JP2003216196A - Speech encoding device and its method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To enhance quality of speech when background noise such as outdoor noise is superposed in a digital mobile radio communication system such as a car telephone and a cellular telephone. <P>SOLUTION: A speech encoding device is provided with a noise superposed section detection means for judging whether a signal to be inputted from a telephone microphone is a signal of a speech section or a signal of a noise section, a speech encoder A for encoding the signal of the speech section and a speech encoder B for encoding the signal of the noise section, when the noise superposed section detection means detects the noise section, control information regarding a decoding processing of the noise section is determined and transmitted. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention, in a digital mobile radio communication system such as a car phone or a mobile phone, improves coding quality when background noise such as outdoors is superimposed and improves transmission quality of noise-superimposed voice. Regarding the technology to make.

In recent years, with the improvement of communication technology, digital mobile radio communication systems such as car phones and mobile phones have become widespread. Along with this, an audio signal processing device that compresses an audio signal with high efficiency has been required.

[0003]

2. Description of the Related Art In a digital mobile radio communication system, it is desirable to encode a voice signal of 4 kHz band at a bit rate of about 4 to 8 kbps in order to effectively use radio frequencies. As a voice coding method corresponding to this,
The CELP method is known.

The CELP method analyzes a speech signal based on a linear prediction theory and extracts a parameter representing a frequency characteristic. At the same time, the driving excitation signal is waveform-wise encoded by vector quantization. Further, on the receiving side, the encoded voice transmitted on the transmission path is decoded by the procedure opposite to that on the transmitting side.

[0005]

By the way, the above CE
In the LP system, since the audio signal is compressed to a low bit rate and at the same time the reproduced audio quality is maintained, encoding (band compression) based on the audio generation model is performed. When encoding, an unnatural reproduced sound may be output. That is, in the conventional method, for a noise signal having a property different from that of voice,
Encoding processing is performed on the assumption that it has the same properties as speech. For this reason, a signal having only background noise is encoded and reproduced as an unnatural sound even though there is no frequency correlation.

Further, conventionally, when encoding voice,
The adaptive codebook is referenced based on the waveform of the voice to detect index information of similar waveform patterns. However, when noise is superimposed on the voice, there is no waveform pattern similar to the adaptive codebook, and there is no choice but to select a waveform pattern that is not very similar, which is unnatural when decoded. There is a problem that it is output as voice.

When the air conditioning noise is taken as an example of the background noise, the spectrum of the original sound of the air conditioning noise shows a substantially flat characteristic as shown in FIG. 13, and the time variation is small. On the other hand, in the reproduced sound of the air conditioning sound, as shown in FIG. 14, the peak of the spectrum envelope fluctuates for each frame. The inventor of the present invention paid attention to the fact that the fluctuation of the spectrum envelope causes unnaturalness in hearing, and investigated the cause of this spectrum fluctuation. That is, in the conventional decoder, the excitation signal is generated from the adaptive codebook and the noise codebook, and the decoding process is performed by passing the excitation signal through the synthesis filter. We analyzed whether it was due to the signal generation processing or the synthesis filter. As a result, no temporal fluctuation was found in the spectrum of the excitation signal. On the other hand, in the case of the synthesis filter, the variation shown in FIG. 15 appeared.

Therefore, the present invention has been made in view of the above problems, and discriminates between a signal containing only noise and a signal containing voice, and an encoding process or a decoding process of a signal containing only noise. It is a first object of the present invention to provide a device that outputs a audibly natural reproduced sound by differentiating the noise and suppressing the characteristics of the synthesis filter when only noise is decoded.

A second object is to provide a technique capable of preventing the influence of noise and performing a high quality encoding process when encoding a signal in which noise is superimposed on voice.

[0010]

The present invention adopts the following means in order to solve the above problems. This will be described with reference to the drawings.

(Means for Solving the First Problem) First, means for solving the first problem will be described with reference to the principle diagram of FIG.

The speech decoding system of the present invention comprises a noise weight section detecting means 1, a speech decoding means 2, a noise decoding means 3 and a noise control means 4. The noise superposed section detecting means 1 has a function of monitoring the signal encoded on the transmitting side and discriminating whether the section is a voice section containing voice or a noise section containing only noise. For example,
The noise superimposition section detection means 1 may detect the power of the signal from the encoded signal and determine whether this power is equal to or greater than a preset threshold value. That is, if the power of the coded signal is equal to or higher than the threshold value, it is determined that the speech section,
If the power of the encoded signal is less than the threshold value, it may be determined to be a noise section. Further, the gain of the encoded signal may be used instead of the power.

The speech decoding means 2 has a function of decoding the coded signal into a waveform signal when the noise superposed section detection means 1 discriminates the coded signal in the speech section. Specifically, a codebook 3a that registers a waveform pattern for each index information, a drive sound source 3b that excites the waveform pattern read from this codebook 3a, and an excitation signal that is output from this drive sound source 3b are filtered. And a synthesis filter 3c for processing.

The noise decoding means 3 has a function of decoding the coded signal into a waveform signal when the noise superimposed section detecting means 1 discriminates the coded signal in the noise section. Specifically, a codebook 3a that registers a waveform pattern for each index information, a drive sound source 3b that excites the waveform pattern read from this codebook 3a, and an excitation signal that is output from this drive sound source 3b are filtered. And a synthesis filter 3c for processing.

The noise control means 4 controls the filter coefficient of the synthesis filter 3c of the noise decoding means 3 when the noise superimposition interval detection means 1 determines the coded signal in the noise interval,
It has the function of suppressing the frequency characteristics of noise. Specifically, the noise control unit 4 has a function of determining a positive value of 1 or less to be multiplied by 1 or less by the filter coefficient.

When a post filter 9 for amplifying the amplitude of the decoded signal output from the synthesizing filter 3c is provided at the subsequent stage of the noise decoding means 3 and the speech decoding means 2, this post filter 9 is It has a function of passing the decoded signal in the noise section output from the noise decoding means 3 as it is.

Next, the speech coding system of the present invention will be described. The speech coding system of the present invention includes a noise superimposition section detection means 5, a speech coding means 6, a noise coding means 7, and a control information generation means 8.

The noise superposed section detection means 5 monitors the coded signal coded on the transmission side and determines whether the signal is a voice section signal containing voice or a noise section signal containing only noise. Has a function to identify. Specifically, by analyzing the waveform characteristics included in the encoded signal to determine whether the power of the signal is less than a threshold value or whether the gain of encoding is less than the threshold value, noise The section may be identified.

Further, the voice coding means 6 has a function of, when the noise superposed section detection means 5 discriminates the voice section, coding into the index information for specifying the waveform of this section. Specifically, the voice encoding unit 6 includes a codebook that registers a waveform pattern for each index information.

The noise coding means 7 has a function of, when the noise superposed section detection means 5 discriminates the noise section, coding into the index information for specifying the waveform of this section.
Like the voice encoding means 6, this encoding means is provided with a codebook for registering a waveform pattern for each index information.

The control information generation means 8 generates control information for decoding processing of this noise section when the noise superposed section detection means 5 determines the noise section, and uses this control information as a coded signal of the noise section. It has a function of being added to and transmitted to the receiving side. Specifically, it is information that controls the filter coefficient of the synthesis filter used on the decoding side based on the waveform characteristics of noise. For example, the filter coefficient should be multiplied by 1
The following positive values are used as control information.

(Means for Solving the Second Problem) Next, means for solving the second problem will be described with reference to the principle diagram of FIG.

This means is applied to the coding system, and includes the noise superimposition section detecting means 10, the inverse filter means 11, the noise removing means 12, the pitch period detecting means 13,
And voice encoding means 14.

The noise superposed section detection means 10 monitors the signal input from the transmitter and detects a voice section containing only voice,
It has a function of discriminating between a noise section including only noise and a noise superposition section in which noise is superimposed on speech.

The inverse filter means 11, when the noise-superimposed section detecting means 10 determines the noise-superimposed section, performs linear prediction analysis on the noise-superimposed section to obtain a linear prediction coefficient, and uses the linear prediction coefficient as a filter coefficient. It has a function of performing filtering processing. The prediction residual signal output from the inverse filter means 11 is input to the noise removing means 12.

The noise removing means 12 has a function of removing a noise portion from the prediction residual signal. As the noise removing means 12, for example, a low pass filter can be used.

The pitch period detecting means 13 has a function of obtaining an autocorrelation function of the residual signal output from the noise removing means 12 and detecting a pitch period at which the autocorrelation function has a maximum value. That is, the prediction residual signal is shifted by a specific cycle, and the specific cycle in which the correlation between each prediction residual signal and the original prediction residual signal is maximum is detected as the pitch cycle.

The voice coding means 14 has a function of coding the waveform of the noise superposed section based on the pitch cycle detected by the pitch cycle detection means 13.

[0029]

A system for solving the first problem of the present invention will be described.

(Operation of system for solving first problem)
In the voice decoding system of the present invention, when the coded signal coded on the transmitting side is received, the noise superposed section detection means 1 is a coded signal of a noise section containing only noise, or a code of a voice section containing speech. It is determined whether or not the signal is a digitized signal.

If the coded signal is a coded signal in the voice section, the coded signal is input to the voice decoding means 2. The voice encoding means 2 encodes the encoded signal into a waveform signal.

If the coded signal is a coded signal in the noise section, this coded signal is input to the noise decoding means 3. Then, the noise decoding means 3 detects a waveform pattern corresponding to the index information from the codebook 3a and excites this waveform pattern via the driving sound source 3b. Then, the excitation signal is input to the synthesis filter 3c. At the same time, the noise control means 4 multiplies the filter coefficient of the synthesis filter 3c by a positive value of 1 or less to synthesize the synthesis filter 3c.
Notify to. The synthesizing filter 3c performs a filtering process on the excitation signal based on the filter coefficient notified from the noise control means 4, and outputs a decoded signal. As a result, the waveform in the noise section is reproduced without unnaturally enhancing the frequency characteristic.

Further, the noise decoding means 3 may be processed by setting the gain of the noise section to "0". Further, when the post filter 9 is provided at the subsequent stage of the synthesis filter 3c, the post filter 9 passes the peak of the noise waveform output from the synthesis filter 3c without emphasizing it (without processing anything).

Next, in the speech coding system of the present invention,
When a signal is input from the transmitter, the noise superposed section detection means 5 determines whether the input signal is a signal in a voice section including voice,
Alternatively, it is identified whether the signal is in a noise section including only noise.

If the input signal is a signal in the voice section, the voice encoding means 6 discriminates a waveform pattern similar to the waveform in the voice section, encodes this waveform pattern into index information for specifying, and receives the index information. Send to the side.

If the input signal is a signal in the noise section, the noise encoding means 7 discriminates a waveform pattern similar to the waveform in the noise section and encodes this waveform pattern into index information for specifying. At the same time, the control information generation means 8 generates control information regarding the decoding process of the noise section and adds it to the index information. In particular,
The control information generating means 8 performs linear prediction analysis on the input signal in the noise section to determine the frequency characteristic, multiplies the obtained filter coefficient by a positive value of 1 or less, and sets the filter coefficient of the synthesis filter used on the receiving side. decide. Then, this filter coefficient is transmitted as control information to the receiving side together with index information.

A system for solving the second problem of the present invention will be described below. (Operation of the system for solving the second problem) In the voice coding system of the present invention, the noise superimposition section determination means 10 monitors the signal input from the transmitter and determines whether the voice section includes only the voice. , It is determined whether it is a noise section including only noise or a noise superposition section in which noise is superimposed on speech.

When the noise-superimposed section is determined, the inverse filter means 11 obtains the prediction coefficient of the noise-superimposed section,
Filtering processing using this prediction coefficient as a filter coefficient is performed and a prediction residual signal is output. This prediction residual signal is input to the noise removing means 12 and the noise portion is removed.

The prediction residual signal from which the noise portion has been removed by the noise removing means 12 is input to the pitch cycle detecting means 13. The pitch period detecting means 13 obtains the autocorrelation function of the prediction residual signal, and detects the pitch period at which this autocorrelation function has the maximum value.

Then, the voice encoding means 14 discriminates a waveform pattern similar to the waveform of the noise superposed section on the basis of the pitch period detected by the pitch period detection means 13, and encodes this waveform pattern into index information for specifying the waveform pattern. To do.
As a result, the audio signal can be encoded without being affected by noise.

[0041]

Embodiments of the present invention will be described with reference to the drawings. <First Embodiment> A first embodiment of the present invention will be described with reference to the drawings.

FIG. 3 is a block diagram showing the structure of the speech decoding system according to the first embodiment. The speech coding system according to the first embodiment includes a noise superposition detection determiner 1 as a noise superposition section detecting means, a speech decoder A (2) as a speech coding means, and a speech decoder B as a noise decoding means. (3) and the reception code separation unit 15 are provided.

The speech decoder A (2) and the speech decoder B
The term (3) means that the CELP method is adopted as the decoding method. The reception code separation unit 15 has a function of separating the coded signal received from the transmission side into power information, index information, and a synthesis filter coefficient.

The noise superimposition detection / determination unit 1 compares the power information separated by the reception code separation unit 15 with a preset threshold value, and if the power information is greater than or equal to the threshold value, determines the coded signal as a voice section, If the power information is less than the threshold, it has a function of determining the coded signal as a noise section. Further, the noise superimposition detection / determination unit 1 has a function of inputting the coded signal of the voice section to the voice decoder A (2) and inputting the coded signal of the noise section to the voice decoder B (3). ing.

The voice decoder A (2) decodes the coded signal in the voice section. Specifically, conventional CE
It has the same configuration and function as the LP type decoder,
The description is omitted.

The voice decoder B (3) decodes the coded signal in the noise section. Here, FIG. 4 shows an internal configuration and a peripheral configuration of the speech decoder B (3). In the figure, the speech decoder B (3) includes an adaptive codebook 30a, a noise codebook 31a, a driving excitation 3b, and a synthesis filter 3c.
Is equipped with. The LPC coefficient correction unit 4 as the noise control means of the present invention is connected to the synthesis filter 3c.

The adaptive codebook 30a is for registering the waveform pattern of a waveform signal having periodicity and index information, and has the function of updating the waveform pattern with the decoded waveform signal.

The random codebook 31a registers the waveform pattern of a waveform signal having no periodicity and index information. The adaptive codebook 30a and the noise codebook 31a define the amplification factors (gains) of the waveform patterns read from them, respectively, and the driving sound source 3b is read from the adaptive codebook 30a and the noise codebook 31a. It has a function of exciting the waveform pattern according to each gain.

The synthesis filter 3c filters the excitation signal output from the driving sound source 3b,
It is for decoding into a waveform signal. The filter coefficient of the synthesis filter 3c is determined on the transmission side. That is, on the transmitting side, the original waveform signal is subjected to linear prediction analysis to calculate linear prediction coefficients, and the linear prediction coefficients are transmitted to the receiving side as filter coefficients. As a result, the speech decoder B (3)
A filter coefficient is detected from the encoded signal and this filter coefficient is used as the filter coefficient of the synthesis filter 3c.

The LPC coefficient correcting section 4 has a function of receiving the judgment result of the noise superposition detection judging device 1 and correcting the filter coefficient of the synthesizing filter 3c. Specifically, as shown in the following formula, the filter coefficient of the synthesis filter 3c is
It has a function of correcting a filter coefficient by multiplying by a positive number of 1 or less. α′i = gi × αi (0.0 <g ≦ 1.0) As a result, the frequency characteristic of the synthesis filter 3c can be made substantially flat (see FIG. 12).

The operation of the speech decoding system will be described below. In the voice decoding system, the reception code separation unit 15 receives the coded signal coded on the transmission side.

The reception code separation unit 15 separates the coded signal into power information, index information and filter coefficients, and inputs the power information to the noise superimposition detection and determination unit 1.
The noise superposition detection determiner 1 determines whether the power information is equal to or more than the threshold value or less than the threshold value. Here, if the power information is greater than or equal to the threshold value, the noise superimposition detection determination unit 1
The encoded signal is determined to be a signal in the voice section, and the power information, the index information, and the filter coefficient separated by the reception code separator 15 are input to the voice decoder A (2). The voice decoder A (2) decodes the coded signal into a voice waveform based on these pieces of information.

On the other hand, when the power information is less than the threshold value,
The noise superimposition detection / determination unit 1 determines that the encoded signal is a signal in a noise section, inputs the power information and the index information separated by the reception code separation unit 15 to the speech decoder B (3), and at the same time sets the filter coefficient. Notify the LPC coefficient correction unit 4.

In the speech decoder B (3), the adaptive codebook 30a or the noise codebook 31 is used based on the index information.
A is searched for the corresponding waveform pattern. Then, the driving sound source 3b excites the waveform pattern according to the gain of each codebook, and inputs the excitation signal to the synthesis filter 3c.

Here, the LPC coefficient correcting section 4 corrects the filter coefficient by multiplying the filter coefficient by a positive value of 1 or less. Then, the corrected filter coefficient is used as the synthesis filter 3c.
Notify to.

The synthesis filter 3c includes the LPC coefficient correction unit 4
The excitation signal output from the driving sound source 3b is filtered according to the filter coefficient notified from the above and decoded into a noise waveform. As described above, according to the first embodiment, the spectrum of the synthesis filter 3c can be made to have a substantially flat characteristic by controlling the filter coefficient when the signal in the noise section is encoded, and the characteristic of the noise waveform can be Unnatural enhancement can be prevented, and the reproduction of noise that is annoying to the ear can be suppressed. Therefore, it is possible to improve the voice quality of mobile communication such as mobile phones and car phones.

<Embodiment 2> In Embodiment 2, an example in which the system of the present invention is applied to an encoder will be described. Figure 5
FIG. 1 is a schematic configuration diagram of a voice encoding system.

In the figure, the speech encoding system comprises a speech encoder A (6), a speech encoder B (7), and a noise superposition detection decision unit 5. The noise superimposition detection / determination device 5 is
The power of the waveform signal input from the transmitter is detected. If this power is greater than or equal to the threshold value, it is determined to be the waveform signal in the voice section containing speech. If the power is less than the threshold value, the waveform signal in the noise section containing only noise. It has the function to judge. Then, the noise superimposition detection and determination unit 5 has a function of inputting the waveform signal of the voice section to the voice encoder A (6) and inputting the waveform signal of the noise section to the voice encoder B (7).

The voice encoder A (6) has a function of encoding the waveform signal in the voice section and is a conventional CELP system encoder. The voice encoder B (7) has a function of encoding the waveform signal in the noise section.

FIG. 6 shows the internal structure and the peripheral structure of the speech coder B (7). In the figure, a speech coder B
(7) includes an adaptive codebook 70a, a noise codebook 71a, a driving sound source 7b, a synthesis filter 7c, an LPC analysis section 7e, and an error minimization section 7d.

The adaptive codebook 70a registers a waveform pattern of a waveform having periodicity and index information for identifying each waveform pattern. The noise codebook 71a is
A waveform pattern having a waveform having no periodicity and index information for identifying each waveform pattern are registered.

The driving sound source 7b has a function of exciting the waveform pattern detected from the adaptive codebook 70a and the waveform pattern detected from the noise codebook 71a according to the gain of each codebook.

The synthesis filter 7c has a function of performing a filtering process using the linear prediction coefficient of the waveform signal in the noise section as a filter coefficient. The error minimization unit 7d compares the waveform signal output from the synthesis filter 7c with the waveform of the input noise signal, optimizes the index information and the amplification factor (gain) of the waveform pattern, and then the noise codebook 71.
It has a function of updating the contents of a.

The LPC analysis section 7e has a function of performing linear prediction analysis on the input waveform to calculate linear prediction coefficients and inputting the linear prediction coefficients to the synthesis filter 7c as filter coefficients.

Further, the code encoder B (7) is connected to the code transmitter 16 and the LPC coefficient corrector 8. The code transmitting unit 16 has a function of transmitting the power information, the index information, and the filter coefficient encoded by the voice encoder B (7) to the receiving side.

The LPC coefficient correction unit 8 has the same function as that of the above-described first embodiment, and has a function of correcting the filter coefficient of the synthesis filter 7c used when decoding the coded signal in the noise section. have. Specifically, correction is performed by multiplying the filter coefficient by a positive value of 1 or less. In response to this, the code transmission unit 16 transmits the filter coefficient corrected by the LPC coefficient correction unit 8 together with other coded signals.

The operation of the speech coding system according to the second embodiment will be described below. When the waveform signal is input from the transmitter, the noise superimposition detection determination unit 5 detects the power of the waveform signal and determines whether it is equal to or more than the threshold value or less than the threshold value. Here, if the power of the waveform signal is equal to or higher than the threshold value, the noise superimposition detection determination unit 5 determines that the waveform signal is a waveform signal in the voice section, and the waveform signal is determined by the voice encoder
Input to (6).

The voice encoder A (6) encodes the waveform information into index information, power information, and filter coefficient using a codebook, and transmits it to the receiving side. Further, when the power of the input waveform is less than the threshold value, the noise superimposition detection determination unit 5 determines that the waveform signal is a waveform signal in a noise section,
This waveform signal is input to the voice encoder B (7).

In the speech coder B (7), the adaptive codebook 70
It has a function of searching a and the noise codebook 71a based on the waveform of the noise section and detecting a similar waveform pattern. Further, the speech coder B (7) is adapted to the adaptive codebook 70a.
Alternatively, the waveform pattern read from the noise codebook 71a is input to the driving sound source 7b.

The driving sound source 7b excites the waveform pattern and inputs it to the synthesis filter 7c. Here, the LPC analysis unit 7e performs a linear prediction analysis on the input waveform signal to calculate a linear prediction coefficient. Then, the LPC analysis unit 7e notifies the synthesis filter 7c of the linear prediction coefficient.

The synthesis filter 7c performs a filtering process using the linear prediction coefficient as a filter coefficient on the excitation signal input from the driving sound source 7b. Error minimizing unit 7d
Compares the decoded signal output from the synthesizing filter 7c with the input waveform signal, and obtains the optimum index information and the gain of the waveform pattern to minimize the error between the two, the adaptive codebook 70a and the noise code. Notify the book 71a.
Then, each codebook updates the registered content and the gain based on the index information and the gain notified from the error minimizing unit 7d, and notifies the code transmitting unit 16 of the updated index information. Furthermore, the LPC coefficient correction unit 8
The linear prediction coefficient (filter coefficient) calculated by the C analysis unit 8 is multiplied by a positive value of 1 or less to correct the filter coefficient. Then, the LPC coefficient correction unit 8 notifies the code transmission unit 16 of the corrected filter coefficient.

The code transmitting unit 16 receives the index information and the power information notified from the speech coder B (7) and LP.
The filter coefficient notified from the C coefficient correction unit 8 is notified to the receiving side.

As a result, on the receiving side, the spectrum of the synthesis filter can be made to have a flat characteristic by performing the decoding process using the corrected filter coefficient.
It is possible to prevent the waveform in the noise section from being unnaturally decoded.

As described above, according to the second embodiment, when decoding the noise section, the spectrum of the synthesis filter can be made to have a flat characteristic, the frequency characteristic of the noise section is not unnatural, and the auditory sense is heard. The noise that is offensive to the ear can be suppressed.

<Third Embodiment> A third embodiment of the present invention will be described below with reference to the drawings. FIG. 7 shows the third embodiment.
3 shows the internal configuration of the speech coder B in FIG.

In the figure, the speech coder B (7) is different from the speech coder B (7) of the second embodiment described above in the adaptive codebook 70a, the noise codebook 71a, the driving sound source 7b, and the synthesis filter 7c. , LPC analysis unit 7e, and error minimization unit 7d. Further, the code transmitter 16 is connected to the voice encoder B (7).

The code transmitting section 16 transmits the adaptive codebook 70a when transmitting the coded signal in the noise section including only the noise section.
It has a function of transmitting "0" as the index information of. Other configurations and functions are similar to those of the above-described second embodiment, and description thereof will be omitted.

FIG. 8 is a block diagram showing a configuration of a speech decoder B (3) corresponding to the speech encoder B (7) of FIG. The speech decoder B (3) has an adaptive codebook 30a, a noise codebook 31a, a driving excitation 3b, a synthesizing filter 3c, and an adaptive postfilter 17 in addition to the configuration of the first embodiment.
Is equipped with.

The adaptive post filter 17 has a function of amplifying the amplitude value without changing the cycle of the waveform. Further, the adaptive codebook 30a is transmitted from the transmitting side by the adaptive codebook 30a.
When the index information “0” is received, the adaptive codebook 3
The gain of 0a is set to "0". Thus, when the waveform signal in the noise section is input, it has a function of searching the noise codebook 31a based on the index information of the noise codebook 31a and reading the corresponding waveform pattern. further,
When the waveform signal in the noise section is input, the adaptive post filter 17 passes the waveform signal without performing any processing.

According to the third embodiment, the noise waveform having no periodicity is encoded and decoded by the noise codebook, so that the noise signal having a flat characteristic having no periodicity causes unnatural periodicity during the decoding process. Can be decoded into a waveform signal that is acoustically natural without adding.

<Fourth Embodiment> FIG. 9 shows the configuration of an encoder B according to the fourth embodiment. The encoder B (7) includes an adaptive codebook analysis unit 18, a noise codebook analysis unit 19, a driving excitation generation unit 20, and an open loop pitch analysis unit 21.

The adaptive codebook analysis unit 18 uses the random codebook 71.
The waveform signal detected from a has a function of performing a filtering process by the long-term prediction synthesis filter 72 and performing a closed loop process of calculating the pitch period of the waveform signal (see FIG. 10).

On the other hand, the open loop pitch analysis section 21 is activated when encoding a noise superposed section in which a noise waveform is superposed on a speech waveform, and the short-term prediction inverse filter 11, low-pass filter LPF 12 and autocorrelation detection are performed. It includes a unit 13b, a maximum correlation value detection unit 13c, and a delay unit 13a (see FIG. 11).

The short-term prediction inverse filter 11 has a function of performing an inverse filtering process using a linear prediction coefficient of a waveform signal as a filter coefficient and outputting a prediction residual signal. The low-pass filter LPF12 has a function of removing the waveform of the noise part from the prediction residual signal.

The delay unit 13a changes the cycle of the prediction residual signal to
It has the function of shifting by a specific period. The autocorrelation detection unit 13b has a function of detecting the correlation value between the original prediction residual signal and the prediction residual signal shifted by the delay unit 13a by the specific period.

The maximum correlation value detector 13c includes a delay unit 13a.
Has a function of detecting the amount of delay (cycle) having the largest correlation by shifting by a specific cycle. This delay is
The driving sound source 7b is notified of the pitch cycle. And
The driving sound source 7b excites the waveform pattern read from the applied codebook 70a based on this pitch period.

As described above, according to the fourth embodiment, the pitch period of the voice waveform on which noise is superimposed can be detected accurately,
It is possible to perform high-quality coding processing that is not affected by the presence or absence of noise, and improve the quality of reproduced speech.

[0088]

According to the present invention, it is possible to prevent unnatural frequency characteristics from being added to noise with a small change in frequency characteristics, and to reduce discomfort during reproduction.

Furthermore, when encoding a signal in which noise is superimposed on speech, a noise component is removed and an accurate pitch period is detected, so that high-quality encoding can be performed. Therefore,
According to the present invention, it is possible to contribute to the improvement of voice quality in mobile communication systems such as mobile phones and car phones.

[Brief description of drawings]

FIG. 1 is a principle diagram of the present invention (1)

FIG. 2 is a principle diagram of the present invention (2)

FIG. 3 is a schematic configuration diagram of a speech decoding system according to the first embodiment.

FIG. 4 is a block diagram of an internal configuration of a speech decoder B.

FIG. 5 is a schematic configuration diagram of a speech encoding system according to a second embodiment.

FIG. 6 is a block diagram of an internal configuration of a speech coder B.

FIG. 7 is a block diagram of the internal configuration of a speech coder B according to the third embodiment.

FIG. 8 is a block diagram of the internal configuration of a speech decoder B according to the third embodiment.

FIG. 9 is a schematic configuration diagram of a speech encoding system according to a fourth embodiment.

FIG. 10 is a block diagram showing an internal configuration of an adaptive codebook analysis unit.

FIG. 11 is an internal configuration block diagram of an open loop analysis unit.

FIG. 12 is a spectrum showing frequency characteristics of a synthesis filter.

FIG. 13 is a diagram showing a spectrum of a source sound of an air conditioning sound.

FIG. 14 is a diagram showing a spectrum of a reproduced sound of an air conditioning sound.

FIG. 15 is a spectrum showing frequency characteristics of a synthesis filter.

[Explanation of symbols]

1 .. Noise superimposition section detection means ..Voice decoding means 3 ... Noise decoding means 3a ... Code book 3b ... Drive sound source 3c ... Synthesis filter 4. Noise control means (LPC coefficient correction unit) 5..Noise superposition section detection means (noise superposition detection judgment device) 6 ... Speech coding means 7 .. Noise coding means 7b ... Drive sound source 7c ... Synthesis filter 7d ... error minimization section 7e ... LPC analysis unit 8 ... Control information generation means (LPC coefficient correction unit) 9 ... Post filter ..Noise-superimposed section detection means 11. Inverse filter means (short-term predictive inverse filter) ..Noise removing means (low-pass filter LPF) ..Pitch cycle detection means 13a ... Delay unit 13b ... Auto-correlation detector 13c ... Correlation maximum value detection unit ..Voice coding means 15 ... Reception code separation unit 16 ... Code transmitter 17 ... Adaptive post filter 18 ... Adaptive codebook analysis unit 19-Noise codebook analysis unit 20. Drive source generator 21 ... Open loop pitch analysis section 30a ... Adaptive codebook 31a ... Noise codebook 70a ... Adaptive codebook 71a ... Noise codebook 72 ... Long-term prediction synthesis filter

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Tomoki Sato             4-1, Kamiodanaka, Nakahara-ku, Kawasaki-shi, Kanagawa             No. 1 within Fujitsu Limited (72) Inventor Tosei Kimura             3-22-8, Hakata Station, Hakata-ku, Fukuoka City, Fukuoka Prefecture             Issue Fujitsu Kyushu Digital Technology Co., Ltd.             Inside the company F-term (reference) 5D045 CC02 CC07                 5K067 AA24 BB03 BB04 BB21 DD45                       EE02 EE10 EE16 FF16 FF25                       FF26 GG01 GG11 HH22 HH23                       HH24 KK13 KK15

Claims (8)

[Claims] 1. A noise superposed section detection means for determining whether a signal input from a transmitter is a signal in a voice section or a signal in a noise section, and a voice for encoding the signal in the voice section. An encoder A and a speech encoder B for encoding a signal in the noise section are provided, and when the noise superposed section detection means determines the noise section,
Determine the control information for the decoding process of this noise interval,
A speech coder that transmits this control information. 2. The speech coding apparatus according to claim 1, wherein the control information is a filter coefficient of a synthesizing filter used for decoding processing on the receiving side. 3. The control information should be multiplied by 1 to a filter coefficient of a synthesis filter used in a decoding process on the receiving side.
The speech encoding device according to claim 1, wherein the following positive values are set. 4. The control information is 1 to be multiplied by a filter coefficient of a synthesis filter used for decoding processing on the receiving side.
The audio encoding device according to claim 1, wherein the control information generating unit determines the positive value based on a frequency characteristic of the noise section, which is the following positive value. 5. A step of determining whether a signal input from a transmitter is a signal in a voice section or a signal in a noise section, a step of encoding a signal in the voice section, and the noise And a step of encoding a signal in a section, and determining a control information related to a decoding process of the noise section when the noise section is determined in the determining step and transmitting the control information. Characteristic speech coding method. 6. The speech coding method according to claim 5, wherein the control information is a filter coefficient of a synthesis filter used for decoding processing on the receiving side. 7. The control information is 1 to be multiplied by a filter coefficient of a synthesis filter used for decoding processing on the receiving side.
The audio encoding method according to claim 5, wherein the following positive values are used. 8. The control information is 1 to be multiplied by a filter coefficient of a synthesis filter used for decoding processing on the receiving side.
6. The following positive value, further comprising the step of determining the positive value based on the frequency characteristic of the noise section.
The described speech coding method.