JPH09114498A - Speech encoding device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To perform an encoding process excellently even if a non-speech signal is inputted by continuously outputting the spectrum parameter of a precedent frame to a linear predictive analyzing means (LPC analyzer) when the non-speech signal lasts. SOLUTION: A buffer memory 1 sends out an input signal in frame units to a subframe divider 7 and a speech discrimination unit 2. A switch control circuit 3 sets a variable (i) indicating the number of successive non-sound frames to 0 when encoding is started. Then, when the speech discrimination unit 2 discriminates as a non-speech signal, the variable (i) is increased by one and it is judged whether or not the variable (i) is a specific number R(e.g. 10). When the variable (i) is larger than the specific number R, a terminal of a changeover switch 4 is closed to a side (a). The LPC analyzer 5 continuously outputs the spectrum parameter of the precedent frame.

Description

Detailed Description of the Invention

[0001]

[0001] The present invention relates to a speech coding apparatus,
More specifically, the present invention relates to an audio encoding device for compressing and recording or transmitting an audio signal by digital information.

[0002]

2. Description of the Related Art As a widely used means for efficiently compressing a speech signal, a speech signal is encoded using a linear prediction parameter representing a spectral envelope and an excitation parameter corresponding to the linear prediction residual signal. There is a method to make it. A speech coding method using such a linear prediction method can obtain a relatively high quality synthesized speech with a small transmission capacity. Therefore, various application methods are actively studied in combination with the recent progress in hardware technology. And is being developed.

Among them, as a method for obtaining good sound quality, "Improved speech quality and e" by Kleijin et al.
fficient vector quantization in SELP "(ICASP'88 s
4.4, pp.155-158, 1988),
CELP (Code Excited Linear Predictive Codin) which uses an adaptive codebook obtained by repeating past sound source signals
g) The method is well known.

FIG. 7 is a block diagram showing the configuration of a code driven linear predictive coding apparatus including the above adaptive codebook.

As shown, from the input terminal, for example, 8k
An original voice signal sampled at Hz (that is, 1/8 ms per sample) is input, and a voice signal having a predetermined frame interval (for example, 20 ms, that is, 160 samples) is stored in the buffer memory 51.

The buffer memory 51 sends the original audio signal in frame units to an LPC (Linear Predictive Coding) analyzer 55.

The LPC analyzer 55 performs a linear prediction analysis (LPC analysis) on the original speech signal, extracts a linear prediction parameter α which is a spectral parameter representing a spectral characteristic, and outputs it to a synthesis filter 56 and a multiplexer 68. Send out.

Further, the sub-frame divider 57 receives the original audio signal input from the buffer memory 51 on a frame-by-frame basis, at a predetermined sub-frame interval (for example, 5 m).
s, that is, 40 samples). That is, in the above example, four subframe signals from the first subframe to the fourth subframe are created from the original audio signal of one frame.

Next, the delay L and the gain β of the adaptive codebook are determined by the following processing.

First, in the delay circuit 61, the input signal of the synthesizing filter 56 in the preceding sub-frame, that is, the driving sound source signal, is delayed by a delay corresponding to the pitch period to create an adaptive code vector.

For example, the assumed pitch period is 40 to 16
Assuming 7 samples, 12 samples with a delay of 40 to 167 samples
Eight types of signals are created as adaptive code vectors and stored in the adaptive code book 62.

At this time, the switch 66 is in an open state, each adaptive code vector is multiplied by the gain value varied by the multiplier 63, and then passed through the adder 67 to be input to the synthesis filter 56 as it is. .

The synthesizing filter 56 performs the synthesizing process using the linear prediction parameter α and sends the synthesized vector to the subtractor 58. The subtracter 58 generates an error vector by subtracting the original speech vector and the synthetic vector, and sends the obtained error vector to the perceptual weighting filter 59.

The perceptual weighting filter 59 performs weighting processing on the error vector in consideration of perceptual characteristics, and sends it to the error evaluator 60.

The error evaluator 60 calculates the mean square of the error vector, searches for the adaptive code vector having the smallest mean square value, and sends the delay L and the gain β to the multiplexer 68. In this way, the delay L and the gain β of the adaptive codebook 62 are determined.

Subsequently, the index i and the gain γ of the probability codebook 64 are determined by the following processing.

The probability codebook 64 has a dimension number corresponding to the subframe length (that is, 40 dimensions in the above example).
Are stored in advance, for example, in the form of 512 types, and each is assigned an index. At this time, the switch 66 is in a closed state.

First, the optimum adaptive code vector determined by the above processing is multiplied by the optimum gain β in the multiplier 63 and then sent to the adder 67.

Next, each probability code vector is multiplied by the multiplier 65.
The gain value is varied and multiplied by, and then input to the adder 67. The adder 67 adds the optimal adaptive code vector multiplied by the optimal gain β and each probability code vector,
The result is input to the synthesis filter 56.

The subsequent processing is performed in the same manner as the adaptive codebook parameter determination processing. That is, the synthesizing filter 56 performs the synthesizing process using the linear prediction parameter α, and sends the synthesized vector to the subtractor 58.

The subtractor 58 generates an error vector by subtracting the original speech vector and the synthetic vector, and sends the obtained error vector to the perceptual weighting filter 59.

The perceptual weighting filter 59 performs a weighting process on the error vector in consideration of perceptual characteristics, and sends it to the error evaluator 60.

The error evaluator 60 calculates the mean square of the error vector, searches for the probability code vector having the smallest mean square value, and sends the index i and the gain γ to the multiplexer 68. In this way, the index i and the gain γ of the probability codebook 64 are determined.

The multiplexer 68 has a quantized linear prediction parameter α, an adaptive codebook delay L, a gain β, a probability codebook index i, and a gain γ.
Is to multiplex each.

Next, the operation of the speech decoding apparatus corresponding to the above speech encoding apparatus will be described in detail with reference to FIG. FIG. 8 is a block diagram showing a configuration of a decoding device corresponding to the code driven linear predictive coding device of FIG.

In the figure, the demultiplexer 78 is
The received signal is decomposed into a linear prediction parameter α, an adaptive codebook delay L and gain β, a probability codebook index i and a gain γ, and the linear prediction parameter α is used as a synthesis filter, and the adaptive codebook delay L and gain are used. β is output to the adaptive codebook 72 and the multiplier 73, and the index i and the gain γ of the probability codebook are output to the probability codebook 74 and the multiplier 75, respectively.

Based on the delay L of the adaptive codebook output from the demultiplexer 78, the adaptive code vector of the adaptive codebook 72 is selected. Here, the adaptive codebook 72 has the same content as the adaptive codebook 62 in the above-mentioned encoding device. That is, the past driving sound source signal is input to the adaptive codebook 72 via the delay circuit 71. The multiplier 73 uses the received gain β to obtain the adaptive code gain interpolation circuit 76.
The adaptive code vector input via the is amplified and sent to the adder 79.

Based on the index i of the probability codebook output from the demultiplexer 78,
A probability code vector in the probability code book 74 is selected. Here, the probability code book 74 has the same contents as the contents of the probability code book 64 in the above encoding device. The multiplier 75 uses the received gain γ to
The probability code vector input via the probability code gain interpolation circuit 77 is amplified and sent to the adder 79.

The adder 79 adds the amplified probability code vector and the amplified adaptive code vector, and sends them to the synthesis filter 80 and the delay circuit 71.

The synthesizing filter 80 performs synthesizing processing using the received linear prediction parameter α as a coefficient, and outputs it as a synthetic speech signal.

The speech coding apparatus based on the linear prediction analysis as described above has an advantage that high-quality coding performance can be obtained at a relatively low bit rate. The speech coding apparatus based on such a linear predictive analysis is configured on the premise of almost periodic voiced sound produced by humans, and it is said that an appropriate analysis length of one frame is around 20 ms.

[0032]

However, the conventional speech coder as described above cannot satisfactorily encode a non-speech signal other than a speech signal, and in particular, when background noise or the like is mixed, it becomes sharp. There was a problem that the sound quality deteriorates.

Mobile phones, voice recorders, and the like are considered as fields of application of the above-mentioned voice encoding device, and these are used in various environments including the case where background noise is mixed. Therefore, the problem of sound quality deterioration is an essential issue that must be solved in order to realize an attractive product.

The present invention has been made in view of the above circumstances, and it is an object of the present invention to provide a voice encoding device with good sound quality that can be encoded well even if a non-voice signal is input.

[0035]

In order to achieve the above object, the speech coder according to the first invention determines whether an input signal divided into predetermined frame intervals is a speech signal or a non-speech signal. A voice discriminating means for discriminating, a linear prediction analysis means for outputting the spectrum parameter of the input signal,
Control for causing the linear prediction analysis unit to continuously output the spectrum parameter in a predetermined preceding frame as the spectrum parameter of the input signal when the result of the determination by the voice determining unit is a non-voice signal continuously for a predetermined number of frames. Means, a driving sound source signal generating means for generating a driving sound source signal corresponding to the linear prediction residual signal, and a synthesis filter for synthesizing voice from the driving sound source signal based on the spectral parameter.

The speech coding apparatus according to the second aspect of the invention comprises speech discrimination means for discriminating whether the input signal divided into a predetermined frame interval is a speech signal or a non-speech signal,
Linear prediction analysis means for outputting the spectrum parameter of the input signal, and when the discrimination result by the voice discrimination means is a non-voice signal, until the next voice signal is discriminated within a range not exceeding a predetermined number of frames. Control means for buffering the input signal to cause the linear prediction analysis means to perform linear prediction analysis of the input signal at once, and driving excitation signal generation means for generating a driving excitation signal corresponding to the linear prediction residual signal, And a synthesis filter for synthesizing voice from the drive sound source signal based on the spectrum parameter.

Therefore, in the speech coder according to the first aspect of the invention, the speech discrimination means discriminates whether the input signal divided into the predetermined frame intervals is a speech signal or a non-speech signal, and the linear prediction analysis means is the above-mentioned. When the spectrum parameter of the input signal is output and the discrimination result by the voice discriminating means is a non-voice signal continuously over a predetermined number of frames, the control means determines the spectrum parameter of the input signal by the linear predictive analysis means. Continuously output the spectrum parameter in the preceding frame, the driving sound source signal generation means generates a driving sound source signal corresponding to the linear prediction residual signal, the synthesis filter based on the spectrum parameter from the driving sound source signal voice. To synthesize.

In the speech coder according to the second aspect of the invention, the speech discrimination means discriminates whether the input signal divided into the predetermined frame intervals is a speech signal or a non-speech signal, and the linear prediction analysis means is the above-mentioned. When the spectrum parameter of the input signal is output and the discrimination result by the voice discriminating means is the non-voice signal, the control means inputs the signal until it is discriminated to be the next voice signal within a range not exceeding the predetermined number of frames. The signal is buffered, the input signal is collectively subjected to linear prediction analysis by the linear prediction analysis unit, the driving excitation signal generation unit generates a driving excitation signal corresponding to the linear prediction residual signal, and the synthesis filter is the spectrum. Speech is synthesized from the driving sound source signal based on the parameters.

[0039]

Embodiments of the present invention will be described below with reference to the drawings. 1 to 4 show a first embodiment of the present invention, and FIG. 1 is a block diagram showing a configuration of a speech coding apparatus.

As shown in FIG. 1, the output terminal of the buffer memory 1 connected to the input terminal is branched into three, and the first output terminal is connected to the subtracter 8 via the subframe divider 7. The second output end is connected to the input end of the changeover switch 4, and the third output end is the voice discriminator 2 which is a voice discriminating means.
It is connected to the switch control circuit 3 which is a control means for controlling the changeover switch 4 via.

One output terminal a of the changeover switch 4 is connected to the synthesizing filter 6, and the other output terminal b is connected to the synthesizing filter 6 via the LPC analyzer 5 which is a linear predictive analysis means. There is.

The output terminal of the subtractor 8 is connected to the input terminal of the error evaluator 10 via the perceptual weighting filter 9. Further, the output terminal of the error evaluator 10 is connected to the adaptive codebook 12 and the probability. It is connected to the codebook 14 and also to the multipliers 13 and 15.

The adaptive codebook 12 has a multiplier 13
Is connected to the first input terminal of the adder 17 via
The probability codebook 14 is connected to the second input terminal of the adder 17 via the multiplier 15 and the switch 16.

The output terminal of the adder 17 is connected to the input terminal of the subtractor 8 via the synthesis filter 6 and the adaptive codebook 12 via the delay circuit 11.
It is connected to the.

The multiplexer 18 is connected to the voice discriminator 2, the LPC analyzer 5, and the error evaluator 10.

In the above speech coding apparatus, the driving excitation signal generating means for generating the driving excitation signal corresponding to the linear prediction residual signal is the delay circuit 11, the adaptive codebook 12, the probability codebook 14, and the multiplication. Vessels 13,15,
The switch 16 and the adder 17 are included.

Next, FIG. 2 is a block diagram showing a more detailed structure of the voice discriminator 2.

The output signal of the buffer memory 1 input to the voice discriminator 2 is branched into two so that one is input to the frame energy analysis circuit 2a and the other is input to the initial frame energy analysis circuit 2b. Has become.

The frame energy analysis circuit 2a is connected to the first input terminal which is the + terminal of the adder 2c, and the initial frame energy analysis circuit 2b is connected to the-of the adder 2c.
The initial frame energy analysis circuit 2b is connected to each of the second input terminals, which are terminals, and is also connected to the threshold value determination circuit 2d.

The output terminal of the adder 2c and the output terminal of the threshold value determining circuit 2d are both connected to the discriminating circuit 2e, and the output of the discriminating circuit 2e is output to the switch control circuit 3. It has become.

Next, the signal flow in the configuration shown in FIGS. 1 and 2 will be described.

From the input terminal, for example, 8 kHz (that is,
An original audio signal sampled at 1/8 ms per sample) is input, and an audio signal at a predetermined frame interval (for example, 20 ms, ie, 160 samples) is stored in the buffer memory 1.

The buffer memory 1 sends the input signal to the sub-frame divider 7 and the voice discriminator 2 in frame units.

The voice discriminator 2 discriminates whether the input signal of the frame is voice or non-voice by a method as described below, for example.

In the voice discriminator 2 having the structure shown in FIG. 2, the frame energy analysis circuit 2a calculates the frame energy Ef of the input frame input signal by the following mathematical formula.

[0056]

(Equation 1) Here, s (n) indicates an input signal in sample n, and N indicates a frame length.

Further, the initial frame energy analysis circuit 2b calculates the frame energy Eb at the time of starting the encoding by using the same mathematical expression as the mathematical expression 1.

The threshold decision circuit 2d decides a threshold according to the magnitude of the background noise energy. For example, as shown in FIG. 3, as the background noise energy increases in dB units, the threshold value is determined based on the relationship of decreasing the threshold value in dB units. Then, the result is sent to the discrimination circuit 2e.

In the adder 2c, the frame energy Ef
By inputting as a positive value and inputting the initial frame energy Eb as a negative value and adding them,
From the frame energy Ef to the initial frame energy E
b is subtracted, and the subtraction result is sent to the discrimination circuit 2e.

Then, the discriminating circuit 2e compares the input subtraction result with the threshold value and discriminates that the frame input signal is in the voice section if the subtraction result is larger than the threshold value, and is in the non-voice section otherwise. To determine.

Returning to FIG. 1, the subframe divider 7 is
The input signal of the frame is divided into predetermined subframe intervals (for example, 5 ms, that is, 40 samples). That is, four subframe signals from the first subframe to the fourth subframe are created from the input signal of one frame.

The LPC analyzer 5 performs a linear prediction analysis (LPC analysis) on the input signal, extracts a linear prediction parameter α which is a spectral parameter representing a spectral characteristic, and sends it to the synthesis filter 6 and the multiplexer 18.

Next, the operation of the switch control circuit 3 will be described with reference to the flowchart of FIG. FIG. 4 is a flowchart showing the operation of the speech coding apparatus.

When encoding is started, a variable i indicating the number of consecutive non-voice frames is set to 0 (step S1).

Next, it is determined whether the discrimination result by the voice discriminator 2 is voice (v) or non-voice (uv) (step S2).

If the determination result in step S2 is non-voice, the variable i is incremented by 1 (step S2).
3), it is determined whether the variable i is larger than a predetermined number R (for example, 10) (step S4).

If the variable i is larger than the predetermined number R (eg, 10) in step S4, the changeover switch 4
Is closed to the side a (step S5), and the spectrum parameter of the preceding frame is continuously used (step S6). After that, the process waits for the next frame (step S7).

On the other hand, when the discrimination result in the above step S2 is voice, after the variable i indicating the number of consecutive non-voice frames is reset to 0 (step S8), the terminal of the changeover switch 4 is closed to the side b. (Step S9), LP
The C analyzer 5 performs LPC analysis to update the spectrum parameter (step S10). Then, the process proceeds to step S7 and waits for the processing of the next frame.

If the variable i is not larger than the predetermined number R (for example, 10) in step S4, the process proceeds to step S8.

Returning to the explanation of FIG. 1, the delay L and the gain β of the adaptive codebook 12 and the index i and the gain γ of the probability codebook are determined by the same method as the method described in the conventional example.

That is, first, the delay L and the gain β of the adaptive codebook 12 are determined by the following processing.

In the delay circuit 11, the input signal of the synthesizing filter 6 in the preceding sub-frame, that is, the driving sound source signal, is delayed by a pitch period to create an adaptive code vector.

For example, the assumed pitch period is 40 to 16
Assuming 7 samples, 12 samples with a delay of 40 to 167 samples
Eight types of signals are created as adaptive code vectors and stored in the adaptive codebook 12.

At this time, the switch 16 is in an open state, and each adaptive code vector is multiplied by the gain value varied by the multiplier 13, and then passed through the adder 17 to be input to the synthesis filter 6 as it is. .

The synthesizing filter 6 performs the synthesizing process using the linear prediction parameter α, and subtracts the synthesized vector from the subtractor 8
To send to. The subtracter 8 generates an error vector by subtracting the original speech vector from the synthesized vector, and sends the obtained error vector to the auditory weighting filter 9.

The perceptual weighting filter 9 performs weighting processing on the error vector in consideration of perceptual characteristics,
It is sent to the error evaluator 10.

The error evaluator 10 calculates the mean square of the error vector, searches for an adaptive code vector having the smallest mean square value, and sends the delay L and the gain β to the multiplexer 18. Thus, the delay L and the gain β of the adaptive codebook 12 are determined.

Subsequently, the index i and the gain γ of the probability codebook 14 are determined by the following processing.

The probability codebook 14 has the number of dimensions corresponding to the subframe length (that is, 40 dimensions in the above example).
Are stored in advance, for example, in the form of 512 types, and each is assigned an index. At this time, the switch 16 is in a closed state.

First, the optimum adaptive code vector determined by the above processing is multiplied by the optimum gain β in the multiplier 13, and then sent to the adder 17.

Next, each probability code vector is multiplied by the multiplier 15
The variable is multiplied by the gain value and input to the adder 17. The adder 17 adds the optimal adaptive code vector multiplied by the optimal gain β and each probability code vector,
The result is input to the synthesis filter 6.

Subsequent processing is performed in the same manner as the adaptive codebook parameter determination processing. That is, the synthesis filter 6 performs the synthesis process using the linear prediction parameter α, and sends the synthesized vector to the subtractor 8.

The subtracter 8 generates an error vector by subtracting the original speech vector and the synthetic vector, and sends the obtained error vector to the perceptual weighting filter 9.

The perceptual weighting filter 9 performs a weighting process on the error vector in consideration of the perceptual characteristic, and sends it to the error evaluator 10.

The error evaluator 10 calculates the mean square of the error vector, searches for the probability code vector having the smallest mean square value, and sends the index i and the gain γ to the multiplexer 18. Thus, the index i and the gain γ of the probability codebook 14 are determined.

The multiplexer 18 quantizes the linear prediction parameter α, the adaptive codebook delay L and gain β, the probability codebook index i and gain γ.
Are multiplexed and transmitted.

The decoding operation of the speech decoding apparatus corresponding to the speech coding apparatus as described above is the same as that described in the conventional example.

From the voice discriminator 2 to the multiplexer 1
The voice / non-voice discrimination result v / uv is output to 8 so that the v / uv information is also included in the encoding parameter to be transmitted, and the corresponding decoding device is provided with the same switch as this encoding device. With a control circuit and changeover switch,
By controlling the changeover switch based on the information of / uv, it is possible to configure a variable bit rate coding device / decoding device capable of coding with higher efficiency.

According to the first embodiment as described above, it is determined whether or not the input signal is a voice signal, and when the non-voice signal continues for a predetermined number of frames, the LPC analyzer is given a predetermined lead. By continuously outputting the linear prediction parameter in the frame, the distortion of the coded speech due to the switching of the linear prediction parameter in the non-voice signal is reduced, even if a non-voice signal such as background noise is mixed, It becomes a high-quality speech coder that can satisfactorily encode a speech signal.

FIG. 5 and FIG. 6 show the second embodiment of the present invention, and FIG. 5 is a block diagram showing the configuration of the speech coding apparatus. In the second embodiment, a description of the same parts as those in the first embodiment will be omitted, and only different points will be mainly described.

The speech coding apparatus of the second embodiment is almost the same as that shown in FIG. 1, but as shown in FIG. 5, the input terminal has the same function as that of the buffer memory 1. The first buffer memory 21 is connected.

The output end of the first buffer memory 21 is 3
The first output end is connected to the subtractor 8 via the subframe divider 7, and the second output end is connected to the second
The buffer control circuit 22 is connected to the buffer memory 23, and the third output end is the control means for controlling the second buffer memory 23 via the voice discriminator 2 which is the voice discrimination means.
It is connected to the.

The second buffer memory 23 is connected to the synthesis filter 6 via the LPC analyzer 5 which is a linear predictive analysis means.

The other parts are the same as in FIG.

Next, the signal flow in the configuration shown in FIG. 5 will be described.

From the input terminal, for example, 8 kHz (that is,
An original audio signal sampled at 1/8 ms per sample is input, and an audio signal having a predetermined frame interval (for example, 20 ms, that is, 160 samples) is stored in the first buffer memory 21.

The first buffer memory 21 sends the input signal to the subframe divider 7 and the voice discriminator 2 in frame units. The voice discriminator 2 discriminates whether the input signal of the frame is voice or non-voice by, for example, the method described in the first embodiment.

The subframe divider 7 inputs the input signal of the frame at a predetermined subframe interval (for example, 5 m).
s, that is, 40 samples). That is, four subframe signals from the first subframe to the fourth subframe are created from the input signal of one frame.

The LPC analyzer 5 performs a linear prediction analysis (LPC analysis) on the input signal, extracts a linear prediction parameter α representing a spectral characteristic, and sends it to the synthesis filter 6 and the multiplexer 18.

Next, the operation of the buffer control circuit 22 will be described with reference to FIG. FIG. 6 is a flowchart showing the operation of the speech coding apparatus.

When encoding is started, a variable i indicating the number of continuous non-voice frames is set to 0 (step S2).
1).

Next, it is determined whether the discrimination result by the voice discriminator 2 is voice (v) or non-voice (uv) (step S22).

If the determination result in step S22 is non-voice, buffering is performed in the second buffer memory 23 (step S23), the variable i is incremented by 1 (step S24), and the variable i is set to a predetermined number. R (eg 1
0) is smaller than 0 (step S25).

If the variable i is smaller than the predetermined number R (for example, 10) in step S25, the contents of the second buffer memory 23 are collectively LPC analyzed (step S26), and then the next frame is processed. On the other hand, when the variable i is equal to or more than the predetermined number R in step S25, the process goes to step S27 without performing step S26.

On the other hand, when the discrimination result in the above step S22 is voice, after resetting the variable i indicating the number of continuous non-voice frames to 0 (step S28), L
The LPC analysis is performed by the PC analyzer 5 to update the spectrum parameter (step S29). Then, the process proceeds to step S27 and waits for the processing of the next frame.

The delay L of the adaptive codebook 12, the gain β, the index i of the probability codebook, and the gain γ are
It is determined in the same manner as the method described in the above conventional example.

The multiplexer 18 multiplexes and transmits the quantized linear prediction parameter α, the adaptive codebook delay L, the gain β, the probability codebook index i, and the gain γ.

The decoding operation of the speech decoding apparatus corresponding to the speech coding apparatus as described above is the same as that described in the conventional example.

Also in this embodiment, the v / uv information may be included in the coding parameter to be transmitted, and the decoding device may have a structure corresponding thereto.

According to the second embodiment as described above, by using the buffer memory, as in the first embodiment described above, a high-quality voice encoding capable of properly encoding a voice signal can be performed. It becomes a device.

The voice discrimination methods in the voice discriminators of the first and second embodiments are described as an example, and are not limited to the above-mentioned means.

In the first and second embodiments, the code-driven linear predictive coding apparatus has been described as an example. However, the linear predictive parameter and the drive excitation signal corresponding to the linear predictive residual signal are used. As long as the encoding device is expressed by parameters, it can be applied to any of them.

[Additional Notes] According to the above-described embodiment of the present invention as described in detail above, the following configuration can be obtained.

(1) A voice discriminating means for discriminating whether the input signal divided into a predetermined frame interval is a voice signal or a non-voice signal, and a linear prediction analysis means for outputting a spectrum parameter of the input signal. When the result of the discrimination by the speech discriminating means is a non-speech signal continuously for a predetermined number of frames, the spectral parameter in the predetermined preceding frame is continuously output to the linear prediction analysis means as the spectral parameter of the input signal. Control means, a driving sound source signal generating means for generating a driving sound source signal corresponding to the linear prediction residual signal, and a synthesis filter for synthesizing voice from the driving sound source signal based on the spectral parameter. Characteristic speech encoding device.

(2) A voice discriminating means for discriminating whether the input signal divided into a predetermined frame interval is a voice signal or a non-voice signal, and a linear predictive analyzing means for outputting the spectrum parameter of the input signal. When the discrimination result by the voice discriminating means is a non-voice signal, the input signal is buffered until the next voice signal is discriminated within a range not exceeding a predetermined number of frames, and the linear predictive analyzing means is provided. Control means for collectively performing linear prediction analysis of the input signal, driving source signal generation means for generating a driving source signal corresponding to the linear prediction residual signal, and voice from the driving source signal based on the spectral parameter. A speech coding apparatus comprising: a synthesizing filter for synthesizing.

(3) The driving sound source signal generating means comprises a delay circuit, an adaptive codebook, and a probability codebook, wherein (1) or (2) above.
3. The speech encoding device according to claim 1.

(4) The voice discriminating means calculates the frame energy of the input signal divided into the frame intervals, and the initial stage of calculating the frame energy of the input signal when the encoding is started. Frame energy analysis means, threshold value determination means for determining a threshold value according to the energy level of the non-voice signal, frame energy calculated by the frame energy analysis means, and initial frame calculated by the initial frame energy analysis means If the result of subtraction is greater than the threshold value by comparing the result of the subtraction by the adder and the threshold value determined by the threshold value determining means with each other, the adder that performs the subtraction by adding the energy with the signs opposite to each other is added. If the frame input signal is in the voice section, The speech coding apparatus according to (1) or (2) above, further comprising: a discriminating unit that discriminates a non-speech section.

According to the invention described in (1) above, a speech coding apparatus with good sound quality can be obtained which can be well coded even if a non-speech signal is input.

According to the invention described in (2) above, the same effect as that of the invention described in (1) above can be obtained by buffering the input signal until it is determined that the audio signal is the next audio signal. You can

According to the invention described in (3) above, the same effect as that of the invention described in (1) or (2) above can be obtained, and the CELP (Code Excited Linear Predictive Codi
ng) method, it is possible to obtain better sound quality.

According to the invention described in (4) above, the same effect as that of the invention described in (1) or (2) above can be obtained, and according to the magnitude of the energy of the non-voice signal,
It is possible to favorably determine whether the input signal is a voice signal or a non-voice signal.

[0122]

As described above, according to the invention as set forth in claim 1, it becomes a speech coding apparatus with good sound quality which can be well coded even if a non-speech signal is inputted.

According to the second aspect of the invention, the same effect as that of the first aspect of the invention can be obtained by buffering the input signal until it is determined that it is the next audio signal. You can

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a speech encoding apparatus according to a first embodiment of the present invention.

FIG. 2 is a block diagram showing a configuration of a voice discriminator in the voice encoding device according to the first embodiment.

FIG. 3 is a diagram showing an example of a relationship between a threshold value determined by a threshold value determination circuit of the voice discriminator and background noise energy in the first embodiment.

FIG. 4 is a flowchart showing the operation of the speech encoding apparatus according to the first embodiment.

FIG. 5 is a block diagram showing a configuration of a speech encoding apparatus according to a second embodiment of the present invention.

FIG. 6 is a flowchart showing the operation of the speech encoding apparatus according to the second embodiment.

FIG. 7 is a block diagram showing a configuration of a conventional speech encoding device.

8 is a block diagram showing a configuration of a speech decoding apparatus corresponding to the speech encoding apparatus of FIG. 7.

[Explanation of symbols]

2 ... Voice discriminator (voice discriminating means) 3 ... Switch control circuit (control means) 5 ... LPC analyzer (linear prediction analyzing means) 6 ... Synthesis filter 10 ... Error evaluator 11 ... Delay circuit (driving source signal generating means) Part: 12 ... Adaptive codebook (part of driving sound source signal generating means) 14 ... Stochastic codebook (part of driving sound source signal generating means) 18 ... Multiplexer 22 ... Buffer control circuit (control means) α ... Linear prediction parameter (Spectral parameter)

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification number Office reference number FI technical display location H03M 7/42 9382-5K H03M 7/42

Claims (2)

[Claims] 1. A voice discriminating means for discriminating whether an input signal divided into a predetermined frame interval is a voice signal or a non-voice signal, a linear prediction analysis means for outputting a spectrum parameter of the input signal, and When the discrimination result by the voice discriminating means is a non-voice signal continuously for a predetermined number of frames, the linear predictive analyzing means continuously outputs the spectrum parameter in a predetermined preceding frame as the spectrum parameter of the input signal. Control means, drive source signal generation means for generating a drive source signal corresponding to the linear prediction residual signal, and a synthesis filter for synthesizing voice from the drive source signal based on the spectral parameter. Speech coding device. 2. A voice discriminating means for discriminating whether an input signal divided into predetermined frame intervals is a voice signal or a non-voice signal, a linear prediction analysis means for outputting a spectrum parameter of the input signal, and If the discrimination result by the voice discriminating means is a non-voice signal, the input signal is buffered until it is discriminated as the next voice signal in a range not exceeding the predetermined number of frames, and then the linear predictive analyzing means is provided. Control means for collectively performing linear prediction analysis of the input signals, driving sound source signal generation means for generating a driving sound source signal corresponding to the linear prediction residual signal, and speech synthesis from the driving sound source signal based on the spectrum parameter. A speech coding apparatus, comprising: