JP3162945B2 - Video tape recorder - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a video tape recorder (V) equipped with a speech speed converter for changing the speech speed of an audio signal.
TR)

[0002]

2. Description of the Related Art A VTR capable of listening to a sound at a normal speed even at a double speed has been commercialized, and its basic configuration is described in, for example, a magazine "Electronics" 1993.
April, pp. 34-37.

[0003]

However, the above-mentioned VTR is a method in which sound can be heard at a normal speed during double-speed playback, but audio information is unconditionally pulled for half an hour and deleted. However, there has been a problem that it is not possible to grasp the contents of the recorded tape from the reproduced audio, and there is a large time lag between the video and the audio.

Accordingly, the present invention can reduce the gap between video and audio during double-speed playback, and also performs speech speed conversion so as to limit the audio speed so that audio information can be obtained as much as possible. For example, a video tape recorder which reproduces audio by the same method as conventional double-speed reproduction when reproducing the audio speed at the speed of fast forward and rewind reproduction (5 times and 9 times). It is.

[0005]

According to the present invention, at the time of double-speed reproduction, compression / expansion processing or deletion processing is performed on an input audio signal depending on whether the reproduced audio signal is an audio section or a silent section. In the double-speed audio reproduction mode in which the speech speed conversion is performed, and in the reproduction at ± N-times speed (N: a natural number of 3 or more), a process of thinning out the audio section of the reproduction audio signal in a predetermined period according to the reproduction double speed is performed. The video tape recorder is provided with control means for setting the N-times speed reproduction mode to be performed.

Further, in the above-mentioned video tape recorder, a speech speed conversion processing means for performing a speech speed conversion processing of an input audio signal, a ring memory in which an output of the speech speed conversion processing means is written in order to perform a double speed audio reproduction mode, and Means for reading data from the ring memory at a constant speed; and a speech speed conversion processing means for the input voice signal in accordance with whether the input voice signal is a voice section or a silent section and the amount of storage in the ring memory. It is characterized by having a speech speed conversion device provided with means for performing compression / decompression processing or deletion processing.

In the above-mentioned video tape recorder, A / D conversion means for sampling an input analog audio signal at a sampling frequency corresponding to a set reproduction speed magnification in order to perform a double speed audio reproduction mode,
A frame memory to which an audio signal output from the D conversion means is input, and a voice speed conversion processing means for performing a voice speed conversion process on the voice signal each time a required number of voice signals are input to the frame memory; A ring memory to which the output of the speech speed conversion processing means is written, a reading means for reading data from the ring memory at a constant speed, and a storage amount calculation for calculating a storage amount of the ring memory based on a write signal and a read signal of the ring memory Means, a speech speed conversion processing means, a section discriminating means for discriminating whether the input speech corresponding to the required number of speech signals inputted to the frame memory is a speech section or a silent section, and a section discriminating means. Signal processing means for performing compression / decompression processing or deletion processing on the required number of audio signals according to the output of the output and accumulation amount calculation means Characterized in that it has a speech speed converting device.

In the above-mentioned video tape recorder, a required number of frames are stored in a frame memory and a frame memory in which an input digital audio signal is written at a speed corresponding to a set reproduction speed magnification in order to perform a double speed audio reproduction mode. Each time a voice signal is input, the voice speed conversion processing means for performing voice speed conversion processing on those voice signals, a ring memory in which the output of the voice speed conversion processing means is written, and a frame memory at the time of 1 × speed reproduction Reading means for reading data from the ring memory based on a read signal having a frequency equal to the writing speed of the memory, and storage amount calculating means for calculating the storage amount of the ring memory based on the write signal and the read signal of the ring memory. And the speech speed conversion processing means outputs the input sound corresponding to the required number of audio signals input to the frame memory. Performs a compression / expansion process or a deletion process on the required number of audio signals according to an output of the interval determination means and an output of the accumulation amount calculation means for determining whether the section is a voice section or a silent section. It is characterized by having a speech speed conversion device provided with signal processing means.

In the above video tape recorder, in order to perform the N-times playback mode, the memory is controlled so that the audio data is written into the memory at the N-times speed and the written data is read at the 1-times speed. I do.

[0010]

According to the present invention, at the time of double speed reproduction,
An adaptive speech speed conversion process is performed, and the VTR is controlled to perform a simple decimation process when the speed is three times or more.

According to the present invention, at the time of double-speed reproduction, compression / expansion processing or deletion processing is performed on an input audio signal depending on whether the input audio signal is an audio section or a silent section.

According to the present invention, at the time of double speed reproduction, the input voice signal is subjected to the speech speed conversion processing by the speech speed conversion processing means. The output of the speech speed conversion processing means is written to the ring memory. Data written to the ring memory is read at a constant speed. In the speech speed conversion processing means, compression / expansion processing or deletion processing is performed on the input voice signal according to whether the input voice signal is a voice section or a silent section and the amount of storage in the ring memory.

According to the present invention, at the time of double speed reproduction, the input analog audio signal is sampled by the A / D converter at a sampling frequency corresponding to the set reproduction speed magnification. The audio signal output from the A / D converter is input to the frame memory. Each time a required number of voice signals are input to the frame memory, the voice speed conversion processing means performs voice speed conversion processing on those voice signals. The output of the speech speed conversion processing means is written to the ring memory. The data written in the ring memory is read based on a read signal having a frequency equal to the sampling frequency at the time of 1 × speed reproduction. The storage amount of the ring memory is calculated by the storage amount calculation means based on the write signal and the read signal of the ring memory.

According to the present invention, at the time of double-speed reproduction, the section discriminating means determines whether the input voice corresponding to the required number of voice signals input to the frame memory is a voice section or a silent section. Then, compression / expansion processing or deletion processing is performed on the required number of audio signals in accordance with the output of the section determination means and the output of the accumulation amount calculation means.

According to the present invention, at the time of double speed reproduction, an input digital audio signal is written to the frame memory at a speed corresponding to the set reproduction speed magnification. Each time a required number of voice signals are input to the frame memory, the voice speed conversion processing means performs voice speed conversion processing on those voice signals. The output of the speech speed conversion processing means is written to the ring memory. Data written to the ring memory is read at a constant speed based on a read signal. The storage amount of the ring memory is calculated by the storage amount calculation means based on the write signal and the read signal of the ring memory.

In the speech speed conversion processing means, the input voice corresponding to the required number of voice signals input to the frame memory is determined by the section determining means as a voice section or a silent section. Then, compression / expansion processing or deletion processing is performed on the required number of audio signals in accordance with the output of the section determination means and the output of the accumulation amount calculation means.

According to the present invention, at the time of ± N times speed reproduction, a part of the sound is expanded and the remaining signal is thinned out.

[0018]

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

FIG. 23 is a schematic block diagram of a VTR embodying the present invention. A monaural audio signal reproduced from a horizontal track of a tape T is picked up by an audio head H and input to an equalizer amplifier 111. The audio signal equalized and amplified by the equalizer amplifier 111 is supplied to a terminal b of a switching circuit 115 and a speech speed conversion IC.
112. The output of the speech speed conversion IC 112 is supplied to a terminal a of the switching circuit 115. The switching circuit 115 selects the terminal a or b based on a command from the system microcomputer 114, and outputs the selected output via the mute circuit 116. That is, when the audio output from the tape T is output from the terminal b, an audio signal having a speed proportional to the tape speed is output, and from the terminal a, the reproduced audio signal is subjected to processing such as compression, decompression, and deletion processing. Signal is output. The speech speed conversion IC 112 is controlled by the system microcomputer 114 and performs the above processing in cooperation with a memory (dynamic RAM) 113.

Next, the operation of FIG. 23 will be described with reference to the flowchart of FIG.

First, the system microcomputer 114 sets the VT
R determines whether or not it is in the reproduction mode (S1). If it is determined in step 1 that the playback mode is set, then it is determined in step 2 (S2) whether the double speed playback key 117 has been pressed.
This is performed in 2). When the double speed reproduction key is pressed, the tape speed is doubled, and the video and the audio are reproduced. When the double speed reproduction key 117 is depressed with respect to the voice, the double speed adaptive speech speed process (S3).
Is performed.

This adaptive speech speed processing will be described with reference to FIG. 25. As shown in FIG. 25 (a), a speech sound of "VTR using speech speed conversion" is reproduced at time T as shown in FIG. If this is applied to the double-speed adaptive speech rate processing, the time is 1/2 (T / 2) as shown in FIG.
, But for this purpose the silence between the words of the normal-speed speech, i.e., the part that is not speaking, is deleted, and the words are reproduced at the same speed as the normal-speed speech. Connected. At this time, there is no complete silence due to ambient noise such as noises, but the surrounding noise is distinguished from the target voice by adaptively changing the threshold for silence determination according to the voice situation. ing. If the playback time is not reduced by half just by removing silence,
The reproduction time is reduced to half by adaptively compressing a part of the voice according to the length of the silent section. Then, when it is determined in step 6 (S6) that the reproduction has ended, the reproduction ends.

If the double speed reproduction key has not been depressed in step 2 (S2), then it is determined in step 4 (S4) whether or not the. ± .N speed mode has been set. The ± N double speed mode corresponds to, for example, fast forward reproduction or rewind reproduction. If it is determined in step 4 that the mode is not the ± N speed mode, the process returns to step 6.

If the mode is ± N times speed mode in step 4,
The compression process by the simple thinning process is performed (S5), and the process proceeds to step 6.

The simple decimation process will be described with reference to FIG. 26. For example, in the case of a reproduced audio signal at a 3 × speed, as shown in FIG. When a simple thinning process is performed on the reproduced signal, the "video signal" in the T2 period is expanded and reproduced in the T1 time period, as shown in FIG. You. That is, 3
The audio signal of the double speed reproduction is converted to a standard speech speed by thinning out 2/3 (thinning period: TD). In other words, T
1: T2 = 3: 1.

Similarly, when the input signal is reproduced at 5 × speed, 4/5 is thinned out and converted to the standard speech speed (T1: T2 =
5: 1), when the input signal is reproduced at 9 × speed, 8/9 is thinned out and converted to the standard speech speed (T1: T2 = 9: 1).

Referring to FIG. 5, a simple decimation process at the time of reverse N-times speed reproduction (at the time of -N times speed reproduction) will be described. As shown in FIG. Convert to speed. That is, FIG. 27 (a) shows an N-times reverse video signal, and FIG. 27 (b) shows a case where it is thinned out. (T1: T2 =
5: 1), in the case of 9 × speed reverse reproduction, 8/9 is thinned out and converted to the standard speech speed (T1: T2 = 9: 1). FIG.
At 7, TD indicates the period to be decimated.

As described above, the adaptive speech speed conversion process is performed at the time of double speed reproduction, and the simple thinning process is performed at the time of triple speed or higher. Even if the conversion is performed, the amount of voice to be deleted is increased, the signal processing is not much different from simple thinning, and there is no point in making an adaptive speech rate that requires complicated signal processing, and it is also difficult to hear. Because.
Therefore, by performing the adaptive speech speed conversion process at the time of double speed reproduction and performing the simple thinning process at the time of triple speed or more, the configuration as the speech speed conversion IC is simplified. In addition, the commercial value of the VTR can be improved.

FIG. 1 shows an overall configuration of a speech speed conversion device corresponding to a portion for performing adaptive speech speed conversion in the speech speed conversion IC.

After the input audio signal is amplified by the ALC amplifier 1, it is sent to the A / D converter 2 and converted into, for example, a 12-bit digital signal. The standard sampling frequency of the A / D converter 2 is, for example, 8 KHz. At the time of 2 × speed reproduction, the sampling frequency fs of the A / D converter 2
AD is 16 KHz.

The output of the A / D converter 2 is a DSP (Digital
l Signal Processor) 4 and also to the level detector 3. The level detection unit 3 includes the A / D conversion unit 2
When the A / D-converted data reaches the maximum value of the conversion range, the ALC (automatic level control) signal is output to A
Output to LC amplifier 1. Thereby, the ALC amplifier 1
Is controlled so that the input signal of the A / D converter 2 does not exceed the maximum range. That is, when the playback tape speed of the VTR changes, the input signal level of the ALC amplifier 1 also changes. Therefore, by automatically adjusting the amplifier gain based on the output of the level detector 3, the A / D
The input signal of the converter 2 is prevented from exceeding the maximum range.

The DSP 4 includes a frame memory 5 having a capacity capable of storing two frames of voice signals, and a voice speed conversion unit 6 for performing voice speed conversion processing on a frame basis for voice signals stored in the frame memory 5. ing. Here, it is assumed that one frame is composed of 200 pieces of sampling data.

Of the first half area and the second half area in the frame memory 5, one frame of the audio signal stored in one area is processed by the speech speed conversion unit 6, and at the same time, A The signal from the / D converter 2 is accumulated. When the signal for one frame is accumulated in the other area, the data in the area is processed by the speech speed conversion unit 6 and the data already processed is stored. The signal from the A / D conversion unit 2 is stored in the one of the areas.

The data output from the speech speed converter 6 is written to the ring memory 7 based on a write clock. The data written in the ring memory 7 is read based on a read clock. The signal read from the ring memory 7 is converted into an analog signal by the D / A converter 8, then amplified by the amplifier 10, and output as an audio output signal.

The sampling frequency fs of the D / A converter 8
DA is 8 KHz. The frequency of the read clock of the ring memory 7 is also 8 KHz. As the ring memory 7, a memory of 21845 × 12 bits, that is, a memory of 21845 words is used. Therefore, the maximum time during which data can be stored in the ring memory 7 (the maximum delay time of the output time with respect to the input signal) is 21
845 x 1/8000 = 2.73 seconds.

A write clock for the ring memory 7 is input to an up-counting input terminal (UP) of the up-down counter 9. A read clock for the ring memory 7 is input to a down-counting input terminal (DOWN) of the up-down counter 9. The up / down counter 9 calculates the difference between the total number of input write clocks and the total number of input read clocks (the ring memory 7).
Count) and count the count value to 15 bits
As a digital signal. The output of the up / down counter 9 is sent to the speech speed converter 6.

FIG. 2 shows a detailed configuration of the speech speed converter 6.

The audio signal read from the frame memory 5 is sent to the power calculator 11, and the average power value P of the audio signal for one frame is calculated. The average power value P represents the amplitude of each audio signal in one sampled frame as i0, i1,... IN-1 (where N = 2
00) is obtained by the following equation 1.

[0039]

(Equation 1)

The average power value P obtained by the power calculator 11 is sent to the comparator 12. The threshold value Th is sent from the threshold value memory 13 to the comparison unit 12, and the average power value P is equal to or larger than the threshold value Th (P ≧ Th) or the average power value P is smaller than the threshold value Th. (P <Th) is determined. The comparison unit 12 outputs a signal indicating that the current frame is a voice section when the average power value P is equal to or larger than the threshold Th (P ≧ Th). Is a signal indicating that is a silent section.

The threshold value Th is, for example, 2 when the number of quantization bits of the A / D converter 2 is 12 bits.
It is set to 12. Note that the threshold value T is calculated as follows.
h may be changed. That is, as shown by a dotted line in FIG. 2, a power steady state detection and threshold value updating unit 14 is provided. The power steady state detection and threshold update unit 14 calculates the average power value P from the power calculation unit 11.
Is determined over a predetermined number of frames (for example, 40 frames), and if it is constant (steady state), a value twice the average power value P at that time is stored in the threshold memory 13. Write and update the threshold value Th. However, the maximum value of the updated threshold value is limited to a predetermined value, for example, 214. By doing so, it is possible to treat the constantly occurring noise as a silent section.

Further, the voice section and the silent section of the input signal may be determined based on the cumulative power Pa of the voice signal of each frame and the given threshold value as shown in the following equation (2). .

[0043]

(Equation 2)

The output of the comparing section 12 is sent to the conditional branching section 15. The output of the ring memory accumulated amount state determination unit 16 is input to the conditional branch unit 15. Also, conditional branching unit 1
5 is supplied with an audio signal from the frame memory 5 via the power calculator 11. Furthermore, conditional branching unit 1
5, a pause continuation length setting memory 17 is connected. In the pause continuation length setting memory 17, a pause continuation length Tdel (a silence deletion start point discrimination value) for determining a deletion start point of a silent section is set.

Based on the amount of storage sent from the up / down counter 9, the ring memory storage state determination unit 16 determines that the state of the ring memory 7 has just reached the overflow state and that the state of the ring memory 7 is under. Detects the state immediately before the flow.

That is, the overflow detection data memory 18 stores overflow detection data Tmax, and the underflow detection data memory 19 stores underflow detection data Tmin.
The overflow detection data Tmax is set to, for example, a value 21645 smaller than the total number of words (TOTAL) 21845 of the ring memory 7 by 200. The underflow detection data Tmin is set to, for example, 200.

The accumulated amount sent from the up / down counter 9 is equal to the overflow detection data Tmax.
At this point, the ring memory storage amount state determination unit 16 outputs a detection signal immediately before overflow. When the accumulated amount sent from the up / down counter 9 becomes equal to or less than the underflow detection data Tmin, the ring memory accumulated amount state determination unit 16 outputs a detection signal immediately before the underflow. The conditional branch unit 15 determines that the ring memory 7 is in the state immediately before the overflow when the immediately before overflow detection signal is input, and the ring memory 7 is in the state immediately before the underflow when the immediately before underflow detection signal is input. Is determined.

The condition branching unit 15 determines whether a voice section or a silent section has been sent from the comparing unit 12, a detection signal regarding the ring memory state sent from the ring memory storage amount state determining unit 16, and a pause continuation. Length setting memory 1
Based on the pause duration Tdel set to 7, the following six cases are classified. And
In response, the multiplexer 20 is controlled to send the audio signal to a predetermined processing unit. (1) First Case (case 1) When it is determined that the input signal is in the voice section and the ring memory 7 is not in the state immediately before the overflow, the first case occurs.

In this case, the audio signal is sent to the pitch compression / expansion means 23 via the multiplexer 20.
The pitch compression / expansion means 23 performs variable speech control (VSC). Assuming that the reproduction speed magnification is n, the pitch compression / expansion means 23 performs the expansion / compression processing on the input signal at a compression ratio larger than 1 / n. As the decompression compression method used here, for example, an overlap addition method (Pointer Interval Control Overlap
and Add: PICOLA), TDHS (TimeDomain Ha
rmonic Scaling) method. Pitch expansion / compression means 23
The signal subjected to the decompression and compression processing is sent to the ring memory 7 via the demultiplexer 27, and is written to the ring memory 7 according to a write clock.

During double-speed playback of a VTR, the A / D
The sampling frequency fsAD of the converter 2 is 16 KHZ, and the sampling frequency fsDA of the D / A converter 8 is 8 KHZ. Therefore, the pitch is restored and output.

In the conventional general time-axis expansion compression, compression is performed at a compression rate of 1/2 at the time of double speed reproduction of a VTR. In other words, two pitch periods are thinned out to one pitch period. For this reason, the output sound is twice as fast as the standard sound speed. In other words, in the normal reproduction of the double speed reproduction, the output audio is twice the standard audio speed. However, the pitch remains unchanged.

On the other hand, the pitch expansion / compression means 23 provided in the speech speed converter 6 shown in FIG.
Set to a larger value. Here, the compression ratio is 2/3
Is set to In other words, three pitch periods are thinned out to two pitch periods. Therefore, the output audio is
This is 3/2 times the standard audio speed. Again, the pitch is
It remains as it is. As described above, when the data is compressed at the compression rate of 2/3, it is 2 / 3-1 as compared with the case of the compression rate of 1/2.
The signal will be expanded by / 2 = 1/6. This extension is the amount of storage in the ring memory 7.

A method of compressing an input signal at a compression ratio of 2/3 using PICOLA will be briefly described with reference to FIG. First, a pitch period is extracted from an input signal.
Let the extracted pitch period be Tp. The waveform A is weighted linearly from 1 to 0 (weight function K1), and a waveform A 'is created. A weight (weight function K2) from 0 to 1 is assigned to the waveform B,
A waveform B 'is created.

Then, the waveforms A ′ and B ′ are added to generate a waveform A ′ * B ′ having a length Tp.
These weights are added to maintain continuity at connection points before and after the waveform A '* B'. Next, the pointer is moved by 3Tp, which is a length determined based on the compression ratio, and the same operation is performed. Thus, two waveforms A ′ * B ′ and C are obtained from the three waveforms A, B, and C. In this way, the signal for three pitch periods becomes 2
The signal is compressed to a pitch period signal.

As a decompression method by the pitch decompression / compression means 23, as shown in FIGS. 17A and 17B, decompression processing is performed in units of a predetermined fixed frame length Ts without extracting a pitch. You may do so. The fixed frame length Ts is set to, for example, a length of 200 input data. FIG. 17 shows an example in which 3Ts is changed to 2Ts.

In the method of FIG. 17 (a), of the waveforms A, B and C having the fixed frame length Ts, the waveform A is given a weight (weight function K1) linearly going from 1 to 0. Thus, a waveform A ″ is created.
Weight (weight function K2) toward the waveform B "
Is created.

Then, those waveforms A "and B" are added to create a waveform A "* B" having a length Ts.
These weights are added to maintain continuity at connection points before and after the waveform A "* B". And the next waveform C
Is output as is. As a result, two waveforms A "* B" and C are obtained from the three waveforms A, B, and C. In this way, the signal for 3Ts is compressed into a signal for 2Ts.

In the method shown in FIG. 17 (b), the waveform A of the waveforms A to C having the fixed frame length Ts has a weight (weight function K3) from the top, for example, 20 data points linearly from 0 to 1. To obtain a waveform A ″.
A weight B (weight function K4) that linearly goes from 1 to 0 is applied to the first to 200th input data to obtain a waveform B ". Then, the waveform C is deleted. The following three waveforms D to F
, The same processing is performed. In this way,
A signal consisting of three waveforms AC (or DF) is 2
Compressed into a signal consisting of two waveforms A "and B" (or D "and E"). That is, the signal for 3Ts is 2
It is compressed to a signal of Ts.

When the decompression and compression processing in units of the fixed frame length is used, the sound quality is reduced, but the processing amount is reduced as compared with the case where the decompression and compression processing for each pitch cycle is used.

When this speech speed conversion device is applied to an English language learning device (at the time of 1 × speed reproduction), the sampling frequency fsAD of the A / D converter 2 is 8 KHZ, and
The sampling frequency fsDA of the / A conversion unit 8 is 8 KHZ
It is. In this case, the pitch compression / expansion means 23 expands the audio signal at a compression ratio of 3/2, for example, so that two pitch periods become three pitch periods. That is, the voice section is extended 1.5 times. Therefore, in this case,
3 / 2−1 = 1/2 with respect to the normal playback of 1 × speed playback
Only the signal is expanded, and the expanded amount becomes the storage amount of the ring memory 7. (2) Second case (case 2) When it is determined that the input signal is a voice section and the ring memory 7 is in a state immediately before overflow, the second case
It becomes a case.

In this case, the audio signal is sent to the input signal deleting section 21 via the multiplexer 20, and the audio signal is deleted. Specifically, the writing operation to the ring memory 7 is stopped until the count value of the up / down counter 9 becomes equal to or less than the underflow detection data Tmin, that is, until the ring memory 7 is in a state immediately before the underflow.

When the ring memory 7 is in the state immediately before the underflow, 200 or less, for example, 100 muffling signals (signals having a value “0”) are output from the muffling insertion unit 22, and the muffling signals are output from the demultiplexer 27. Is sent to the ring memory 7 via the. The reason why the mute signal is written in the ring memory 7 is to prevent a click sound from being generated at the end of the sound signal due to sound deletion. (3) Third Case (case 3) When it is determined that the input signal is a silent section, the duration of the silent section is less than the set pause duration Tdel, and the ring memory 7 is not in the state immediately before overflow. This is the third case.

In this case, the same processing as in the first case is performed. However, in the case of the third case, the expansion / compression processing may be performed at a compression ratio of 1 / n, where n is the reproduction speed magnification. That is, in the third case, the decompression and compression processing is performed at a compression ratio of 1 / n or more. (4) Fourth case (case 4) When it is determined that the input signal is a silent section, the duration of the silent section is less than the set pause duration Tdel, and the ring memory 7 is in a state immediately before overflow. , The fourth case.

In this case, the same processing as in the second case is performed. (5) Fifth Case (case 5) It is determined that the input signal is a silent section, the duration of the silent section is longer than the set pause duration Tdel, and the ring memory 7 is not in the state immediately before the underflow. Sometimes it is the fifth case.

In this case, the audio signal is sent to the input signal deleting section 25 via the multiplexer 20, and the audio signal is deleted. Specifically, the writing operation to the ring memory 7 is stopped. However, in order to prevent the start part (unvoiced section) of the voice section from being lost, and to prevent a click sound from being generated at a gap due to the deletion of the voice, the waveform synthesis / insertion unit 26 performs waveform synthesis. Insertion processing is performed.

The waveform synthesizing and inserting process performed by the waveform synthesizing and inserting unit 26 will be described with reference to FIGS. In the method according to FIG. 4A, the waveform combining and inserting unit 26
A memory 31 and a second memory 32 are provided. At the start of the input signal deletion process by the input signal deletion unit 25, a predetermined length T equal to or less than one frame length from the deletion start point.
s, for example, an input signal for one frame is stored in the first memory 31.
Are sequentially stored in address order. Next, the first memory 31
Is multiplied with the content A of the first memory 31 by a function K1 that linearly changes from 1 to 0 as the address of the first memory 31 increases. Then, the multiplication result A ′ is written into the first memory 31 again.

An input signal of a predetermined length Ts immediately before the end point of the input signal deletion section by the input signal deletion section 25 is
The data is sequentially stored in the second memory 32 in the order of addresses. next,
As the address of the second memory 32 increases, 0 to 1
Is multiplied by the content B of the second memory 32. Then, the multiplication result B ′ is written into the second memory 32 again. Thereafter, the first memory 31
Is added to the content B 'of the second memory 32 to obtain data A' * B 'having a predetermined length Ts.
Then, the obtained data A ′ * B ′ for the predetermined length Ts is sent to the ring memory 7 via the demultiplexer 27 and written into the ring memory 7.

In the method shown in FIG. 4B, input signals of a predetermined length Ts less than one frame length, for example, one frame, are sequentially stored in the first memory 31 in address order from the deletion start point. Next, a function K3 having a slope that changes linearly from 1 to 0 at the rear end is stored in the content A of the first memory 31.
Is multiplied by Then, the multiplication result A ′ is again the first
The data is written to the memory 31.

The input signal of a predetermined length Ts immediately before the end point of the input signal deletion section by the input signal deletion section 25 is
The data is sequentially stored in the second memory 32 in the order of addresses. next,
The content B of the second memory 32 is multiplied by a function K4 having a slope linearly changing from 0 to 1 at the front end. Then, the multiplication result B ′ is written into the second memory 32 again. Thereafter, the contents A 'of the first memory 31 and the contents B' of the second memory 32 are joined together to obtain data A '+ B' for 2Ts. And the obtained 2Ts
The minute data A ′ + B ′ is sent to the ring memory 7 via the demultiplexer 27 and written to the ring memory 7. FIG. 4B shows an example in which Ts is the length of one frame, but data having half the length of one frame may be used as Ts.

When the input signal deleting unit 25 repeatedly performs the process of deleting the audio signal in the silent section, the ring memory 7 may be in a state immediately before the underflow. In this case, the input signal for the predetermined length Ts is stored in the second memory 32 from the time when the ring memory 7 enters the state immediately before the underflow. Then, based on the data stored in the first memory 31 and the data stored in the second memory 32, the same waveform synthesis insertion processing as described above is performed. (6) Sixth case (case 6) It is determined that the input signal is a silent section, the duration of the silent section is longer than the set pause duration Tdel, and the ring memory 7 is in a state immediately before underflow. Sometimes this is the sixth case.

In this case, the input signal is sent to the thinning processing section 24 via the multiplexer 20. The thinning-out section 24 performs thinning-out processing so that the compression rate is 1 / n, where n is the reproduction speed magnification of the VTR. For example, at the time of double speed reproduction, the input signal is thinned at a compression rate of 1/2, and at the time of triple speed reproduction, the input signal is thinned at a compression rate of 1/3. During 1 × speed reproduction, the input signal is output as it is.

The following method is used as the thinning processing by the 1 / n thinning processing section 24. Here, 2
A description will be given of the case of double speed reproduction as an example.

The pitch of the input signal is extracted using the time axis compression method using PICOLA or TDHS described above,
The pitch data portion is thinned out so that the compression ratio becomes 1/2.

As shown in FIGS. 5A to 5C,
The waveform may be thinned every predetermined time Ts without performing the pitch extraction.

In the method shown in FIG. 5A, the waveforms B and D of the waveforms A to D are thinned out, and a signal composed of the waveforms A and C is obtained.

In the method of FIG. 5B, the waveforms B and D are thinned out of the waveforms A to D. The waveform A has a slope rising from 0 to 1 at the front end (function K4).
Is multiplied by a function having a slope (function K3) having a slope falling from 1 to 0 at the rear end to generate a waveform A ′. Further, the waveform C ′ is formed by multiplying the waveform C by a function having a slope (function K4) rising from 0 to 1 at the front end and a slope (function K3) falling from 1 to 0 at the rear end. Is done. In this way, a signal consisting of four waveforms A to D is compressed into a signal consisting of two waveforms A 'and C'.

In the method shown in FIG. 5C, the waveform A is weighted linearly from 1 to 0 (weight function K1), and a waveform A 'is created. A weight (weight function K2) from 0 to 1 is assigned to the waveform B,
A waveform B 'is created. Then, those waveforms A ′ and B ′ are added to create a waveform A ′ * B ′ having a length Ts.

Similarly, the waveform C is given a weight (function K1) linearly going from 1 to 0, and the waveform C ′
Is created. The waveform D 'is weighted from 0 to 1 (function K2) to generate a waveform D'. Then, those waveforms C ′ and D ′ are added to create a waveform C ′ * D ′ having a length Ts. In this way, a signal consisting of four waveforms A to D becomes two waveforms A '
It is compressed into a signal consisting of * B 'and C' * D '.

As described above, in the case corresponding to the sixth case, the thinning-out process is performed at a compression ratio of 1 / n, where n is the reproduction magnification of the VTR, but the compression ratio is controlled as follows. You may make it.

When the decimation process is performed at a compression ratio of 1 / n, the ratio f between the sampling frequency fsDA of the D / A converter 8 and the sampling frequency fsAD of the A / D converter 2 is obtained.
When sDA / fsAD is equal to the compression ratio 1 / n,
The storage amount of the ring memory 7 does not change. However, the calculation accuracy of the compression ratio 1 / n and the sampling frequency fs
Depending on the clock accuracy of AD and fsDA, fsDA /
It is possible that fsAD does not equal the compression ratio 1 / n.

When fsDA / fsAD becomes larger than the compression ratio 1 / n (fsDA / fsAD> 1 / n),
Assuming that fsDA / fsAD = 1 / a (a> 0), ｛(1
/ A)-(1 / n)}, the compression ratio decreases, the degree of thinning increases, the storage amount of the ring memory 7 decreases, and the storage amount of the ring memory 7 may underflow. .

On the other hand, fsDA / fsAD is equal to the compression ratio 1 /
n (fsDA / fsAD <1 /
In n), assuming that fsDA / fsAD = 1 / a (a> 0), the compression ratio increases by {(1 / n)-(1 / a)}, the degree of thinning decreases, and the ring memory 7 Is increasing.

Therefore, when performing the thinning process,
After confirming the storage amount of the ring memory 7, the compression ratio is controlled as follows. As fsDA / fsAD = 1 / a (a> 0), α that satisfies the condition of (1 / n) −α <1 / a <(1 / n) + α is selected. Here, α is a value of 0 or more and 1 or less, for example, a value in a range of 0.001 to 0.1.

When fsDA / fsAD becomes larger than the compression ratio 1 / n, that is, when the storage amount of the ring memory 7 decreases, the compression ratio is increased from 1 / n to ｛(1 /
n) + α｝. That is, the compression ratio is increased, and the amount of storage in the ring memory 7 is increased.

When fsDA / fsAD becomes smaller than the compression ratio 1 / n, that is, when the storage amount of the ring memory 7 increases, the compression ratio is increased from 1 / n to ｛(1 / n).
n) -α｝. That is, the compression ratio is reduced, and the amount of storage in the ring memory 7 is reduced.

In the above description, the compression ratio is changed based on the storage amount of the ring memory 7, but when the thinning process is performed, the compression ratio is set to {(1 / n) -α} or { (1 / n) + α｝ may be alternately changed.

FIGS. 6 and 7 show the processing procedure by the speech speed converter 6.

The processing performed by the speech speed conversion unit 6 in the case of double speed reproduction of a VTR will be described below. (1) Processing at Start of Reproduction When reproduction is started and the average power value P of the first frame is calculated by the power calculation unit 11 (step 1), the calculated average power value P is equal to or larger than the threshold Th. Whether it is or not is determined based on the output of the comparison unit 12 (step 2).

When the input speech signal starts from a silent section, in the first frame, the average power value P becomes smaller than the threshold value Th, and the process proceeds to step S11. Then, the duration of the silent section (the number of frames in which the silent section continues) is calculated, and it is determined whether or not the calculated duration is equal to or longer than the pause duration Tdel set in the pause duration memory 17 (step). 12). This pause continuation length T
“del” is set to, for example, a length of four frames in terms of the number of frames.

In the processing for the first frame,
Since the continuation length of the silent section is less than the pause continuation length Tdel, it is determined whether or not the ring memory 7 is in the state immediately before the underflow based on the output of the ring memory storage amount state determination unit 16 (steps 13 and 14). .

In the processing for the first frame,
Since the ring memory 7 is in a state immediately before the underflow, the frame data is thinned out by the thinning-out processing unit 24 at a compression ratio of 1/2 (step 28), and the compressed data after the thinning-out processing is written to the ring memory 7. . Thereafter, the process returns to step 1. (2) Description of the Process in the First Case In step 2, when it is determined that the average power value P is equal to or larger than the threshold Th, it is determined that the current frame is a voice section, and the process proceeds to step 3. In step 3,
Whether or not the previous frame was a deletion section is determined by the first flag F
1 is determined based on the state. If the previous frame is not a deletion section, it is determined whether or not the ring memory 7 is in the state immediately before overflow based on the output of the ring memory storage amount state determination unit 16 (steps 6 and 7). If the previous frame is a section to be deleted, after the processes of steps 4 and 5 are performed, it is determined whether or not the ring memory 7 is in a state immediately before overflow (steps 6 and 7). Steps 4 and 5 will be described later.

If it is determined in step 7 that the current state is not the state immediately before the overflow, the first case occurs, and the pitch compression / expansion means 23 compresses the current frame data on the time axis at a compression ratio of 2/3 ( Step 8). The compressed data is sent to and written to the ring memory 7. Thereafter, the process returns to step 1. (2) Description of the Second Case Process When it is determined in step 2 that the average power value P is equal to or greater than the threshold Th, the frame transmitted this time is determined to be a voice section, and step 3 is performed. Proceed to. In step 3, it is determined whether or not the previous frame is a deletion section based on the state of the first flag F1. If the previous frame is not the deletion section, it is determined whether or not the ring memory 7 is in the state immediately before the overflow based on the output of the ring memory storage amount state determination unit 16 (step 6,
7). If the previous frame is a section to be deleted, after the processes of steps 4 and 5 are performed, it is determined whether or not the ring memory 7 is in a state immediately before overflow (steps 6 and 7). Steps 4 and 5 will be described later.

If it is determined in step 7 that the state is just before the overflow, the second case occurs,
The input signal is deleted by the input signal deletion unit 21 until the underflow detection signal is output from the ring memory storage amount state determination unit 16 (step 9). That is, writing to the ring memory 7 is stopped until the ring memory 7 is in a state immediately before underflow.

When the ring memory 7 is in the state immediately before underflow, the silence insertion section 22 writes a predetermined number of silence signals "0" of 200 or less into the ring memory 7 (step 10). Then, the process returns to step 1.

Instead of the processing in step 10 described above, FIG.
The processing shown in FIG. 9A or FIG. 9B may be performed. The method shown in FIG. 9A will be described. From the time when it is determined in step 7 that the state is immediately before the overflow, for example, for the waveform A corresponding to 200 input signals, the weight (linear) going from 1 to 0 linearly ( Weight function K
1) is applied to obtain a waveform A '. In addition, for waveform B corresponding to 200 input signals from immediately before the underflow to 200 before the underflow, weights from 0 to 1 (weight function K2)
To obtain a waveform B ′.

Then, the obtained two waveforms A ′ and B ′ are added to form a waveform A ′ * of 200 lengths.
Create B '. Then, 200 signals for this waveform A ′ * B ′ are written in the ring memory 7. The detection 200 times before the immediately before the underflow is performed based on the count value of the up / down counter 9. As a result, it is possible to effectively prevent a click sound from being generated at a gap between audio signals before and after the audio deletion section.

The method shown in FIG. 9B will be described. When it is determined in step 7 that the state is immediately before the overflow, for example, the waveform A for 100 input signals is linearly changed from 1 to 0. A weight A (weight function K1) is applied to obtain a waveform A '. Also, for waveform B for 100 input signals from immediately before the underflow to 100 before, the weights from 0 to 1 (weight function K
By adding 2), a waveform B 'is obtained. Then, 200 signals obtained by combining the obtained two waveforms A ′ and B ′ are written in the ring memory 7.

In step 9 described above, if it is determined that the state is immediately before the overflow, the input signal is deleted by the input signal deletion unit 21 until the underflow detection signal is output from the ring memory storage state determination unit 16. However, the data stored in the ring memory 7 may be deleted so that the ring memory 7 is in a state immediately before the underflow.

More specifically, the write start address of the ring memory 7 is changed from the address (point C) in the state immediately before the overflow shown in FIG. 18A, as shown in FIG. Is jumped to the address (point A) where the state immediately before the underflow occurs. Therefore, in the process of step 9, the data stored at the addresses from the point A to the point C is deleted. Thereafter, as shown in FIG. 18C, after the mute signal is written in step 10, the input data is written.

If the data stored in the ring memory 7 is deleted in step 9 so that the ring memory 7 is in the state immediately before the underflow, as described above, the mute signal is written in the ring memory 7 in step 10. 19 (a) or FIG. 19 (b).

Now, as shown in FIG. 18B, the write start address of the ring memory 7 is changed from the address (point C) in the state immediately before the overflow shown in FIG. It is assumed that the user jumps to the address (point A) in the immediately preceding state. For data S stored from a point A to a predetermined number, for example, an address 200 points ahead (point B in FIG. 19A), FIG.
As shown in (a), a waveform S ′ is obtained by applying a weight (weight function K1) linearly going from 1 to 0. Further, as shown in FIG. 19A, a weight (weight function K2) from 0 to 1 is applied to 200 pieces of input data (waveform T) written to the ring memory 7 thereafter. ,
A waveform T 'is obtained.

Then, the obtained two waveforms S ′ and T ′ are added together, and a waveform S ′ * having a length of 200
Create T '. Then, 200 signals for this waveform S ′ * T ′ are written into the ring memory 7 from the point A. As a result, it is possible to effectively prevent a click sound from being generated at a gap between audio signals before and after the storage data deletion section.

The method shown in FIG. 19B will be described. Data stored from a point A in FIG. 18B to a predetermined number of addresses, for example, 100 addresses ahead (point B in FIG. 19B). S is weighted linearly from 1 to 0 (weight function K1) to obtain a waveform S '. Also,
Thereafter, a weight (weight function K2) from 0 to 1 is applied to 100 pieces of input data (waveform T) written to the ring memory 7 to obtain a waveform T '. Then, 200 signals obtained by combining the obtained two waveforms S ′ and T ′ are written into the ring memory 7 from the point A. (3) Description of the process as the third case When it is determined in step 2 that the average power value P is smaller than the threshold Th, the continuation length of the silent section up to this time is calculated (step 11) and calculated. Pause duration Tde whose duration is set in the pause duration memory 17
It is determined whether it is equal to or greater than 1 (step 12). And
If it is determined that the duration of the silent section is less than the pause duration Tdel, the ring memory storage amount state determination unit 1
Based on the output of No. 6, it is determined whether or not the state is immediately before underflow (steps 13 and 14).

When the ring memory 7 is not in the state immediately before the underflow, it is determined whether or not the state is immediately before the overflow based on the output of the ring memory storage amount state determination section 16 (steps 6 and 7). If the state is not the state immediately before the overflow, a third case occurs, and the pitch compression / expansion means 23 determines that the current frame data is 2/3.
(Step 8). The compressed data is sent to and written to the ring memory 7. Thereafter, the process returns to step 1. (4) Description of Processing in Fourth Case When it is determined in step 2 that the average power value P is smaller than the threshold value Th, the continuation length of the silent section up to this time is calculated (step 11). Pause duration Tde whose duration is set in the pause duration memory 17
It is determined whether it is equal to or greater than 1 (step 12). And
If it is determined that the duration of the silent section is less than the pause duration Tdel, the ring memory storage amount state determination unit 1
Based on the output of No. 6, it is determined whether or not the state is immediately before underflow (steps 13 and 14).

If the ring memory 7 is not in the state immediately before underflow, it is determined whether or not it is in the state immediately before overflow based on the output of the ring memory storage amount state determination section 16 (steps 6 and 7). If the state is immediately before the overflow, the fourth case occurs, and the input signal is deleted by the input signal deletion unit 21 until the underflow detection signal is output from the ring memory storage amount state determination unit 16 (step 9). That is, the ring memory 7
Ring memory 7 until
Writing to is interrupted.

When the ring memory 7 is in the state immediately before underflow, the silence insertion section 22 writes a predetermined number of silence signals "0" of 200 or less into the ring memory 7 (step 10). Then, the process returns to step 1. (5) Description of Process as Fifth Case When it is determined in step 2 that the average power value P is smaller than the threshold Th, the continuation length of the silent section up to this time is calculated (step 11) and calculated. Pause duration Tde whose duration is set in the pause duration memory 17
It is determined whether it is equal to or greater than 1 (step 12). And
When it is determined that the duration of the silent section is equal to or longer than the pause duration Tdel, the ring memory storage amount state determination unit 1
6, it is determined whether or not the state is immediately before the underflow (steps 15 and 16).

When the ring memory 7 is not in the state immediately before the underflow, the fifth case is set, and the first flag F1 indicating that the current frame is a deletion section by the input signal deletion section 25 is set (step 17). The first flag F1 has been reset (F1 = 0) in the initial setting when the power is turned on. Then, it is determined whether or not the second flag F2 indicating whether or not the current frame is the first frame of the deletion section by the input signal deletion unit 25 has been reset (step 18).

The second flag F2 has been reset (F2 = 0) in the initial setting when the power is turned on. Then, when the processing for the first frame of the deletion section by the input signal deletion unit 25 is completed, it is set to (F2 = 1). Then, when the processing for a series of deletion sections by the input signal deletion unit 25 is completed, reset (F2 =
0).

Therefore, when the current frame is the first frame of the section to be deleted by the input signal deleting section 25, the second flag F2 is reset (F2 = 0). When the second flag F2 is reset, the current frame data is stored in the first memory 31 by the waveform synthesis insertion unit 26 (step 19). Further, the writing of the current frame data to the ring memory 7 is stopped by the input signal deleting unit 25 (step 20).
That is, the current frame data is deleted. And
After the second flag F2 is set (F2 = 1) (Step 21), the process returns to Step 1.

Further, when a silent section continues,
The process proceeds to Step 16 through Steps 2, 11, 12, and 15, and it is determined whether or not the ring memory 7 is in the state immediately before the underflow based on the output of the ring memory storage amount state determination unit 16.

When the ring memory 7 is not in the state immediately before the underflow, the current frame is stored in the input signal deleting section 25.
A first flag F1 indicating that the section is a deletion section is set (step 17). Then, it is determined whether or not the second flag F2 indicating whether or not the current frame is the first frame of the deletion section by the input signal deletion unit 25 has been reset (step 18).

In this case, since the second flag F2 is set (F2 = 1), it is determined that the current frame is not the first frame of the deletion section by the input signal deletion unit 25. In this case, the current frame data is stored in the second memory 32 by the waveform synthesis insertion unit 26 (step 22). Further, the writing of the current frame data to the ring memory 7 is stopped by the input signal deleting unit 25 (step 23). Then, the process returns to step 1.

When the silent section continues and the ring memory 7 is not in the state immediately before the underflow, steps 2, 11, 12, 15, 16, 17, and
The processing of 18, 22, and 23 is repeated. That is,
The frame data in the second memory 32 is updated, and the writing of the frame data to the ring memory 7 is stopped.

After that, when the frame data of the voice section is input, in step 2, the average power value P
Is greater than or equal to the threshold Th, it is determined based on the state of the first flag F1 whether or not the previous frame was a deletion section by the input signal deletion unit 25 (step 3). In this case, since the first flag F1 is set (F1 = 1), it is determined that the previous frame was a deletion section by the input signal deletion unit 25, and the process proceeds to Step 4. In step 4, the deletion process by the input signal deletion unit 25 is stopped, and the waveform synthesis insertion process by the waveform synthesis insertion unit 26 is performed.

That is, as already described with reference to FIG. 4A, the content of the first memory 31 is multiplied by a function that changes linearly from 1 to 0, and the content of the second memory 32 is multiplied by 0 to 1 Is multiplied by a function that changes linearly, and the results of both multiplications are added. The result of this addition (corresponding to A ′ * B ′ in FIG. 4A) is sent to the ring memory 7 via the demultiplexer 27 and written into the ring memory 7.

Thereafter, the first flag F1 and the second flag F2 are reset (F1 = F2 = 0) (step 5). Then, the process proceeds to Step 6.

By the way, in the case where the above-described deletion processing by the input signal deletion unit 25 is repeatedly performed on the continuous silent section, the ring memory 7
May be in the state immediately before underflow. In this case, the result of step 16 is YES, and the
Move on to In step 24, it is determined whether or not the previous frame is a deletion section by the input signal deletion unit 25 by the first flag F.
1 is determined based on the state.

In this case, since the first flag F1 has been set (F1 = 1), the routine proceeds to step 25, where the current frame data is stored in the second memory 32. Then, the deletion processing by the input signal deletion unit 25 is stopped, and the waveform synthesis insertion processing by the waveform synthesis insertion unit 26 is performed (step 26). Then, the first flag F1 and the second flag F2 are reset (F1 = F2
= 0) (step 27), and then proceed to step 1.

The waveform synthesizing and inserting process performed by the waveform synthesizing and inserting unit 26 in step 26 is substantially the same as the waveform synthesizing and inserting process described in step 4 except that the frame data stored in the second memory 32 is The difference from the processing described in step 4 above is that the frame data is after the ring memory 7 is in the state immediately before the underflow.

The processing in step 25 is omitted, and
If the answer is YES in step 24, the second memory 32
Alternatively, the process may proceed to step 26 without storing the current frame data. In this case, in the waveform synthesis insertion process performed in step 26, the second
The frame data before the underflow immediately before state (the previous frame data) stored in the memory 32 is used.

Further, the processing of step 22 may be omitted, and a step of storing frame data in the second memory 32 may be added between step 3 and step 4. In this case, step 4
In the step 19, the first memory 31
And the contents recorded in step 3 and step 4
The waveform synthesis insertion process is performed on the basis of the contents recorded in the second memory 32 in the step added between. (6) Description of the Process in the Sixth Case When it is determined in step 2 that the average power value P is smaller than the threshold Th, the continuation length of the silent section up to this time is calculated (step 11). Pause duration Tde whose duration is set in the pause duration memory 17
It is determined whether it is equal to or greater than 1 (step 12). And
When it is determined that the duration of the silent section is equal to or longer than the pause duration Tdel, the ring memory storage amount state determination unit 1
6, it is determined whether or not the state is immediately before the underflow (steps 15 and 16).

When the ring memory 7 is in a state immediately before an underflow, it is determined whether or not the previous frame is a deletion section by the input signal deletion unit 25 based on the state of the first flag F1 (step 24). First flag F1
Is reset (F1 = 0), that is, if the previous frame is not the section to be deleted by the input signal deleting unit 25, the sixth case occurs and the process proceeds to step.
In step 28, the thinning processing section 24 thins the current frame data at a compression ratio of 1/2. The thinned data is sent to the ring memory 7 and written. Thereafter, the process returns to step 1.

That is, even if the duration of the silent section is equal to or longer than the pause duration Tdel, if the ring memory 7 is in the state immediately before underflow and the previous frame is not the section to be deleted by the input signal deleting section 25, the frame The data is not deleted, and after being thinned out at a compression ratio of 1/2, is written to the ring memory 7.

In FIG. 7, in step 12,
It is determined whether or not the duration of the silent section is longer than the set pause duration Tdel.
A, as shown in FIG.
Whether it is less than the reference length T1 (T <T1), or less than the second reference length T2 (T1 <T2) set to be equal to or longer than the first reference length T1 in which the silence duration T is set (T1 ≦ T <T2) ),
Or a second reference length T2 in which a continuation length T of a silent section is set.
Whether (T ≧ T2) or not may be determined. As the first reference length, for example, the length for four frames is
As the second reference length, for example, a length for 40 frames is set.

Then, as shown in FIG. 8, the process may proceed to the following steps according to each determination result.
That is, when the continuation length T of the silent section is less than the set first reference length T1 (T <T1), the process proceeds to step S13. First reference length T1 in which duration T of silent section is set
If it is less than the second reference length T2 (T1 <T2) (T1 ≦ T <T2) set as described above, the process proceeds to step 28, where the thinning is performed by the 1 / n thinning process. The duration T of the silent section is equal to or longer than the set second reference length T2 (T ≧ T
If 2), go to step 15.

FIG. 10 shows the relationship between the input signal and the output signal during double-speed playback, and particularly shows how the input signal in the silent section is deleted. FIGS. 11 and 12 show a data write start point to the ring memory 7, a data read start point from the ring memory 7, and points A to H in FIG.
3 shows the state of the ring memory 7 in FIG.

In FIG. 10, at the start of double-speed playback, the input signal is in a silent section and the ring memory 7 is empty (see FIG. 11A). After the data is thinned out at a compression ratio of に よ っ て by 24, the data is written to the ring memory 7.

When the accumulated amount Tm of the ring memory 7 reaches the underflow detection data Tmin, reading of data from the ring memory 7 is started (FIG. 11).
(B)).

When the frame data for the voice section a of the input signal is transmitted (point A), the frame data is compressed by the pitch compression / expansion means 23 at a compression ratio of 2/3. Frame data is decompressed on the basis of compression at a compression ratio of 1/2 at which the lengths of the input signal and the output signal match. In this sense, the decompression process is described in FIG. Then, the compressed data is written to the ring memory 7. At the point A, as shown in FIG. 11C, the accumulated amount TmA remains at Tmin.

Output signal a for speech section a of the input signal
1 is read out with a delay of the accumulated amount TmA at the point A. Then, at the point in time when the voice section a of the input signal has been input (point B), as shown in FIG. 11D, the accumulation amount Tmin at point A, which is the start point of the current compression section,
The sum of the compressed data of the voice section a from the point A to the point B and the decompression amount StB with respect to the compression at the compression ratio of 1/2 is the storage amount TmB (= StB + Tmin) of the ring memory 7. Therefore, the output signal a1 for the speech section a of the input signal
Is output when TmB (= StB + Tmin) has elapsed from point B.

The frame data of a silent section shorter than the pause duration Tdel following the voice section a of the input signal is also compressed by the pitch compression / expansion means 23 at a compression rate of 2/3. When a voice section b is input following the silent section, the frame data of the voice section b is also compressed by the pitch compression / expansion means 23 at a compression rate of 2/3.

At the point in time when the voice section b of the input signal has been input (point C), as shown in FIG. 11 (e), the accumulated amount Tm at point A which is the start point of the current compression section.
The sum of in and the decompression amount StC of the compressed data corresponding to the input signals from point A to point C with respect to 圧 縮 compression is the accumulated amount TmC (= StC + Tmin) of the ring memory 7. Therefore, the output signal b1 for the voice section b of the input signal ends being output at the point when TmC (= StC + Tmin) has elapsed from the point C.

When a signal in a silent section having a length equal to or longer than the pause duration Tdel is transmitted following the voice section b of the input signal, the signal reaches the pause duration Tdel (D
(Point), the frame data is compressed by the pitch compression / expansion means 23 at a compression ratio of 2/3.

At point D, as shown in FIG. 11 (f), the accumulation amount Tmin at point A, which is the start point of the current compression section,
Of the compressed data corresponding to the input signal from the point A to the point D and the decompression amount StD with respect to the 圧 縮 compression is the accumulated amount TmD (= StD + Tmin) of the ring memory 7.
Therefore, the output signal for the silent section between the voice section b of the input signal and the point D is TmD (= StD +
Tmin), the output ends at the point when the minute has elapsed.

The frame data in the silent section after the pause duration Tdel is deleted by the input signal deletion unit 25 until the storage amount of the ring memory 7 becomes equal to or less than the underflow detection data Tmin. The length Std of the pause deletion part is D from the point A which is the start point of the current compression section.
It becomes equal to the extension StD of the compressed data corresponding to the input signal up to the point with respect to 圧 縮 compression. Input signal deletion unit 2
After the deletion process is performed by step 5, a synthesized waveform for preventing a click sound is inserted by the waveform synthesis insertion unit 22, but the inserted synthesized waveform portion is omitted in FIG.

The last point (E
12), the accumulated amount TmE of the ring memory 7 is equal to or less than the underflow detection data Tmin, as shown in FIG. Here, an example is shown in which the accumulated amount TmE has become equal to the underflow detection data Tmin.

The frame data for the silent section from the point E is decimated by the decimating processing section 24 at a compression ratio of 、, and then written to the frame memory 7. And
When the signal of the voice section c is input (point F), the frame data of the voice section c is compressed by the pitch compression / expansion means 23 at a compression rate of 2/3. That is, a new compression section is started. Then, the compressed data is stored in the ring memory 7.
Is written to.

At the point F, as shown in FIG. 12 (h), the accumulation amount TmF of the ring memory 7 is the same as that at the point E.
in.

The output signal c for the voice section c of the input signal
1 is output with a delay of the accumulation amount Tmin at the point F. Pause duration Tde following the voice section c of the input signal
Frame data in a silent section less than 1 (a silent section from the voice section c to the point G) is also compressed by the pitch compression / expansion means 23 at a compression ratio of 2/3.

At point G, as shown in FIG. 12 (i), the accumulated amount Tmin at point F which is the start point of the current compression section.
Of the compressed data corresponding to the input signal from the point F to the point G and the decompression amount StG with respect to the 圧 縮 compression is the accumulated amount TmG (= StG + Tmin) of the ring memory 7.
Therefore, the output signal of the input signal for the silent section from the voice section c to the point G is TmG (= StG +
Tmin), the output ends at the point when the minute has elapsed.

The frame data of the silent section after the pause duration Tdel is deleted by the input signal deletion unit 25 until the storage amount of the ring memory 7 becomes the underflow detection data Tmin. Length S of this pause deletion part
td is equal to the decompression amount StG of the compressed data corresponding to the input signal from the point F to the point G, which is the start point of the current compression section, with respect to 圧 縮 compression.

The last point (H
12), the accumulated amount TmH of the ring memory 7 is equal to or less than the underflow detection data Tmin, as shown in FIG. Here, an example is shown in which the accumulated amount TmH has become equal to the underflow detection data Tmin.

The frame data for the silent section from the point H is thinned out by the thinning-out section 24 at a compression ratio of 1/2, and then written into the frame memory 7. And
When the signal of the voice section d is input, the frame data of the voice section d is compressed by the pitch compression / expansion means 23 at a compression rate of 2/3. Then, the decompressed data is written to the ring memory 7.

FIG. 13 shows the relationship between the input signal and the output signal at the time of double-speed reproduction, and particularly shows how the input signal is deleted when the state immediately before the overflow occurs.
FIG. 14 shows the ring memory 7 at each of points S to U in FIG.
The state of is shown.

From a certain point of time to a point T, a voice section a,
Frame data for a series of input signals including b, c, etc. and a silent section is compressed by the pitch compression / expansion means 23 at a compression rate of 2/3 (expanded for compression at a compression rate of 1/2). And In this case, the ring memory 7
The elongation accumulates.

At the input start point (point S) of the voice section b, as shown in FIG. 14A, the accumulation amount Tmin at the start point of the compression processing of the series of input signals and the compression processing of the compression processing are performed. The sum of the compressed data corresponding to the input signal from the start point to the point S and the decompression amount StS with respect to 圧 縮 compression is the accumulated amount TmS (= StS + Tmin) of the ring memory 7.
Therefore, the output signal b1 for the voice section b is started to be output at the point when TmS (= StS + Tmin) has elapsed from the point S.

It is assumed that at the time point (point T) when the compressed data corresponding to the input signal of the voice section c is written into the ring memory 7, the ring memory 7 is in a state immediately before overflow. That is, it is assumed that, at the point T, the accumulated amount of the ring memory 7 is equal to or larger than the overflow detection data Tmax.

At point T, as shown in FIG. 14 (b), the accumulated amount Tmin at the start point of the compression processing for the series of input signals and the input signal from the compression processing start point to point T are determined. The sum of the corresponding compressed data and the decompression amount StT with respect to １ ／ compression is the accumulated amount TmT of the ring memory 7.
(= StT + Tmin). In other words, the total number of words in the ring memory 7 is set to TOTAL, the overflow detection data is set to Tmax, and TOTAL and Tmax are used.
Is Dmin, the accumulated amount Tmt at point T is
Since it is equal to Tmax, it becomes TOTAL-Dmin.

Therefore, the output signal corresponding to the series of input signals starts from point T and accumulates TmT (= StT + Tmi).
n) Output is completed at the point of time delayed.

At the point T, when the ring memory 7 enters the state immediately before overflow, the input signals thereafter are unconditionally deleted by the input signal deletion section 21 until the ring memory 7 enters the state immediately before underflow. You.
After the deletion process is performed by the input signal deletion unit 21, silence is inserted by the silence insertion unit 22, but the inserted silence part is omitted in FIG. After the ring memory 7 enters the state immediately before overflow (T
Point), and the frame data is deleted, as shown in FIG.
It is assumed that the ring memory 7 is in a state immediately before underflow (accumulated amount TmU = Tmin) at point U as shown in FIG. In this case, an input signal including four silent sections and three voice sections d, e, and f from point T to point U is deleted. Therefore, the input signal from point T to point U does not appear as an output signal.

When the signal of the voice section g is inputted after the point U, the frame data for this voice section is compressed at a compression rate of 2/3 by the pitch compression / expansion means 23 (compared to the compression at the compression rate of 1/2). After that, the data is written into the ring memory 7. The output signal g for the voice section g is started to be output with a delay of the accumulation amount Tmin of the ring memory 7 at the point U.

In the above embodiment, the voice section and the silent section of the input signal are determined based on the average power value P of each frame, but may be determined based on the average amplitude of each frame. . In this case, as shown in FIG. 15, an average amplitude calculator 11A for calculating an average amplitude value for each frame is provided instead of the power calculator 11 in FIG. When the number of quantization bits of the D conversion unit 2 is 12 bits, for example, the value 26
Is set. And the average amplitude calculation unit 1
The comparison unit 12A compares the average amplitude value calculated by 1A with the threshold value of the threshold value memory 13A, thereby determining whether it is a voice section or a silent section.

That is, if the average amplitude value is equal to or greater than the threshold value, it is determined to be a voice section, and if the average amplitude value is less than the threshold value, it is determined to be a silent section. The average amplitude value W for each frame is obtained by calculating the amplitude of each audio signal in one sampled frame as i0, i1,... IN-1 (where N = 2
00), it is calculated based on the following Equation 3.

[0154]

(Equation 3)

The other processing is the same as the processing performed by the speech speed conversion unit 6 in FIG. 2, and a description thereof will be omitted.

Note that, in this case, the threshold value may be changed as follows. That is,
As shown by a dotted line in FIG. 15, an average amplitude steady state detection and threshold value updating unit 14A is provided. The average amplitude steady state detection and threshold update unit 14A includes an average amplitude calculation unit 11
It is determined whether or not the average amplitude value W from A is constant over a predetermined number of frames. If the average amplitude value W is constant (steady state), a value twice the average amplitude value W at that time is stored in the threshold memory 13A. To update the threshold. However, the maximum value of the updated threshold value is limited to a predetermined value, for example, 28.

The speech section and the silence section of the input signal may be determined on the basis of the cumulative amplitude Wa of the speech signal of each frame represented by the following equation 4 and a given threshold value. .

[0158]

(Equation 4)

The voice section and the silence section of the input signal are detected as the signal periodicity of each frame, and if the detected cycle is within a predetermined pitch range of the voice signal, the section is a voice section. If the detected cycle is outside the predetermined pitch cycle range of the audio signal, it may be determined to be a silent section.

In this case, as shown in FIG.
Is provided with a pitch cycle detecting section 11B for detecting the periodicity of each frame based on the autocorrelation method, and the pitch memory range of the voice signal is set in the threshold value memory 13B. You. Then, the cycle detected by the pitch cycle detecting unit 11B and the threshold memory 13
The comparison unit 12B compares the pitch cycle range of the audio signal set to B.

The pitch period range of the audio signal to be set is:
It depends on the reproduction speed, and is set, for example, in the range of 66 × n (Hz) to 320 × n (Hz) at the time of n × speed reproduction. Therefore, at the time of double speed reproduction, the pitch cycle range of the audio signal is set to a range of 132 Hz to 640 Hz. Other processes are the same as the processes performed by the speech speed conversion unit 6 in FIG. 2, and thus description thereof is omitted.

Further, the speech section and the silent section of the input signal may be determined by comparing the power spectrum of the signal of each frame with the power spectrum of the steady state.

In this case, as shown in FIG.
Is provided with a power spectrum calculator 11C for calculating a power spectrum for one or a plurality of predetermined frequency bands for each frame.
The power spectrum in the steady state for the predetermined one or a plurality of frequency bands is stored in the power spectrum storage unit 1.
3C.

The contents of the power spectrum storage unit 13C are as follows:
When the power spectrum steady state detection unit 14B detects that the power spectrum is in the steady state based on the change state of the power spectrum calculated by the power spectrum calculation unit 11C, the power spectrum is updated to the detected power spectrum in the steady state.

The input signal is a power spectrum calculator 11C.
, A power spectrum for one or more predetermined frequency bands is calculated for each frame.
Then, the comparison unit 12C compares the calculated power spectrum with the power spectrum in the steady state stored in the power spectrum storage unit 13C.

If the calculated power spectrum fluctuates with respect to the steady-state power spectrum, the frame is determined to be a voice section. Conversely, if the calculated power spectrum does not fluctuate from the steady-state power spectrum, the frame is determined to be a silent section.

Specifically, the power spectrum storage unit 13
C includes the predetermined one based on the steady-state power spectrum for the predetermined one or more frequency bands.
Alternatively, threshold values for a plurality of frequency bands are stored. Then, it is stored in the power spectrum storage unit 13C. The power spectrum for the predetermined one or more frequency bands calculated by the power spectrum calculation unit 11C is compared with the corresponding threshold value stored in the power spectrum storage unit 13C, so that the input signal is It is determined whether the section is a section or a silent section.

For example, it is assumed that the power spectrum in the steady state is a power spectrum of only noise as shown in FIG. Also, it is assumed that the power spectrum of the voice without noise is shown in FIG. In the steady state, when noise shown by the power spectrum of FIG.
When an audio signal having the power spectrum shown by 1 (b) is input, the power spectrum is obtained by combining both power spectra as shown in FIG. 21 (c).

Therefore, for example, a frequency band f having a relatively small power in a power spectrum in a steady state.
The powers for a and fb increase significantly in the power spectrum of the voice section. That is, by comparing the steady-state power in one or more frequency bands having relatively small power in the steady-state power spectrum with the power in the one or more frequency bands of the power spectrum of the input signal, Is a voice section or a silent section.

If it is known that the noise in the steady state is a noise in a high frequency band, a power spectrum for a low frequency band (for example, a frequency band of 4 KHz or less) where the influence of the noise is small is calculated. Depending on whether the calculated power spectrum is equal to or greater than a predetermined threshold,
It is also possible to determine whether the input signal is a voice section or a silent section.

Further, by comparing the average power value P of each frame with the threshold value Th, when discriminating between a voice section and a silent section, the threshold value is determined based on the storage amount of the ring memory 7. Th may be changed. That is, the threshold value T is set such that the smaller the storage amount of the ring memory 7, in other words, the larger the empty area of the ring memory 7, the smaller the missing portion of the voice section.
h is reduced. This makes the output sound closer to nature.

That is, as shown in FIG. 22, a threshold adjusting means 51 is provided. The threshold adjusting unit 51 obtains the storage amount of the ring memory 7 from the ring memory storage amount state determination unit 16. Then, the storage time Tm is calculated by dividing the obtained storage amount of the ring memory 7 by the sampling frequency of the D / A converter 8. Then, the threshold value Th is determined based on the calculated accumulation time Tm, and the content of the threshold value memory 13 is updated.

More specifically, the storage amount of the ring memory 7 obtained from the ring memory storage amount state determination unit 16 is 800, which is the sampling frequency of the D / A conversion unit 8.
By dividing by 0, the accumulation time Tm is obtained.
Then, the threshold value Th for the accumulation time Tm is obtained based on the data of the threshold value Th for the accumulation time Tm created in advance.

The following table shows an example of the data of the threshold Th with respect to the accumulation time Tm when the number of quantization bits of the A / D converter 2 is 12 bits.

[0175]

[Table 1]

When discriminating between a voice section and a silent section by comparing the power accumulated value Pa of each frame with a threshold value, the average amplitude value W of each frame is compared with the threshold value. When discriminating between the voice section and the silent section, the power section of each frame is compared with the threshold value by comparing the amplitude cumulative value Wa of each frame with the threshold value. Is also determined in the same manner as described above.
The threshold value may be changed based on the accumulated amount of.

Further, the pause continuation length Tdel for determining the silence section deletion start point may be changed based on the storage amount of the ring memory 7. That is,
The pause continuation length Tdel is increased such that the smaller the storage amount of the ring memory 7, in other words, the larger the empty area of the ring memory 7, the smaller the number of silence sections to be deleted. This makes the output sound closer to nature.

That is, as shown in FIG. 22, a pause duration adjusting means 52 is provided. Pause continuation length adjusting means 52
Obtains the storage amount of the ring memory 7 from the ring memory storage amount state determination unit 16. And the obtained ring memory 7
Is divided by the sampling frequency of the D / A converter 8 to calculate the accumulation time Tm. Then, based on the calculated accumulation time Tm, the pause duration Tde
is determined, and the contents of the pause continuation length setting memory 17 are updated.

More specifically, the storage amount of the ring memory 7 obtained from the ring memory storage amount state determination unit 16 is 800, which is the sampling frequency of the D / A conversion unit 8.
By dividing by 0, the accumulation time Tm is obtained.
Then, the pause duration Tdel for the accumulation time Tm is obtained based on the data of the pause duration Tdel for the accumulation time Tm created in advance.

The following table shows an example of the data of the pause duration Tdel with respect to the accumulation time Tm at the time of double speed reproduction of the VTR.

[0181]

[Table 2]

In the above, the case where the input signal is an analog signal has been described. However, the present invention can be applied to a case where the input signal is digital data. For example,
When a compressed digital audio signal is sent from an IC memory, a magnetic disk, a digital communication line, or the like, the compressed digital audio signal is expanded and the PC
The converted PCM audio signal is temporarily stored in a buffer. Thereafter, the PCM audio data is read from the buffer at a speed corresponding to the set reproduction speed magnification, and is sent to the frame memory 5 in FIG.

The adaptive speech rate conversion processing has been described in detail above. Next, a specific memory control operation for speech rate conversion by the simple thinning method will be described. That is, the speech speed conversion IC 112 controls the memory 113 as follows to perform the speech speed conversion.

FIG. 28 shows a memory control operation at the time of triple speed reproduction, in which the audio signal is written at the triple speed, reading is started at the same time as the writing, and reading is performed so that the audio has the same speed as the normal speed reproduction. finish. That is, memory control is performed so that writing is performed in a cycle T / 3 that is 1/3 of the read cycle T, and accordingly, the T0 period in the drawing is thinned out. Similarly, at the time of 5 × speed reproduction, the writing time is T / 5, and at the time of 9 × speed reproduction, the writing time is T / 5.
/ 9.

FIG. 29 shows the read / write timing of the memory at the time of the reverse 5 × speed reproduction.
Double speed cycle, W is write period, * 2 is T / 5, * is T /
It is 6. The timing of the write address of the memory is 5 × speed, the write cycle period is 5/6 of the 5 × speed forward reproduction, and the write cycle counter with 6 times in the T period circulates 5 times, and Writing is performed only during the period when the value is 0. For this reason, as shown in the figure, the writing and reading go round with a slight shift, and the reading returns to the original state five times, so that the reproduced sound can be heard smoothly without changing the content at the point α. As shown in the figure, writing and reading are performed in the order of d and D, e and E, f1 is F1, and F2 is f2. In FIGS. 28 and 29, the broken line indicates the stepped state of the write address, and the solid line indicates the stepped state of the read address.

[0186]

As described above, in the VTR according to the present invention, the adaptive speech speed conversion processing is performed at the time of double speed reproduction, and the simple thinning processing is performed at the time of triple speed reproduction or more. In addition to simplifying the configuration of the IC for use, the appropriate speech speed conversion processing is automatically selected, so that the commercial value of the VTR can be improved. Further, according to the present invention, there is an advantage that the processing load can be reduced at the time of double-speed reproduction, the gap between the video and the audio can be reduced, and the capacity of the memory for storing the audio signal does not become enormous.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an overall configuration of a speech speed conversion device.

FIG. 2 is a block diagram illustrating a configuration of a speech speed conversion unit.

FIG. 3 shows that an input signal is compressed at a compression ratio of 2 /
FIG. 3 is an explanatory diagram showing a method of performing compression in Step 3;

FIG. 4 is an explanatory diagram for describing processing by a waveform synthesis processing unit;

FIG. 5 is an explanatory diagram for explaining various thinning processing methods performed by a thinning processing unit.

FIG. 6 is a flowchart illustrating a processing procedure by a speech speed conversion unit;

FIG. 7 is a flowchart illustrating a processing procedure by a speech speed conversion unit;

FIG. 8 is a flowchart illustrating a modified example of the processing procedure by the speech speed conversion unit and corresponding to FIG. 7;

FIG. 9 is an explanatory diagram for explaining a process that can be replaced with the process of step 10 in FIG. 6;

FIG. 10 is a time chart showing a relationship between an input signal and an output signal at the time of double-speed reproduction, and particularly showing a state in which an input signal in a silent section is deleted.

FIG. 11 shows a starting point of data writing to the ring memory 7,
FIG. 11 is a schematic diagram showing a state of starting reading data from the ring memory 7 and a state of the ring memory 7 at points A to D in FIG. 10.

12 is a schematic diagram showing the state of the ring memory 7 at points E to H in FIG.

FIG. 13 shows a relationship between an input signal and an output signal at the time of 2 × speed reproduction.
6 is a time chart illustrating a state in which an input signal is deleted.

14 is a diagram showing a ring memory 7 at each of points S to U in FIG. 13;
It is a schematic diagram which shows the state of.

FIG. 15 is a block diagram illustrating a modified example of a circuit for determining a voice section and a silent section, corresponding to FIG. 2;

FIG. 16 is a block diagram showing another modified example of the circuit for distinguishing between a voice section and a silent section, and corresponds to FIG. 2;

FIG. 17 shows a compression ratio of 2 /
FIG. 3 is an explanatory diagram showing a method of performing compression in Step 3;

FIG. 18 is an explanatory diagram for explaining a process that can be replaced with the process of step 9 in FIG. 6;

19 is an explanatory diagram for explaining a process that can be replaced with the process of step 10 of FIG. 6 when the process of FIG. 18 is adopted as the process of step 9 of FIG. 6;

FIG. 20 is a block diagram showing still another modification of the circuit for discriminating between a voice section and a silent section, corresponding to FIG. 2;

FIG. 21 is a graph showing a power spectrum in a steady state, a power spectrum of a voice without noise, and a power spectrum of a voice section.

FIG. 22 is a block diagram showing a speech speed conversion unit to which a threshold adjustment unit and a pause duration adjustment unit are added.

FIG. 23 is a main circuit block diagram of a video tape recorder embodying the present invention.

24 is a diagram showing a flowchart for explaining the operation of the circuit block diagram of FIG. 23;

FIG. 25 is a diagram for explaining the concept of adaptive speech speed conversion processing.

FIG. 26 is a diagram for explaining the concept of speech speed conversion by simple thinning processing.

FIG. 27 is a diagram for explaining the concept of speech speed conversion by simple thinning processing during reverse playback.

FIG. 28 is a diagram for explaining a memory control method for realizing a simple thinning process.

FIG. 29 is a diagram for explaining a memory control method for implementing a simple thinning process during reverse playback.

[Explanation of symbols]

 2 A / D converter 4 DSP 5 Frame memory 6 Speech speed converter 7 Ring memory 8 D / A converter 9 Up / down counter 11 Power calculator 11A Average amplitude calculator 11B Pitch cycle detector 11C Power spectrum calculator 12, 12A, 12B, 12C Comparison unit 15 Conditional branching unit 16 Ring memory storage amount state determination unit 21, 25 Input signal deletion unit 23 Pitch compression / expansion unit 24 Decimation processing unit 51 Threshold adjustment unit 52 Pause continuation length adjustment unit 112 Talk speed Conversion IC 114 microcomputer (microcomputer)

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁷ , DB name) G10L 21/04

Claims (5)

(57) [Claims] At the time of double speed reproduction, speech speed conversion is performed for performing compression / decompression processing or deletion processing on an input audio signal according to whether the reproduced audio signal is a voice section or a silent section. A double-speed audio playback mode and an N-times playback mode for performing a process of thinning out a voice section of a predetermined period of a playback audio signal in accordance with the playback double speed during ± N-times playback (N: a natural number of 3 or more). A video tape recorder comprising control means for setting. 2. A speech memory according to claim 1, further comprising: a speech speed conversion processing means for performing speech speed conversion processing of an input speech signal in order to perform a double speed speech reproduction mode; Means for reading out data at a constant speed from the voice signal, and the speech speed conversion processing means compresses and decompresses the input voice signal according to whether the input voice signal is a voice section or a silent section and the amount of storage in the ring memory. A video tape recorder comprising a speech speed converter provided with means for performing processing or deletion processing. 3. An A / D converter for sampling an input analog audio signal at a sampling frequency corresponding to a set reproduction speed magnification in order to perform a double speed audio reproduction mode. A frame memory to which an audio signal output from the means is input, each time a required number of audio signals are input to the frame memory, a voice speed conversion processing means for performing a voice speed conversion process on those voice signals,
A ring memory into which the output of the speech speed conversion processing means is written;
A reading unit that reads data from the ring memory at a constant speed; and a storage amount calculation unit that calculates a storage amount of the ring memory based on a write signal and a read signal of the ring memory. The section discriminating means for discriminating whether the input speech corresponding to the required number of speech signals inputted to the frame memory is a speech section or a silent section, and the output of the section discriminating means and the output of the accumulation amount calculating means, A video tape recorder comprising a speech speed conversion device including signal processing means for performing compression / decompression processing or deletion processing on a required number of audio signals. 4. A frame memory in which an input digital audio signal is written at a speed corresponding to a set reproduction speed magnification in order to perform a double speed audio reproduction mode. Each time a signal is input, a speech speed conversion processing means for performing speech speed conversion processing on those audio signals, a ring memory in which the output of the speech speed conversion processing means is written, and a writing to a frame memory at the time of 1 × speed reproduction A read unit that reads data from the ring memory based on a read signal having a frequency equal to the speed; and a storage amount calculation unit that calculates a storage amount of the ring memory based on a write signal and a read signal of the ring memory. The speech rate conversion processing means determines whether the input voice corresponding to the required number of voice signals input to the frame memory is a voice section or a silent section. And a signal processing means for performing compression / expansion processing or deletion processing on the required number of audio signals in accordance with the output of the section determination means and the output of the storage amount calculation means. A video tape recorder comprising a speech speed conversion device. 5. The apparatus according to claim 1, further comprising a memory control means for controlling the memory so as to write the audio data into the memory at N times speed and to read out the written data at 1 times speed in order to perform the N times speed reproduction mode. A video tape recorder characterized in that: