order of the parameters output by the spectrum parameter converting means 62 to the spectrum parameter inverse converting means 64. The spectrum parameter inverse conversion means 64 is the order reduction means 63.
Narrow band spectrum parameter 4
It returns to the same region as and is output as the broadband spectrum parameter 13.

With such a configuration, even if the narrow band spectrum parameter 4 is a parameter in which the synthesis becomes unstable when the lowest order to the predetermined order is taken out,
The same effects as in Example 14 can be obtained.

Example 16. Example 14 and Example 15
In, the strong poles were suppressed by reducing the order, but a method that gives a similar effect, such as using an autocorrelation coefficient as a spectral parameter and applying a lag window to it, can be used. With this configuration, the same effect as that of the fourteenth embodiment can be obtained by another means. The wideband spectrum estimating means 7 of the above thirteenth to sixteenth embodiments can also be applied to the conventional configuration of Document 1 or the like. For example, the entire configuration in the case of applying to Document 1 is the one in which the inverse filter 5, the wideband sound source estimation means 14, and the post filter means 61 are removed from FIG. In the case of such a configuration, the thirteenth embodiment
The effects newly generated in Nos. 16 to 16 can be added to the conventional technique.

Example 17 In the following embodiments, an example will be described in which the present invention is applied to a device that restores wideband speech based on coded information such as transmission. FIG. 18 is a block diagram of a wideband speech restoring device according to a seventeenth embodiment of the present invention. In the figure, 101 is a narrow band speech code, 102 is a separating means, and 1
Reference numeral 03 is a narrow band spectrum code, 104 is a narrow band excitation code, 105 is a wide band spectrum decoding means, 106 is a wide band excitation decoding means, 107 is a narrow band spectrum decoding means,
Reference numeral 108 is a narrow band sound source decoding means, and 109 is a narrow band speech decoding means. Others are the same as those in the first to sixteenth embodiments, and description thereof will be omitted. Also in this embodiment, a good wideband sound source signal is obtained without reanalysis.

The operation of the embodiment of the present invention will be described below with reference to FIG. First, the narrow band speech code 101
Is input to the separation means 102 and the narrow band speech decoding means 109. This narrow band speech code 101 is, for example, 8 KHz.
The narrowband voice signal, which is sampled in the above and is limited to the telephone band of 300 Hz to 3.4 KHz, is separately encoded, and is input from the storage medium or the communication path. In the separating means 102, the narrowband speech code 10
1 is the narrowband spectrum code 103 and the narrowband excitation code 10
4, and outputs narrowband spectrum code 103 to wideband spectrum decoding means 105 and narrowband excitation code 104 to wideband excitation decoding means 106.

The narrowband spectrum decoding means 107 in the wideband spectrum decoding means 105 decodes the narrowband spectrum code 103 and outputs the obtained narrowband spectrum parameter 4. The narrow band spectrum decoding means 10
In No. 7, it suffices to perform the reverse process of the encoding process of the narrowband spectrum parameter used when the narrowband speech code 101 was encoded. Then, the broadband spectrum decoding means 1
The wideband spectrum estimating means 7 in 05 estimates the wideband spectrum parameter 13 using the narrowband spectrum parameter 4. As the wideband spectrum estimating means 7, the method described in the above-described embodiments can be used.

The narrow band excitation decoding means 108 in the wide band excitation decoding means 106 decodes the narrow band excitation code 104 and outputs the obtained narrow band excitation signal 6. Then, the wideband sound source estimation means 14 in the wideband sound source decoding means 106.
However, using the narrow band source signal 6
To estimate. The broadband sound source estimating means 14 may be a zero padding means or the like. Narrowband sound source decoding means 108
Then, the reverse process of the encoding process of the narrowband excitation signal used when the narrowband speech code 101 was encoded may be performed.

The synthesis filter 17 performs synthesis filter processing on the wideband sound source signal 16 using the wideband spectrum parameter 13 to generate a temporary wideband speech signal. The band-pass filter 18 performs a band-pass filter process on the temporary wide-band speech signal to extract a component other than the band having the narrow-band speech component. When the band of the wideband audio signal is 0 Hz to 7.3 KHz, 0 Hz to 300 Hz and 3.4
A 7.3 KHz component is extracted from KHz.

On the other hand, the narrowband speech decoding means 109 decodes the input narrowband speech code 101 and outputs the obtained narrowband speech signal 1 to the upsampling means 19.
This decoding process may be the reverse of the encoding process used when the narrowband speech code 101 was encoded. Next, the upsampling means 19 outputs the narrowband audio signal 1
Is upsampled M times. The signal generated by upsampling has the same sampling frequency as that of the wideband speech signal 20 and has the same narrowband component as that of the narrowband speech signal 1. Then, the output of the bandpass filter 18 and the output of the upsampling means 19 are added to generate a wideband audio signal 20.

With such a configuration, when a narrow band speech code is received from a storage medium or a communication path, it is not necessary to reanalyze the narrow band speech, so that there is an effect that restoration can be performed with a small processing amount. Further, since distortion due to interpolation at the time of synthesis or windowing at the time of analysis is not superimposed, there is an effect that a better quality wideband speech can be restored. It also has the same effect as the first embodiment.

The narrowband speech decoding means 109 may be configured to input the narrowband spectrum parameter 4 and the narrowband excitation signal 6 and synthesize the narrowband speech signal 1, or conversely, the narrowband speech decoding means 109. A configuration is also possible in which the narrowband spectrum parameter 4 and the narrowband excitation signal 6 calculated as the intermediate parameters of the decoding process in the above are input to the wideband spectrum decoding means 105 and the wideband excitation decoding means 106.
In this case, it is possible to omit redundant processing, and there is an effect that wideband speech can be restored with a smaller processing amount. Also,
When the pitch period code and the power code can be separated from the narrowband speech code 101, the pitch period and the power information are decoded from these codes, and the wideband spectrum parameter 13 and the pitch period and the power information are used to read the literature. 1
It is also possible to have a configuration in which a temporary wideband synthesized sound is generated by the same method as.

Example 18. FIG. 19 shows Example 18 of the present invention.
FIG. 3 is a configuration diagram of wideband sound source decoding means 106 of the wideband speech restoration device of FIG. In the figure, the new parts are 110 narrow band pitch code, 111 narrow band power code, 112 wide band pitch decoding means, 113 wide band pitch period, 114
Broadband power decoding means of 115, broadband power of 115, 11
6 sound source generation means. Others are the same as those in the seventeenth embodiment, and description thereof will be omitted.

In this embodiment, the separation means 102 allows the narrow band pitch code 110 and the narrow band power code 111 to be easily obtained.
It is limited to the case where the narrowband speech code 101 that can be separated is input. In this case, the configuration of FIG. 19 is meaningful. The operation of the embodiment of the present invention will be described below with reference to FIG. As narrowband excitation code 104, narrowband pitch code 110 and narrowband power code 111 are input to wideband excitation decoding means 106.

Wideband pitch decoding means 112 in wideband excitation decoding means 106 estimates wideband pitch period 113 using narrowband pitch code 110. As an estimation method, the narrow band pitch period may be decoded from the narrow band pitch code 110 and its value may be multiplied by M, but the result is held as a table and the table component corresponding to the narrow band pitch code 110 is stored. You can ask for it by reading. Next, the wideband power decoding means 114 in the wideband excitation decoding means 106.
Is the wideband power 11 using the narrowband power code 111.
Estimate 5. As an estimation method, the narrow band power may be decoded from the narrow band power code 111 and the value thereof may be multiplied by g, but the result is held as a table and the table component corresponding to the narrow band power code 111 is stored. You may ask for it by reading it.

The sound source generating means 116 outputs a signal in which fixed sound sources are lined up with the wide band pitch period 113 as a repeating period, and finally the output signal of the sound source generating means 116 is multiplied by the wide band power 115 to obtain a wide band sound source signal. 1
Output as 6.

With this configuration, in addition to the effect of the seventeenth embodiment, the wideband source signal 16 is directly generated without decoding the narrowband source signal, so that the restoration can be performed with a small processing amount. There is.

Example 19 20 shows a nineteenth embodiment of the present invention.
FIG. 3 is a configuration diagram of wideband sound source decoding means 106 of the wideband speech restoration device of FIG. In the figure, the new parts are 117 narrowband adaptive excitation code, 118 narrowband driving excitation code, 119 wideband adaptive excitation decoding means, 120 wideband driving excitation decoding means, 121 narrowband adaptive excitation decoding means, and 122 It is a narrow band drive excitation decoding means. Others are the same as those described above, and description thereof will be omitted.

In this embodiment, the separating means 102 can easily convert the narrowband speech code inputted from the narrowband speech code 11 into the narrowband adaptive excitation code 11.
This is limited to the case where the narrowband speech code 101 that can separate the 7 and the narrowband driving excitation code 118 is input. In this case, the configuration of FIG. 20 is meaningful. The operation of the embodiment of the present invention will be described below with reference to FIG. As narrowband excitation code 104, narrowband adaptive excitation code 117 and narrowband driving excitation code 118 are input to wideband excitation decoding means 106.

The narrowband adaptive excitation decoding means 121 in the wideband adaptive excitation decoding means 119 uses the narrowband adaptive excitation code 1
Decode 17 and output the obtained narrowband adaptive lag length 25 and narrowband adaptive gain 26. The wideband adaptive excitation estimation means 30 in the wideband adaptive excitation decoding means 119 generates and outputs a wideband adaptive excitation signal 32 from the narrowband adaptive lag length 25 and the narrowband adaptive gain 26. The operation of the wideband adaptive sound source estimation means 30 is the same as in the second embodiment.

The narrow band drive excitation decoding means 122 in the wide band drive excitation decoding means 120 is the narrow band drive excitation code 1
Decode 18 and output the obtained narrowband drive source signal 24. Wideband drive excitation estimation means 27 in wideband drive excitation decoding means 120 estimates wideband drive excitation signal 29 from this narrowband drive excitation signal 24 and outputs it. The operation of the wideband driving sound source estimation means 27 is the same as in the second to fourth embodiments.
Is the same as. Finally, the wideband adaptive excitation signal 32 and the wideband driving excitation signal 29 are added and output as the wideband excitation signal 16.

With this configuration, in addition to the effects of the second to fourth embodiments and the seventeenth embodiment, the wideband excitation signal 16 is directly generated without decoding the narrowband excitation signal. The effect is that restoration can be performed with a small amount of processing. Furthermore, since the frequencies of the fundamental frequency and its harmonic components are correctly aligned at positions of integral multiples, the connection between the narrowband component and the restored wideband component in the finally restored wideband speech signal is good, and high-quality wideband speech is produced. There is an effect that can be restored. In addition, since voiced unvoiced information and pitch period information are not used, it is possible to represent a sound source with an intermediate property, so that the effects of voiced unvoiced decision errors and pitch extraction errors that are likely to occur in narrowband speech signals with superimposed noise are reduced. In addition, a good wideband sound source can be estimated even near the voiced / unvoiced boundary, and stable and natural wideband speech can be restored.

Example 20. 21 shows Embodiment 20 of the present invention.
FIG. 3 is a configuration diagram of wideband sound source decoding means 106 of the wideband speech restoration device of FIG. In the figure, new parts are 123 narrow band long cycle prediction code, 124 wide band long cycle prediction parameter (code) decoding means, 125 narrow band long cycle prediction parameter (code) decoding means, 126 narrow band long cycle prediction code. Residual code, 127 wideband long period prediction residual decoding means, and 128 narrow band long period prediction residual decoding means. Others are the same as those described above, and description thereof will be omitted.

In this embodiment, the narrow band long period predictive code 1 can be easily converted from the narrow band speech code input by the separating means 102.
It is limited to the case where the narrow band speech code 101 that can separate the H.23 and the narrow band long cycle prediction residual code 126 is input. In this case, the configuration of FIG. 21 is meaningful. The operation of the embodiment of the present invention will be described below with reference to FIG. As the narrow band excitation code 104, the narrow band long period prediction code 123 and the narrow band long period prediction residual code 126 are input to the wide band excitation decoding means 106.

Narrowband long cycle prediction parameter decoding means 1 in wide band long cycle prediction parameter (code) decoding means 124
25 decodes the narrow band long cycle prediction code 123,
Narrow band long period delay 38 which is one of the obtained prediction codes
And the narrow band long cycle prediction coefficient 39 which is another prediction code is output. The wide band long period prediction parameter estimation unit 44 in the wide band long period prediction parameter decoding unit 124 uses the narrow band long period delay 38 and the narrow band long period prediction coefficient 39 to
A wideband long-period delay 45, which is one of long-period prediction codes,
A wideband long-period prediction coefficient 46, which is one of the other long-period prediction codes, is estimated and output. The operation of the wideband long-period prediction parameter estimation means 44 is the same as that of the fifth embodiment.

The narrow band long cycle prediction residual decoding means 128 in the wide band long cycle prediction residual decoding means 127 decodes the narrow band long cycle prediction residual code 126 and obtains the narrow band long cycle prediction residual. The difference signal 41 is output. Wideband long-period prediction residual estimation means 4 in wideband long-period prediction residual decoding means 127.
2 estimates the wideband long-period prediction residual signal 48 from this narrow-band long-period prediction residual signal 41 and outputs it. The operation of the wideband long-period prediction residual estimation means 42 is the same as in the fifth to seventh embodiments. Finally, the long cycle synthesis filter 47
Is a wide band long period delay 45 and a wide band long period prediction coefficient 46.
Is used to perform long-period synthesis filtering on the wideband long-period prediction residual signal 48, and the obtained signal is output as the wideband excitation signal 16.

With this configuration, in addition to the effects of the fifth to seventh embodiments and the seventeenth embodiment, the wideband sound source signal 16 is directly generated without decoding the narrowband sound source signal. The effect is that restoration can be performed with a small amount of processing.

Example 21. Examples 17 to 20
In, the wideband spectrum parameter 13 is estimated after decoding the narrowband spectrum parameter 4 from the narrowband spectrum code 103. However, by referring to the wideband spectrum codebook by the narrowband spectrum code 103, the wideband spectrum parameter 13 can be directly determined. A configuration for calculating is also possible. With this configuration, in addition to the effects of the seventeenth to twentieth embodiments, there is an effect that restoration can be performed with a smaller processing amount.

Example 22. FIG. 22 is a block diagram of a wideband voice restoration device according to an embodiment of the present invention. The new parts in the figure are the narrowband power decoding means 129 and the wideband normalized excitation decoding means 130. The wideband spectrum estimating means 7 is the same as that of the eleventh embodiment, and the other points are the same as those described above, and the description thereof is omitted.

The operation of the embodiment of the present invention will be described below with reference to FIG. Narrowband power decoding means 129
Outputs a narrowband excitation power 52 obtained by decoding the power-related portion from the narrowband amplitude information included in the narrowband excitation code 104 to the wideband spectrum estimating means 7. The wideband spectrum estimating means 7 uses the narrowband spectrum parameter 4 and the narrowband sound source power 52,
Wideband spectrum parameter 13 and wideband sound source power 5
Estimate 8. The wideband normalized excitation decoding means 130 estimates a power source-normalized wideband excitation signal using a portion other than the portion related to the narrowband power included in the narrowband excitation code 104, and as a wideband normalized excitation signal 56. Output. As the processing in the wide band normalized sound source decoding means 130, the same processing as in the eighteenth to twentieth embodiments can be used. Then, this wideband normalized sound source signal 5
6 is multiplied by the wide band sound source power 58 to generate a wide band sound source signal 16.

With this structure, the effects of the eleventh embodiment and the eighteenth to twentieth embodiments can be combined. It is also possible to adopt a configuration in which the wide band spectrum estimating means 7 estimates only one of the wide band spectrum parameter 13 and the wide band sound source power 58 as in the ninth and tenth embodiments.

Example 23. Examples 17 to 22
In the above, the post filter means 61 may be inserted between the synthesis filter 17 and the band pass filter 18.
Further, the post filter means 61 and the band-pass filter 18 may be arranged in opposite positions, or the post-filter means 61 may be applied to the wide band audio signal 20. With this configuration, the narrow band speech decoding means 1
When the post-filter processing is performed within 09, the continuity between the narrow band portion and the restored band can be improved. Further, the effects of the twelfth embodiment and the seventeenth to twenty-second embodiments can be combined.

Example 24. A configuration in which the wide band excitation decoding means 106 is removed from FIG. 18 and a post filter means 61 is inserted between the synthesis filter 17 and the band filter 18 is also possible. Further, the post filter means 61 and the band-pass filter 18 may be arranged in opposite positions, or the post-filter means 61 may be applied to the wide band audio signal 20.
This configuration is described in Reference 1 in Example 17 of the present invention, Example 12 of the present invention.
When the post-filter processing is performed in the narrow band speech decoding means 109, the continuity between the narrow band portion and the restored band can be improved.

[0140]

As described above, according to the present invention,
Wideband source signal is estimated using narrowband source signal,
Since this is used to synthesize a wideband voice signal, the characteristics of a narrowband source signal can be satisfactorily given to the wideband source signal, and there is little dependence on the speaker and stable wideband voice with natural sound quality is obtained. There is an effect that can be restored.

Further, since the zero padding means for inserting a predetermined number of zeros between each sample of the narrow band excitation signal is used as the wide band sound source estimation means, there is no need for voiced unvoiced judgment or pitch extraction, and voiced unvoiced judgment error. It is possible to estimate a good wideband sound source that is not affected by the pitch extraction error and to restore a stable and natural wideband sound.

Further, as the wideband excitation estimating means, the wideband adaptive excitation signal and the wideband excitation signal are estimated using the narrowband adaptive excitation code and the narrowband driving excitation signal, and the wideband excitation signal is generated from this. Therefore, the characteristics relating to the strength and fluctuation of the pitch periodicity of the narrow band sound source signal are well reflected in the wide band sound source signal, and there is an effect that a wide band sound of good sound quality can be restored without any pulse-like sound. Furthermore, since the fundamental frequency and the frequencies of its harmonic components are correctly aligned at positions of integral multiples, the connection between the narrowband component and the restored wideband component in a wideband speech signal is good, and the practical property of pitch periodicity can also be restored. , Has the effect of restoring high-quality wideband speech.

As the wideband sound source estimating means, the wideband long-period prediction code and the wideband long-period residual signal are estimated using the narrowband long-period prediction code and the narrowband long-period residual signal, and these are used. Since the wideband source signal is synthesized, the characteristics of the pitch periodicity and the variation of the narrowband source signal are well reflected in the wideband source signal,
There is an effect that it is possible to restore wide-band voice with good sound quality without pulse-like sound. Furthermore, since the fundamental frequency and the frequencies of its harmonic components are correctly aligned at positions of integral multiples, the connection between the narrowband component and the restored wideband component in the finally restored wideband audio signal is good, and the actual pitch periodicity The characteristics can also be incorporated, and there is an effect that high-quality wideband speech can be restored.

Further, since the wideband spectrum parameter and / or the wideband amplitude information are estimated by using the narrowband spectrum parameter and the narrowband amplitude information, the narrowband amplitude information is used for estimating the wideband spectrum parameter. This has the effect that a good spectrum can be estimated more stably and a wideband speech having a more correct amplitude can be restored.

Furthermore, since the wide band speech signal estimated by using the narrow band speech signal is post-filtered, the pitch periodicity is emphasized and the spectral envelope is enhanced when the quality of the restored wide band speech signal is insufficient. There is an effect that the sound quality can be improved by emphasizing the poles of.

Furthermore, since the wideband speech signal is synthesized by expanding the narrowband spectrum parameter and using it as the wideband spectrum parameter, there is an effect that a rough wideband spectrum can be reconstructed very easily. In addition, a memory for accumulating the codebook is unnecessary, which has the effect of reducing the amount of calculation.

Furthermore, since the wideband spectrum parameter is obtained by inversely converting the narrowband spectrum parameter up to a predetermined order into a spectrum parameter, there is an effect that a rough wideband spectrum can be reconstructed very easily. . In addition, a memory for accumulating the codebook is unnecessary, which has the effect of reducing the amount of calculation.

Further, according to the present invention, the narrowband synthesized speech is generated using the narrowband speech code and the wideband speech signal is estimated, and the wideband speech is converted into an upsampled signal of the narrowband synthesized speech or the narrowband synthesized speech. Since the components of the band other than the narrowband synthesized sound of the signal are extracted and added, wideband speech can be restored even from encoded narrowband speech, and the decoded narrowband speech is not re-analyzed, so the processing amount is small. There is an effect that can be restored with.

Alternatively, the wideband speech signal is synthesized by using the wideband spectrum parameter estimated by using the narrowband spectrum code and the wideband excitation signal estimated by using the narrowband excitation code. There is an effect that the voice can be restored with a small processing amount without having to re-analyze the voice. Further, since distortion due to interpolation at the time of synthesis or windowing at the time of analysis is not superimposed, there is an effect that a better quality wideband speech can be restored.

Further, as the wide band excitation decoding means, the zero padding means for inserting a predetermined number of zeros between the samples of the narrow band excitation decoded by using the narrow band excitation code is used. It also has the effect that a sound source of a proper nature can be restored well, and a wideband voice with stable and natural sound quality can be restored.

Further, as the wideband excitation decoding means, the wideband adaptive excitation signal estimated using the narrowband excitation code and the wideband driving excitation signal are estimated and added to obtain the wideband excitation signal. A wideband excitation signal is directly generated without decoding the signal, and there is an effect that it can be restored with a small amount of processing. In addition, since the characteristics of the pitch periodicity included in the narrowband excitation code and the fluctuation thereof are well reflected in the wideband excitation signal, there is an effect that a wideband speech with good sound quality can be restored.

As the wide band excitation decoding means, the wide band long period prediction code estimated using the narrow band excitation code and the wide band long period residual signal are estimated, and these are used to synthesize the wide band excitation signal. Therefore, the wideband source signal is directly generated without decoding the narrowband source signal, and there is an effect that the restoration can be performed with a small processing amount. Further, since the characteristics of the pitch periodicity included in the narrowband excitation code and the fluctuations are well reflected in the wideband excitation signal,
There is an effect that it is possible to restore wideband sound with good sound quality.

Since the wideband spectrum parameter and / or the wideband amplitude information are estimated using the narrowband spectrum code and the narrowband amplitude information, the narrowband amplitude information is used for estimating the wideband spectrum parameter. Is reflected, the better spectrum can be estimated more stably, and the difference in the narrowband spectrum parameters can be reflected in the estimation of the wideband amplitude information, so that there is an effect that the wideband speech having a more correct amplitude can be restored.

Furthermore, since the post-filtering is performed on the wide-band speech signal estimated using the narrow-band speech code, when the post-filtering process is applied to the narrow-band synthesized sound, it is restored with the narrow-band portion. This has the effect of improving the continuity of the selected band. Further, when the quality of the restored wideband audio signal is insufficient, there is an effect that the sound quality can be improved by emphasizing the pitch periodicity and the pole of the spectrum envelope.

[Brief description of drawings]

FIG. 1 is a configuration diagram of a wideband speech restoring device according to a first embodiment of the present invention.

FIG. 2 is an explanatory diagram illustrating a process of zero padding means according to the first embodiment of the present invention.

FIG. 3 is a configuration diagram of wideband sound source estimating means in a wideband speech restoring device according to a second embodiment of the present invention.

FIG. 4 is an explanatory diagram illustrating an example of an adaptive sound source signal according to the second embodiment of the present invention.

FIG. 5 is a configuration diagram of a wideband driving sound source estimating means in a wideband speech restoring device according to a third embodiment of the present invention.

FIG. 6 is a configuration diagram of wideband driving sound source estimation means in a wideband speech restoration device according to a fourth embodiment of the present invention.

FIG. 7 is a configuration diagram of wideband sound source estimating means in a wideband speech restoring device according to a fifth embodiment of the present invention.

FIG. 8 is a configuration diagram of a wideband driving sound source estimating means in a wideband speech restoring device according to a sixth embodiment of the present invention.

FIG. 9 is a configuration diagram of wideband driving sound source estimation means in a wideband speech restoration device according to a seventh embodiment of the present invention.

FIG. 10 is a configuration diagram of wideband sound source estimating means in a wideband speech restoring device according to an eighth embodiment of the present invention.

FIG. 11 is a configuration diagram of a wideband speech restoring device according to a ninth embodiment of the present invention.

FIG. 12 is a configuration diagram of a wideband speech restoring device according to a tenth embodiment of the present invention.

FIG. 13 is a configuration diagram of a wideband speech restoring device according to an eleventh embodiment of the present invention.

FIG. 14 is a configuration diagram of a wideband speech restoring device according to a twelfth embodiment of the present invention.

FIG. 15 is an explanatory diagram illustrating a relationship between outlines of a narrow band spectrum and a wide band spectrum in Embodiment 13 of the present invention.

FIG. 16 is an explanatory diagram illustrating a relationship between outline shapes of a narrow band spectrum and a wide band spectrum in Embodiment 14 of the present invention.

FIG. 17 is a configuration diagram of wideband spectrum estimating means in a wideband speech restoring device according to a fifteenth embodiment of the present invention.

FIG. 18 is a configuration diagram of a wideband speech restoring device according to a seventeenth embodiment of the present invention.

FIG. 19 is a configuration diagram of a wideband sound source decoding means in a wideband speech restoring device according to an eighteenth embodiment of the present invention.

FIG. 20 is a configuration diagram of wideband speech decoding means in a wideband speech restoring device according to a nineteenth embodiment of the present invention.

FIG. 21 is a block diagram of a wideband speech decoding means in a wideband speech restoring device according to a twentieth embodiment of the present invention.

FIG. 22 is a configuration diagram of a wideband speech restoration device according to example 22 of the present invention.

[Explanation of symbols]

1 narrow band speech signal, 2 analysis means, 3 spectrum analysis means, 4 narrow band spectrum parameters, 5 inverse filter, 6 narrow band excitation signal, 7 wide band spectrum estimation means, 8 vector quantization means, 9 narrow band spectrum codebook, 10 spectrum codes, 11 inverse quantization means, 1
2 wideband spectrum codebook, 13 wideband spectrum parameters, 14 wideband sound source estimation means, 15 zero padding means, 16 wideband sound source signal, 17 synthesis filter, 18
Bandpass filter, 19 upsampling means, 20
Wideband speech signal, 21 Excitation analysis means, 22 Narrowband adaptive codebook, 23 Distortion minimizing means, 24 Narrowband driving excitation signal, 25 Narrowband adaptive lag length, 26 Narrowband adaptive gain, 27 Wideband driving excitation estimating means, 28 Zero padding means, 29 Wideband drive excitation signal, 30 Wideband adaptive excitation estimation means, 31 Wideband adaptive excitation codebook, 32 Wideband adaptive excitation signal, 33 Wideband adaptive lag length, 34 Wideband adaptive gain, 35 Power calculation means, 36 Noise generation means , 37 narrow band long period prediction analysis means, 38 narrow band long period delay, 39 narrow band long period prediction coefficient, 40 long period inverse filter, 41 narrow band long period prediction residual signal, 42
Wide band long period prediction residual estimation means, 43 zero padding means, 44
Wide-band long-period prediction parameter estimation means, 45 wide-band long-period delay, 46 wide-band long-period prediction coefficient, 47 long-period synthesis filter, 48 wide-band long-period prediction residual signal, 4
9 upsampling means, 50 zeroing means, 51
Narrow band power calculation means, 52 narrow band sound source power, 53
Narrowband power-included spectrum code, 54 Excitation normalization means, 55 Narrowband normalized excitation signal, 56 Wideband normalized excitation signal, 57 Wideband power codebook, 58 Wideband excitation power, 59 Wideband excitation power estimation means, 60
Wideband power-included spectral codebook, 61 post-filter means, 62 spectral parameter conversion means, 63
Order reducing means, 64 spectrum parameter inverse transforming means, 101 narrow band speech code, 102 separating means, 10
3 narrow band spectrum code, 104 narrow band excitation code,
105 Wideband spectrum decoding means, 106 Wideband excitation decoding means, 107 Narrowband spectrum decoding means, 10
8 Narrowband Excitation Decoding Means, 109 Narrowband Speech Decoding Means, 110 Narrowband Pitch Codes, 111 Narrowband Power Codes, 112 Wideband Pitch Decoding Means, 113 Wideband Pitch Cycles, 114 Wideband Power Decoding Means, 115
Wideband power decoding means, 116 sound source generating means, 117
Narrowband adaptive excitation code, 118 narrowband drive excitation code,
119 Wideband adaptive excitation decoding means, 120 Wideband driving excitation decoding means, 121 Narrowband adaptive excitation decoding means, 12
2 narrow band driving excitation decoding means, 123 narrow band long cycle prediction code, 124 wide band long cycle prediction parameter decoding means, 125 narrow band long cycle prediction parameter decoding means, 1
26 narrow band long period prediction residual code, 127 wide band long period prediction residual decoding means, 128 narrow band long period prediction residual decoding means, 129 narrow band power decoding means, 130 wide band normalized excitation decoding means.

─────────────────────────────────────────────────── ─── Continued Front Page (58) Fields surveyed (Int.Cl. ⁷ , DB name) G10L 19/00 G10L 21/04

Claims (30)

(57) [Claims] 1. A analyzing means for analyzing a narrow band speech signal to obtain a narrow band spectrum parameter and a narrow band excitation signal, a spectrum estimating means for estimating a wide band spectrum parameter using the narrow band spectrum parameter, and the narrow band. A wideband sound source estimating means for estimating a wideband sound source signal using a sound source signal, and a synthesizing means for generating a wideband speech signal from the estimated wideband spectrum parameter and the wideband sound source signal are provided. Zero padding means for inserting and outputting a predetermined zero value in each sampling interval of the band source signal, noise generating means for generating a noise signal, signal output by the zero padding means and noise generating means A wideband speech restoration apparatus comprising: an addition unit that adds a noise signal. 2. An analyzing means for analyzing a narrow band speech signal to obtain a narrow band spectrum parameter and a narrow band source signal, a spectrum estimating means for estimating a wide band spectrum parameter using the narrow band spectrum parameter, and the narrow band. A wideband sound source estimating means for estimating a wideband sound source signal using a sound source signal, and a synthesizing means for generating a wideband speech signal from the estimated wideband spectrum parameter and the wideband sound source signal are provided. Sound source analysis means for analyzing a narrow band adaptive excitation signal to obtain a narrow band adaptive excitation code and a narrow band driving excitation signal; adaptive sound source estimation means for estimating a wide band adaptive excitation signal using the narrow band adaptive excitation code; A driving sound source estimation means for estimating a wide band driving sound source signal using the driving sound source signal; Wideband audio decompressor, characterized in that an adding means for generating a wideband excitation signal from the source signal and the wideband excitation signal. 3. An analyzing means for analyzing a narrow band speech signal to obtain a narrow band spectrum parameter and a narrow band excitation signal, a spectrum estimating means for estimating a wide band spectrum parameter using the narrow band spectrum parameter, and the narrow band. A wideband sound source estimating means for estimating a wideband sound source signal using a sound source signal, and a synthesizing means for generating a wideband speech signal from the estimated wideband spectrum parameter and the wideband sound source signal are provided. Sound source analysis means for analyzing a band excitation signal to obtain a narrow band long cycle prediction code and a narrow band long cycle prediction residual signal, and estimating a wide band long cycle prediction residual signal using the narrow band long cycle prediction residual signal And long-range prediction residual estimation means for estimating a wide-band long-period prediction code using the narrowband long-period prediction code A wide band speech reconstructing apparatus comprising: a code estimating unit; and a long period synthesizing unit that synthesizes a wide band excitation signal from the estimated wide band long period prediction residual signal and the wide band long period prediction code. 4. Analyzing means for analyzing a narrow band speech signal to obtain narrow band spectrum parameters, narrow band excitation signals and narrow band amplitude information of the narrow band excitation signals, and the narrow band spectrum parameters and the narrow band amplitude information. A spectrum estimating means for estimating a wideband spectrum parameter using the same, a wideband source estimating means for estimating a wideband source signal using the narrowband source signal and the narrowband amplitude information, the estimated wideband spectrum parameter, and a wideband source A wideband speech decompression device comprising a synthesizing means for generating a wideband speech signal from a signal. 5. Analyzing means for analyzing a narrow band speech signal to obtain narrow band spectrum parameters, narrow band excitation signals and narrow band amplitude information of the narrow band excitation signals, and the narrow band spectrum parameters and the narrow band amplitude information. A spectrum estimation means for estimating at least a wideband spectrum parameter and wideband amplitude information using the same, a wideband excitation estimation means for estimating a wideband excitation signal using the narrowband excitation signal and the wideband amplitude information, and the estimated wideband spectrum parameter And a wideband speech restoration device comprising a synthesizing means for generating a wideband speech signal from the wideband sound source signal. 6. An analyzing means for analyzing a narrow band speech signal to obtain a narrow band spectrum parameter, a spectrum estimating means for outputting the wide band spectrum parameter using the narrow band spectrum parameter as it is as the wide band spectrum parameter, and the above-mentioned output. A wideband speech restoration apparatus comprising a synthesizing means for generating a wideband speech signal from a wideband spectrum parameter. 7. An analyzing means for analyzing a narrow band speech signal to obtain a narrow band spectrum parameter, a modification for suppressing a pole of a spectrum envelope of the narrow band spectrum parameter, and a modification for suppressing the pole .
A wideband speech reconstructing device comprising spectrum estimating means for outputting a narrowband spectrum parameter as a wideband spectrum parameter, and synthesizing means for generating a wideband speech signal using the outputted wideband spectrum parameter. 8. A analyzes the narrowband speech signal to obtain a narrowband spectral parameter analysis means, the narrowband spectral parameter is converted to a different area, perform suppress deformation of the poles of the spectral envelope, suppressing the pole
The transformed spectrum envelope is inversely transformed into the same region as the narrowband spectrum parameter to suppress the pole.
The narrowband spectrum obtained by inversely transforming the transformed spectral envelope.
Spectrum estimation means for <br/> output-vector parameter as wideband spectral parameters, wideband speech decompression apparatus comprising combining means for generating a wideband audio signal from the broadband spectrum parameters the output. 9. A narrowband spectrum parameter obtained by decoding a narrowband spectrum parameter from a narrowband spectrum code, suppressing a pole of a spectrum envelope of the decoded narrowband spectrum parameter, and performing a modification of suppressing the pole.
A spectrum decoding means for outputting a broadband spectrum parameter parameter, wideband speech decompression apparatus comprising combining means for generating a wideband speech signal using the wide-band spectral parameters outputted. 10. The method according to claim 9, further comprising wideband sound source estimation means for estimating a wideband sound source signal, wherein the synthesis means is configured to generate a wideband speech signal from the wideband sound source signal and a wideband spectrum parameter. Wideband audio restoration device. 11. A spectrum decoding means for estimating a wideband spectrum parameter using a narrowband spectrum code separated from a narrowband speech code, and a wideband excitation signal using the narrowband excitation code separated from the narrowband speech code. And a synthesizing means for generating a wideband speech signal from the estimated wideband spectrum parameter and the wideband excitation signal, wherein the wideband excitation decoding means is a narrowband excitation restored from a narrowband excitation code. Zero padding means for inserting and outputting a predetermined zero value in each sample interval of the signal, noise generating means for generating a noise signal, signal output by the zero padding means and noise signal generated by the noise generating means A wide-band voice restoration device comprising: an addition unit that adds and. 12. A spectrum decoding means for estimating a wideband spectrum parameter using a narrowband spectrum code separated from a narrowband speech code, and a wideband excitation signal using the narrowband excitation code separated from the narrowband speech code. And a synthesizing means for generating a wideband speech signal from the estimated wideband spectrum parameter and the wideband excitation signal, wherein the wideband excitation decoding means is a narrowband speech coded from an input narrowband speech code. Wideband adaptive excitation decoding means for estimating wideband adaptive excitation signal using band adaptive excitation code, and wideband driving excitation decoding means for estimating wideband driving excitation signal using narrowband driving excitation code separated from input narrowband speech code And generate a wideband source signal from the estimated wideband adaptive source signal and the wideband driving source signal. Wideband audio decompressor, characterized in that it is composed of a calculation unit. 13. A spectrum decoding means for estimating a wideband spectrum parameter using a narrowband spectrum code separated from a narrowband speech code, and a wideband excitation signal using the narrowband excitation code separated from the narrowband speech code. And a synthesizing means for generating a wideband speech signal from the estimated wideband spectrum parameter and the wideband excitation signal, wherein the wideband excitation decoding means is a narrowband speech coded from an input narrowband speech code. Wide-band long-period prediction code decoding means for estimating the wide-band long-period prediction code using the band-length long-period prediction code, and wide-band long-period prediction residual using the narrow-band long-period prediction residual code separated from the input narrow-band speech code. Wideband long-period prediction residual decoding means for estimating a difference signal, the estimated wideband long-period prediction code and wideband long-period prediction residual Signal and wideband speech decompression apparatus, characterized in that it is composed of a summing means for generating a wideband excitation signal from the. 14. A narrowband amplitude information decoding means for estimating narrowband amplitude information using a narrowband excitation code separated from a narrowband speech code, a narrowband spectrum code separated from the narrowband speech code, and the above. A spectrum decoding means for estimating a wideband spectrum parameter and wideband amplitude information using the narrowband amplitude information; a wideband excitation decoding means for estimating a wideband excitation signal using the narrowband excitation code and the wideband amplitude information; A wideband speech restoration device comprising a synthesizing means for generating a wideband speech signal from the wideband spectrum parameter and the wideband sound source signal. 15. decodes the narrowband spectral parameter from a narrowband spectrum code performs suppress deformation of the spectral envelope of the poles of the narrowband spectral parameters the decoded narrowband spectrum was suppressing deformation the pole
And spectrum decoding step of outputting the parameters as a broadband spectrum parameter, wideband speech decompression method comprising the synthetic steps to produce a wideband speech signal using the wide-band spectral parameters outputted. 16. A narrow band in which a pole of a spectrum envelope of a narrow band spectrum parameter obtained by analyzing a narrow band speech signal is modified, and the pole is modified.
Wideband audio decompressor for generating a wideband speech signal using the spectral parameter as a broadband spectral parameters. 17. A wideband sound source estimation means for estimating a wideband sound source signal is provided, and a wideband sound signal is generated from the generated wideband sound source signal and the wideband spectrum parameter.
The wideband speech restoration device according to item 1. 18. A narrow band in which a pole of a spectrum envelope of a narrow band spectrum parameter obtained by analyzing a narrow band speech signal is modified, and the pole is modified to be modified.
Wideband speech decompression method of generating a wideband speech signal using the spectral parameter as a broadband spectral parameters. 19. A voice transmission including a voice encoding device for encoding a narrow band voice signal and outputting a narrow band voice code, and a wide band voice restoring device for receiving the narrow band voice code and generating a wide band voice signal. In the system, the wideband speech decompression device decodes a narrowband spectrum parameter from a narrowband spectrum code included in the received narrowband speech code and suppresses a pole of a spectrum envelope of the decoded narrowband spectrum parameter. And perform deformation to suppress the pole.
A voice transmission system comprising a spectrum decoding means for outputting the narrowband spectrum parameter as a wideband spectrum parameter and a synthesizing means for generating a wideband voice signal by using the outputted wideband spectrum parameter. 20. A voice transmission including a voice encoding step of encoding a narrow band voice signal and outputting a narrow band voice code, and a wide band voice restoring step of receiving the narrow band voice code and generating a wide band voice signal. In the method, in the wideband speech restoration step, a narrowband spectrum parameter is decoded from the narrowband spectrum code included in the received narrowband speech code , and a pole of a spectrum envelope of the decoded narrowband spectrum parameter is suppressed. And perform deformation to suppress the pole.
A voice transmission method comprising a spectrum decoding step of outputting the narrowband spectrum parameter as a wideband spectrum parameter, and a synthesizing step of generating a wideband voice signal using the outputted wideband spectrum parameter. 21. A narrowband speech signal is analyzed to analyze a narrowband speech signal.
And a means for obtaining the spectrum parameter and a pole of the spectral envelope of the narrow band spectral parameter.
The pole structure is smoothed by performing deformation to smooth the structure.
Widen the narrowband spectral parameters that have been transformed
Spectral estimation output as band spectral parameter
Using the output means and the wideband spectral parameters output above.
Wideband speech recovery with synthesizing means for generating banded speech signals
Original device. 22. A narrowband speech signal is analyzed to analyze a narrowband speech signal.
Analysis means to obtain the spectrum parameter and the narrow band spectral parameter is converted into another region,
Perform a modification to smooth the polar structure of the vector envelope,
Spectral envelope modified to smooth the polar structure
To the same region as the narrowband spectral parameters above
Then, the spectrum that has been deformed to smooth the polar structure is
The narrow band spectral parameter obtained by inversely transforming the Toll envelope is widened.
Spectral estimation output as band spectral parameter
From the output wideband spectrum parameter
Wideband speech restoration device having synthesis means for generating speech signal
Place 23. From a narrowband spectrum code to a narrowband spectrum.
Decode the cut-off parameters and use this decoded narrowband spectrum.
Smooth the polar structure of the spectral envelope of the Toll parameter
To make the pole structure smooth.
The narrow band spectral parameters
The spectrum decoding means that outputs as a parameter and the wideband spectrum parameter that is output
Wideband speech recovery with synthesizing means for generating banded speech signals
Original device. 24. Wideband sound source for estimating a wideband sound source signal
Estimating means is provided, and the synthesizing means includes the wideband source signal and the wideband spectrum.
Parameter to generate a wideband speech signal.
24. A wideband speech decompression device according to claim 23, made up. 25. From a narrowband spectrum code to a narrowband spectrum.
Decode the cut-off parameters and use this decoded narrowband spectrum.
Smooth the polar structure of the spectral envelope of the Toll parameter
To make the pole structure smooth.
The narrow band spectral parameters
The spectrum decoding step to output as a parameter and the wideband spectrum parameter output to
Wideband sound with synthesis step to produce banded speech signal
Voice restoration method. 26. A narrow band obtained by analyzing a narrow band speech signal.
The polar structure of the spectral envelope of the band spectral parameters
Performs a smoothing deformation to smooth the pole structure.
The narrowband spectral parameters transformed by
Generate wideband speech signal using as vector parameter
Broadband audio restoration device. 27. Wideband sound source for estimating a wideband sound source signal
Estimating means, the generated wideband source signal and the wideband spectrum
27. A wideband speech signal is generated from the parameters.
The wideband speech restoration device according to item 1. 28. A narrow band obtained by analyzing a narrow band speech signal.
The polar structure of the spectral envelope of the band spectral parameters
Performs a smoothing deformation to smooth the pole structure.
The narrowband spectral parameters transformed by
Generate wideband speech signal using as vector parameter
Wideband audio restoration method. 29. A narrow band sound is obtained by encoding a narrow band speech signal.
A voice coding device for outputting a voice code, and the narrowband voice code
Wideband speech recovery for receiving a signal to generate a wideband speech signal
In voice transmission system comprising a device, the wideband speech decompression apparatus includes a narrowband spectrum included in narrowband voice code thus received
Decode the narrowband spectral parameters from the
Spectral envelope of decoded narrowband spectral parameters
To make the pole structure of the
Narrowband spectral parameters with smoothing deformation
To output as a broadband spectral parameter
Using the digital decoding means and the output wideband spectrum parameters.
A voice transmission system having a synthesizing means for generating a band voice signal.
Tem. 30. A narrow band sound is obtained by encoding a narrow band sound signal.
A voice encoding step for outputting a voice code, and the narrowband sound
Wideband speech that receives a voice code and produces a wideband speech signal
In the voice transmission method including a restoration step, the wideband speech restoration step includes a narrowband spectrum included in the received narrowband speech code.
Decode the narrowband spectral parameters from the
Spectral envelope of decoded narrowband spectral parameters
To make the pole structure of the
Narrowband spectral parameters with smoothing deformation
To output as a broadband spectral parameter
Wide by using the Le decoding step, the wideband spectral parameters outputted
Speech transmission with a synthesis step for producing a band speech signal
Method.