JPH10124097A - Voice recording and reproducing device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain high quality sounds without increasing the amount of computations in a coding process by providing a discriminating means which discriminates input signal, a coding means which codes the signals and a coded data smoothing means which smooth the data obtained by coding non-voice signals. SOLUTION: A coding/combining section has a voice/non-voice discrimination section 17 which discriminates inputted signals to voice signals and non-voice signals employing the frames made by diving digital input signals into a constant length as a unit, a multipulse coding section 19 and a non-voice coding section 20. A frame energy computing section 16 is connected to the input terminal of a coding selection switching switch 18 through the section 17. A first output terminal 'a' of the switch 18 is connected to the section 19 and a second output terminal 'b' is connected to the section 20. The section 20 also acts as a coded data smoothing means.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an audio recording / reproducing apparatus, and more particularly, to an audio recording / reproducing apparatus that performs digital information compression processing on an audio signal for recording and reproducing.

[0002]

2. Description of the Related Art In recent years, an audio signal obtained by a microphone or the like is converted into a digital signal, which is recorded in, for example, a semiconductor memory. At the time of reproduction, the audio signal is read from the semiconductor memory and converted into an analog signal. A digital information recording / reproducing apparatus called a so-called digital recorder which outputs sound as a sound by a speaker or the like has been developed. Japanese Patent Application Laid-Open No. 63-259700 discloses a digital information recording / reproducing apparatus as described above.

[0003] In a storage / reproduction device such as the digital information recording / reproduction device described above, in order to save the amount of data recorded in a semiconductor memory, a highly efficient encoding is performed on a digitized audio signal. Technical means for minimizing the amount of generated data have been proposed.
In particular, in recent years, various audio coding techniques have been developed with the development of digital signal processing techniques, and the recordable time has been dramatically increased. Also, in non-voice sections and unvoiced sections, many variable-rate codings for performing more efficient coding have been proposed.

[0004] As this highly efficient coding method, a code excitation linear prediction coding method (CELP: Code Exci) is used.
ted Linear Predictive Cod
ing), an analysis-synthesis-type speech coding method represented by AD
There are known coding methods such as a waveform coding type compression method such as PCM.

[0005]

However, in the variable rate coding as described above, in a non-voice section or an unvoiced section, good coding quality cannot be obtained because the coding bit rate is further reduced. In particular, there is a problem that the sound quality is rapidly deteriorated when background noise or the like is mixed. Further, the above-described speech coding technology means has a problem that the amount of calculation is large.

In order to provide the audio recording / reproducing apparatus at a lower cost, encoding and decoding are performed by a fixed-point DSP (Digital Signal Pro).
However, in the present situation, the DSP often uses up the computing power of the encoding, and if further processing is added in order to improve the encoding performance, real-time processing cannot be achieved. Was.

SUMMARY OF THE INVENTION The present invention has been made in view of the above problems, and has as its object to provide an audio recording / reproducing apparatus capable of obtaining good sound quality without increasing the amount of calculation in encoding processing.

[0008]

To achieve the above object, a first audio recording / reproducing apparatus according to the present invention comprises: a discriminating means for discriminating an input signal into an audio signal and a non-audio signal; Encoding means for encoding, encoded data smoothing means for smoothing the data obtained by encoding the non-voice signal,
Is provided.

In order to achieve the above object, a second aspect of the present invention is provided.
The audio recording / reproducing apparatus has a discriminating means for discriminating the input signal into a speech signal and a non-speech signal in units of a frame obtained by dividing the digitized input signal into a fixed length, and a non-speech sound source estimating unit. A linear prediction encoding unit that encodes the input signal, and encoded data smoothing unit that smoothes at least one of gain information and linear prediction parameter information of a signal from the non-speech sound source estimation unit. I do.

[0010] In order to achieve the above object, a third aspect of the present invention is provided.
The audio recording / reproducing apparatus according to the first or second audio recording / reproducing apparatus, wherein the encoded data smoothing means smoothes the data obtained by encoding the non-audio signal only when the non-audio frame continues for a predetermined number or more. It is characterized in that

In the first audio recording / reproducing apparatus, the discriminating means discriminates the input signal into a speech signal and a non-speech signal, and the encoding means encodes the input signal. The encoded data smoothing means smoothes the data obtained by encoding the non-voice signal.

In the second audio recording / reproducing apparatus, the discriminating means discriminates the input signal into a speech signal and a non-speech signal in units of frames obtained by dividing the digitized input signal into fixed lengths, The input signal is encoded by predictive encoding means. Further, the coded data smoothing means smoothes at least one of the gain information and the linear prediction parameter information of the signal from the non-speech sound source estimation section in the linear prediction coding means.

[0013] The third audio recording / reproducing apparatus is provided with the first audio recording / reproducing apparatus.
Alternatively, in the second audio recording / reproducing apparatus, the encoded data smoothing means smoothes data obtained by encoding a non-audio signal only when a predetermined number or more of non-audio frames are continuous.

[0014]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing the overall configuration of an audio recording / reproducing apparatus according to one embodiment of the present invention.

In FIG. 1, a microphone 1 is connected to one end of an encoding / decoding unit 5 via a microphone amplifier 2, a low-pass filter 3, and an A / D converter 4. The other end of the encoding / decoding unit 5 is connected to an audio memory 7 via a memory control unit 6.

The speaker 8 is connected to a D / A converter 11 via a power amplifier 9 and a low-pass filter 10, and the D / A converter 11 is connected to one end of the encoding / decoding unit 5. .

Further, a system control unit 12 for controlling the operation of each unit includes an encoding / decoding unit 5, a memory control unit 6,
In addition to the voice memory 7, an operation input unit 13 including operation switches such as recording, playback, and stop, an address counter 14,
It is connected to the display unit 15.

FIG. 2 is a block diagram showing the configuration of the encoding unit in the encoding / decoding unit 5.

The encoding / decoding unit 5 discriminates the input signal into a speech signal and a non-speech signal in units of frames obtained by dividing the digitized input signal into fixed lengths (speech / non-speech signal). A discriminating unit 17), a speech sound source estimating unit (multi-pulse encoding unit 19), and a non-speech sound source estimating unit (non-speech encoding unit 20), and serve as a linear predictive encoding means for encoding the input signal. Fulfill.

In FIG. 2, a frame energy calculation unit 16 is connected to an input terminal c of a coding selection switch 18 as a coding selection unit via a voice / non-voice determination unit 17 as a voice / non-voice determination unit. Have been. The first output terminal a of the encoding changeover switch 18 is connected to the multi-pulse encoding unit 19, and the second output terminal b is connected to the non-speech encoding unit 20.

The configuration of the multi-pulse encoder 19 is as follows.
For example, since it is described in detail in Japanese Patent Publication No. 4-25560, description thereof is omitted here.

FIG. 3 is a block diagram showing the configuration of the non-speech encoding section 20.

The non-speech coding unit 20 serves as a coded data smoothing means. In FIG. 3, the linear prediction analysis unit 21 sends a signal to the gain calculation unit 24 via the energy calculation unit 22. Connected to the multiplexer 25. The random signal generator 23 is connected to the gain calculator 24. The gain calculator 24 is connected to the multiplexer 25.

FIG. 4 is a block diagram showing the configuration of the decoding unit in the encoding / decoding unit 5.

In the figure, a voice / non-voice discriminating section 26 is connected to an input terminal c of a decoding changeover switch 27, and a first output terminal a of the decoding changeover switch 27 is connected to a multi-pulse decoding section 28. The second output terminal b is connected to the non-speech decoding unit 29.

FIG. 5 is a block diagram showing the configuration of the non-speech decoding unit 29. However, the same reference numerals are given to the same components in the configuration diagrams shown in FIGS.

In FIG. 1, a demultiplexer 30 is connected to a gain multiplying unit 31 and a linear prediction synthesizing unit 32. Further, the random signal generation unit 23 includes the gain multiplication unit 3
1 is connected to the linear prediction synthesis unit 32.

Next, the recording / reproducing operation of the audio recording / reproducing apparatus having the above configuration will be described.

In FIG. 1, when an operator performs a recording operation via the operation input unit 13, an audio analog signal input from the microphone 1 is amplified by the microphone amplifier 2, and the low-pass filter 3 outputs the audio analog signal. Unnecessary high frequency components are cut off. An output signal from the low-pass filter 3 is converted into a digital signal by the A / D converter 4. At this time, the system control unit 12 selects and operates the encoding unit in the encoding / decoding unit 5, and encodes the digital signal from the A / D converter 4. The encoded data obtained by this encoding is stored in the audio memory 7 via the memory control unit 6.

When an operator performs a reproducing operation via the operation input unit 13, encoded data is read from the audio memory 7 and supplied to the encoding / decoding unit 5 via the memory control unit 6. Is done. At this time, the system control unit 12
The decoding unit in the encoding / decoding unit 5 is selected and operated, and the encoded data is encoded, so that decoded data is created. Since the decoded data is a digital signal, the D / A converter 11 converts the decoded data into an analog signal. Further, in the low-pass filter 10, unnecessary high-frequency components among frequency components included in the analog signal output from the D / A converter 11 are cut off. Then, the analog signal output from the low-pass filter 10 is amplified by the power amplifier 9, and a reproduction signal is output from the speaker 8.

In the above series of operations, the memory control unit 6 controls the input / output operation of signals between the audio memory 7 and the encoding / decoding unit 5. The address counter 14
Performs a count operation in accordance with address data provided from the system control unit 12 and specifies an address to the audio memory 7.

Next, the encoding operation of the encoding unit in the encoding / decoding unit 5 will be described with reference to the flowchart shown in FIG.

When the encoding / decoding operation is started in the encoding / decoding unit 5 under the control of the system control unit 12,
First, a noise counter NoiseCnt for determining whether an input signal is a voice signal or a non-voice signal is initialized, and NoiseCnt is set to 0 (step S1). Thereafter, when a frame sound is read by the encoding unit of the encoding / decoding unit 5 (step S2), the frame energy is calculated by the frame energy calculation unit 16 (see FIG. 2).
Eng is calculated.

Thereafter, the voice / non-voice discriminating section 17 (see FIG. 2) compares the frame energy Eng with a predetermined threshold value NoiseLev (step S4). Noise counter NoiseC
nt is set to NoiseCnt = 0 again (step S5), and the process proceeds to step S7. If the frame energy Eng is smaller than the threshold NoiseLev, the noise counter
NoiseCnt is incremented (step S6), and the process proceeds to step S7.

In step S7, the voice /
The non-voice discriminating unit 17 determines whether or not the noise counter NoiseCnt is greater than “2”. Here, NoiseCnt ≦
If it is 2, it is determined that the input signal is an audio signal, and the audio signal determination flag SP is set to 1 (step S8).
Thereafter, the encoding changeover switch 18 selects the multi-pulse encoding unit 19 based on the audio signal determination flag SP = 1, and the multi-pulse encoding unit 19 performs multi-pulse encoding (step S9). .

On the other hand, in step S7, NoiseC
If it is determined that nt> 2, the input signal is determined to be a non-voice signal, and the voice signal determination flag SP is set to 0 (step S11). Then, the encoding switch 18 is
The non-speech encoding unit 20 is selected based on the speech signal determination flag SP = 0, and non-speech encoding is performed in the non-speech encoding unit 20 (step S12).

Here, the non-speech coding will be described with reference to FIG.

The linear prediction analysis unit 21 performs a linear prediction analysis on the signal input in frame units, sends out the obtained linear prediction parameters to the multiplexer 25, and outputs the linear prediction residual signal to the energy calculation unit 22. To send to. The energy calculation unit 22 calculates the energy Er of the input linear prediction residual signal according to the following equation 1, and sends it to the gain calculation unit 24.

[0040]

(Equation 1) Where r (n) is the linear prediction residual signal at sample n,
N indicates the frame length.

The random signal generator 23 generates a random signal as a non-voice sound source signal and sends it to the gain calculator 24. The gain calculator 24 calculates the gain of the random signal rand (n) as the non-speech sound source signal created by the random signal generator 23. The gain is determined as follows.

First, a value of a gain candidate value ge that minimizes an error eer between a product of a plurality of gain candidate values ge and the energy of the random signal rand (n) and the energy Er of the linear prediction residual signal. Ask for. That is, the value of the gain candidate value ge that minimizes the error eer obtained by the following Expression 2 is obtained.

[0043]

(Equation 2) In other words, Expression 2 searches for the gain of a random signal such that the energy of the generated sound source signal becomes equal to the energy of the linear prediction residual signal. Then, the value of the gain candidate value ge that minimizes the error eer is sent to the multiplexer 25 as g.

This multiplexer 25 collectively outputs the received linear prediction parameters and g and outputs them as encoded data.

Returning to FIG. 6, after step S9 or step S12, the audio signal determination flag SP and each encoded data are recorded in the audio memory 7 (step S9).
10), and return to step S2.

Next, the operation relating to the encoded data smoothing will be described with reference to the flowchart shown in FIG.

In the present embodiment, the above-described multi-pulse coding is coded with 191 bits per frame, non-voice coding is coded with 81 bits per frame, and the voice signal determination flag SP is coded with 1 bit per frame. Shall be.

First, a counter Cn indicating the current number of frames
At time t, the audio signal determination flag Prev_SP is initialized (step S13), and Cnt = 0 and Prev_SP = 1, respectively.
Thereafter, the value of 1-bit data corresponding to the audio signal determination flag SP is read from the audio memory 7 (step S14). Then, it is determined whether or not the value of the one-bit data is the audio signal determination flag SP = 1 (step S15).

If it is determined in step S15 that the audio signal determination flag SP is not 1, the audio memory 7 stores 31bi.
The reading for t is performed (step S18), and the gain g of this frame is quantized and stored as G [Cnt] = g (step S19). Thereafter, the counter Cnt is incremented (step S20), and the audio signal determination flag is further increased.
Prev_SP = 0 (step S21), and the process returns to step S14.

In step S15, when the audio signal determination flag SP = 1, the audio memory 7
Minute (step S16), the audio signal determination flag
It is determined whether or not Prev_SP is “0” (step S
30). If the audio signal determination flag is not Prev_SP = 0, the process returns to step S14 and waits until Prev_SP = 0.

In step S30, Prev_SP =
0, the quantized gain G [i], ｛i = 0,1,
.., Cnt} are smoothed and rewritten in the voice memory 7 (step S17). Thereafter, the Cnt Prev_SP is initialized again (step S31), and the process returns to step S14.

In the present embodiment, the average value of the gain G of each of the 10 frames before and after the non-voice frame is set as the smoothed gain G 'of that frame. Then, the moving average is sequentially calculated for each frame, and the previous gain G is replaced with the smoothed gain G '.

In addition, the maximum value of the gain G is set to a predetermined value a or less. That is, G [i] = G [i]... G <a G[i]=a... G> a or over a continuous section of the gain G of the non-voice frame. Find the average and calculate each gain G
May be replaced.

The above-mentioned smoothing can be performed at any time after the end of the recording and before the reproduction, but in the present embodiment, the smoothing is automatically performed after the end of the recording as a predetermined timing. Is set as follows. Note that, other than this timing, as a predetermined timing, (1) if smoothing is not performed prior to the start of reproduction, smoothing is automatically started. (2) Start up at a preset time. (3) Automatically executed when auto power off (auto power off). (4) The user intentionally starts up. And so on.

In order to prevent the smoothing from being performed twice, a flag indicating whether or not the smoothing is performed is provided, and the smoothing is performed only when the smoothing is not performed. This improves efficiency.

Further, when smoothing is performed so as not to generate an inconsistent portion in recorded audio data, keys, switches, etc. of an operation unit or the like in the audio recording / reproducing apparatus are locked. By providing the mechanism, it is possible to prevent occurrence of a mismatched portion and malfunction.

As described above, since the gain parameter in the non-speech section is smoothed by the coded data smoothing means of the non-speech coding section 20, continuity with the preceding and succeeding frames is improved, and the audibility is improved. Natural reproduction sound can be obtained. Further, since the encoded data smoothing processing by the encoded data smoothing means is performed on encoded data recorded on a recording medium, it is possible to improve sound quality without increasing the amount of calculation at the time of encoding. Can be.

In the present embodiment, only the gain parameter is smoothed in the above-mentioned coded data smoothing means. However, it is also possible to smooth the spectrum parameter in the non-voice section. .

Next, the decoding operation of the decoding unit in the encoding / decoding unit 5 will be described with reference to the flowchart shown in FIG.

When the decoding processing operation is started in the encoding / decoding section 5 under the control of the system control section 12 (see FIG. 1), first, the speech / non-speech discriminating section 26 (see FIG. 4) The value of the 1-bit data is read from the audio memory 7 (step S22), and it is determined whether the value of the 1-bit data is the audio signal determination flag SP = 1 (step S23).

If it is determined in step S23 that the audio signal determination flag SP is not 1, the audio memory 7 stores 31bi.
The non-speech decoding unit 29 is selected by the decoding changeover switch 27, and non-speech decoding processing is performed in the non-speech decoding unit 29 (step S28).

In FIG. 5, the demultiplexer 30 comprises
The encoded data is separated into a linear prediction parameter and a gain, and each is separated into a linear prediction synthesis unit 32 and a gain multiplication unit 31.
To send to. The random signal generator 23 generates a random signal equal to the frame length, and the gain multiplier 31 amplifies the random signal with the received gain value. Then, the linear prediction synthesis unit 32 performs linear prediction synthesis on the amplified random signal and outputs the result.

On the other hand, if the audio signal determination flag SP = 1 in step S23, 191 bits are read from the audio memory 7 (step S24), and the multi-pulse decoding unit 28 is selected by the decoding changeover switch 27. The multi-pulse decoding unit 28 performs a multi-pulse decoding process (step S25).

Thereafter, the output from the decoding unit in the encoding / decoding unit 5 is sent to the D / A converter 11 (step S26), and the process returns to step S22.

In the present embodiment, the encoding / decoding process in the voice section uses the multi-pulse system, but it is naturally possible to use a system such as the CELP system.

[Appendix] According to the embodiment of the present invention as described in detail above, the following configuration can be obtained. (1) discriminating means for discriminating an input signal into a speech signal and a non-speech signal; encoding means for encoding the input signal;
And a coded data smoothing means for smoothing data obtained by coding the non-voice signal.

(2) a discriminating means for discriminating the input signal into a speech signal and a non-speech signal in units of a frame obtained by dividing the digitized input signal into a predetermined length, and a non-speech sound source estimating unit; Linear prediction encoding means for encoding the input signal; and encoded data smoothing means for smoothing at least one of gain information and linear prediction parameter information of a signal from the non-speech sound source estimating unit. An audio recording / reproducing apparatus characterized by the above-mentioned.

(3) The coded data smoothing means smoothes the data obtained by coding the non-voice signal only when the non-voice frame continues for a predetermined number or more (1) or (2). An audio recording / reproducing apparatus according to claim 1.

(4) The encoding means has a speech sound source estimating unit and a non-speech sound source estimating unit, and when a speech signal is input, the linear prediction parameter and the speech obtained by the speech sound source estimating unit. The sound source estimation parameter is encoded, and when a non-speech signal is input, the linear prediction parameter and the gain of a random signal representing the sound source signal obtained from the non-speech sound source estimator are encoded. 1),
The audio recording / reproducing device according to (2) or (3).

(5) Discriminating means for discriminating an input signal into a speech signal and a non-speech signal in units of a frame obtained by dividing a digitized input signal into a predetermined length, a speech sound source estimating unit and a non-speech sound source estimation Encoding a linear prediction parameter and a speech source estimation parameter obtained by the speech source estimation unit when a speech signal is input, and a linear prediction parameter and the non-speech source when a non-speech signal is input. Linear predictive encoding means for encoding the gain of a random signal representing the sound source signal obtained from the estimating unit; recording means for recording the discrimination result and the encoded data on a recording medium; Reading means for reading the encoded data, decoding means for decoding the voice signal and the non-voice signal based on the read determination result, and a frame for the non-voice signal, which is predetermined. An audio data recording / reproducing apparatus comprising: encoded data smoothing means for smoothing either or both of the linear prediction parameter and the gain in the continuous frame when the number of consecutive frames exceeds a predetermined number.

(6) The coded data smoothing means includes:
The audio recording / reproducing apparatus according to (1), (2), (3), (4) or (5), wherein the non-audio frame is smoothed at a predetermined timing.

(7) The coded data smoothing means includes:
The audio recording / reproducing apparatus according to (1), (2), (3), (4), (5) or (6), wherein smoothing is performed only when smoothing has not been performed.

(8) The operation input cannot be performed while the data obtained by encoding the non-speech signal is smoothed. (1), (2), (3), (4), (5) , (6)
Or the audio recording / reproducing device according to (7).

[0074]

As described above, according to the present invention, it is possible to provide an audio recording / reproducing apparatus capable of obtaining good sound quality without increasing the amount of calculation in the encoding process.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an overall configuration of an audio recording / reproducing apparatus according to an embodiment of the present invention.

FIG. 2 is a block diagram showing a configuration of an encoding unit in an encoding / decoding unit in the audio recording / reproducing apparatus of the embodiment.

FIG. 3 is a block diagram showing a configuration of a non-speech encoding unit in the encoding unit shown in FIG. 2 in the audio recording / reproducing apparatus of the embodiment.

FIG. 4 is a block diagram showing a configuration of a decoding unit in an encoding / decoding unit in the audio recording / reproducing apparatus of the embodiment.

FIG. 5 is a block diagram showing a configuration of a non-speech decoding unit in the decoding unit shown in FIG. 4 in the audio recording / reproducing apparatus of the embodiment.

FIG. 6 is a flowchart showing an encoding processing operation of an encoding unit in an encoding / decoding unit in the audio recording / reproducing apparatus of the embodiment.

FIG. 7 is a flowchart showing a non-speech part smoothing operation of the encoding part in the encoding / decoding part in the audio recording / reproducing apparatus of the embodiment.

FIG. 8 is a flowchart showing a decoding processing operation of a decoding unit in the encoding / decoding unit in the audio recording / reproducing apparatus of the embodiment.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Microphone 2 ... Microphone amplifier 3 ... Low-pass filter 4 ... A / D converter 5 ... Encoding / decoding part 6 ... Memory control part 7 ... Sound memory 8 ... Speaker 9 ... Power amplifier 10 ... Low-pass filter 11 ... D / A Converter 12 ... System control unit 13 ... Operation input unit 16 ... Frame energy calculation unit 17 ... Speech / non-speech discriminating unit 18 ... Coding changeover switch 19 ... Multi-pulse coding unit 20 ... Non-speech coding unit 26 ... Speech / Non-speech discriminator 27 ... Decoding switch 28 ... Multi-pulse decoder 29 ... Non-speech decoder

Claims (3)

[Claims] 1. A discriminating means for discriminating an input signal into a voice signal and a non-voice signal; a coding means for coding the input signal; and coded data for smoothing data obtained by coding the non-voice signal. An audio recording and reproducing apparatus, comprising: a smoothing unit. 2. A non-speech sound source estimating unit comprising: a discrimination means for discriminating the input signal into a speech signal and a non-speech signal in units of a frame obtained by dividing the digitized input signal into a predetermined length; Linear prediction encoding means for encoding an input signal; and encoded data smoothing means for smoothing at least one of gain information and linear prediction parameter information of a signal from the non-speech sound source estimating unit. An audio recording / reproducing apparatus characterized by the above-mentioned. 3. The encoded data smoothing means smoothes data obtained by encoding a non-speech signal only when a non-speech frame continues for a predetermined number or more.
Alternatively, the audio recording / reproducing device according to claim 2.