JPH08292789A - Speech speed changing device - Google Patents

Abstract

PURPOSE: To provide a speech speed changing device in which the degradation of a voice is not generated when the changing of a speech speed is performed. CONSTITUTION: This device is a speech speed changing device provided with a time base compressing and expanding means 14 for performing the time base compression and expansion processing of an input voice and outputting the input voice while making the voice speed of the input voice slow. Moreover, this device is provided with a means 12 discriminating whether a periodicity is present or not in the input voice and means 13, 15 allowing the time base compression and expansion processing by the time base compressing and expanding means 14 to be performed only to sections in which the periodicities are present among the input signal.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a speech speed conversion device for slowing down and outputting a voice speed of input voice.

[0002]

2. Description of the Related Art A speech speed conversion technique for obtaining an output voice at a slower speed than the input voice speed by subjecting an input voice signal to a time axis compression / expansion process has been conventionally known. However, when the input voice is subjected to the time axis compression / expansion process, there is a problem that the voice is deteriorated.

[0003]

SUMMARY OF THE INVENTION It is an object of the present invention to provide a speech speed conversion device which does not cause voice deterioration when speech speed conversion is performed.

[0004]

A first speech speed conversion apparatus according to the present invention comprises time-axis compression / expansion means for time-axis compression / expansion processing of input voice, and outputs the voice at a low voice speed. In the speech speed converting apparatus, a unit for determining whether or not the input voice has periodicity, and a time axis by the time axis compression / expansion unit only for a section of the input signal determined to have periodicity. It is characterized in that it is provided with means for performing compression / decompression processing.

A second voice speed conversion apparatus according to the present invention is a voice speed conversion apparatus which comprises time-axis compression / expansion means for time-axis compression / expansion processing of input voice, and outputs the voice at a low voice speed. , Means for determining whether the input voice has periodicity, means for calculating the duration when it is determined to have periodicity, and of the input signal, the duration of the periodicity is equal to or greater than a predetermined value. It is characterized in that it is provided with means for performing the time axis compression / expansion processing by the time axis compression / expansion means only for the section.

A third speech speed conversion apparatus according to the present invention is provided with a time axis compression means for time axis compression processing of input speech, and in the speech rate conversion apparatus for slowing down and outputting the input speech, A means for determining whether or not the sound has periodicity, a means for calculating the duration when it is determined to have the periodicity, and a duration of the periodicity of the input signal is less than the first predetermined length. For the section, the means for outputting the time axis compression processing by the time axis compression / decompression means without performing the time axis compression processing, and the duration of the periodicity of the input signal that is equal to or greater than the first predetermined length and is greater than the first predetermined length. Of the input signal, the means for performing the time-axis compression process by the time-axis compression means using a relatively large compression rate for a section having a large second predetermined length or less.
It is characterized in that it is provided with means for performing time-axis compression processing by the time-axis compression means using a relatively small compression rate for a section in which the continuity length of the periodicity is equal to or greater than the second predetermined length.

[0007]

In the first speech speed converting apparatus according to the present invention, it is determined whether the input voice has periodicity. Then, the time axis compression / expansion processing is performed by the time axis compression / expansion means only for the section of the input signal that is determined to have periodicity.

In the second speech speed converting apparatus according to the present invention,
It is determined whether the input voice has periodicity. When it is determined that there is periodicity, its continuation length is calculated. Then, in the input signal, the time axis compression / expansion processing is performed by the time axis compression / expansion means only for a section in which the continuity of the periodicity is equal to or greater than a predetermined value.

In the third speech speed converting apparatus according to the present invention,
It is determined whether the input voice has periodicity. When it is determined that there is periodicity, its continuation length is calculated. Of the input signal, the time axis compression processing by the time axis compression / expansion means is not performed for the section where the continuity of the periodicity is less than the first predetermined length. Of the input signals, the duration of periodicity is the first
The time axis compression process is performed by the time axis compression means using a relatively large compression rate for a section that is greater than or equal to the predetermined length and less than the second predetermined length that is greater than the first predetermined length.
Of the input signal, the time axis compression process is performed by the time axis compression means using a relatively small compression rate for a section in which the continuity of the periodicity is equal to or greater than the second predetermined length.

[0010]

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

(1) Description of the first embodiment

A first embodiment of the present invention will be described with reference to FIG. FIG. 1 shows the configuration of a speech speed conversion device that makes an output voice speed slower than an input voice speed.

The speech speed conversion device comprises a voice signal input unit 11, a periodicity determination unit 12, an input switching unit 13, a time axis compression / expansion unit 14, an output switching unit 15, a buffer 16 and a voice signal output unit 17. .

The audio signal input section 11 is provided with, for example, an amplification section, an A / D conversion section, a frame memory and the like. The signal input to the audio signal input unit 11 is amplified,
It is converted into a digital signal and stored in the frame memory. The output of the audio signal input unit 11 is the periodicity determination unit 12
To the input switching unit 13.

In this embodiment, the analog signal is input to the speech speed converter, but a digital signal read from an IC memory or the like may be input to the speech speed converter. In this case, it is not necessary to provide the audio signal input unit 11 with an A / D conversion unit.

The periodicity determination unit 12 detects whether or not the input audio signal has periodicity. The detection of the periodicity of the input signal is performed based on the autocorrelation method as is well known. The periodicity determination unit 12 outputs a first switching signal when it determines that the input audio signal has periodicity, and outputs a second switching signal when it determines that the input signal has no periodicity. The switching signal output from the periodicity determination unit 12 is sent to the input switching unit 13 and the output switching unit 15.

The input switching unit 13 includes a periodicity determining unit 12
The destination of the input signal is switched based on the switching signal from. That is, when the first switching signal is input, the input audio signal is sent to the time axis compression / expansion means 14, and when the second switching signal is input, the input audio signal is sent to the output switching unit 15.

When the input voice speed is a voice reproduced at a speed higher than the standard reproduction speed, the time axis compression / expansion means 14 inputs at a compression ratio α of 1 / n or more, where the reproduction speed multiplication factor is n. Compress the signal. When the input sound is reproduced at the standard reproduction speed, the compression rate α is 1 or more,
Decompress the input signal.

The time axis compression / expansion method used by the time axis compression / expansion means 14 is, for example, a pointer addition control overlap pointer addition method (Pointer Interval Control Overlap a).
nd Add: PICOLA), TDHS (Time Domain Harmonic Scalin
g) There are laws, etc.

Using PICOLA, the input voice signal (the voice data input to the time axis compression / decompression means 14) is compressed at a compression rate of 2.
A method of compressing with / 3 will be briefly described with reference to FIG. First, the pitch period is extracted from the input voice signal. The extracted pitch period is Tp. For the waveform A, a weight that linearly goes from 1 to 0 (weight function K1)
Are added to form a waveform A ′. A weight (weighting function K2) from 0 to 1 is applied to the waveform B to create a waveform B '.

Then, these waveforms A'and B'are added together to form a waveform A '* B' of length Tp.
These weights are added to maintain continuity at the connection points before and after the waveform A ′ * B ′. Then the pointer becomes
The same operation is performed after moving by 3 Tp, which is the length determined based on the compression rate. 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 3 pitch periods is compressed into the signal for 2 pitch periods.

The output switching unit 15 includes a periodicity determining unit 12
One of the two inputs is selected based on the switching signal from and output to the buffer 16. That is, when the first switching signal is input, the output of the time axis compression / expansion means 14 is output to the buffer 16, and when the second switching signal is input, the output of the input switching unit 13 is output to the buffer 16. To do.

The voice data accumulated in the buffer 16 is
It is read at a constant speed and sent to the audio signal output unit 17.

The audio signal output section 17 includes a D / A conversion section. From the output switching unit 15 to the audio signal output unit 17
The digital signal sent to is converted into an analog signal and output from the audio signal output unit 17.

In this embodiment, the case where the voice signal is output as an analog signal from the voice speed converter is shown, but the voice signal may be output as a digital signal from the voice speed converter. In this case, the audio signal output unit 17
It is not necessary to provide a D / A conversion unit in.

When both the input and output signals of the speech speed converter are analog signals, D in the voice signal output section 17 is used.
The sampling frequency of the A / A converter is set to the standard sampling frequency f _SO, and the A / D in the audio signal input unit 11 is set.
The sampling frequency of the converter is set to n · f _SO , where n is the reproduction speed multiplication factor. For example, when the input voice speed is voice reproduced at a speed twice the standard reproduction speed, that is, when the reproduction speed magnification n of the input signal is 2, the A / D in the audio signal input unit 11 is The sampling frequency of the conversion unit is 2f _SO , and D in the audio signal output unit 17
The sampling frequency of the / A converter is f _SO .

When both the input and output signals of the speech speed conversion device are digital signals, assuming that the reproduction speed multiplication factor is n, the audio signal input unit with respect to the output speed of the data output from the audio signal output unit 17 will be described. The input speed of the data input to 11 is set to be n times.

The operation of the speech speed converter will be described. The periodicity determination unit 12 determines whether or not the audio signal input to the audio signal input unit 11 has periodicity. When it is determined that the input signal has periodicity, the first switching signal is output from the periodicity determination unit 12 and sent to the switching units 13 and 15. Therefore, when it is determined that the input audio signal has periodicity, the input audio signal is sent to the time axis compression / expansion means 14.

When the reproduction speed magnification n is, for example, 2, the time axis compression / expansion means 14 uses, for example, a compression ratio of 2.
At / 3, the input signal is compressed. That is, the waveforms are thinned out so that 3 pitches of the input audio signal become 2 pitches. Therefore, the speed of the voice output from the voice signal output unit 17 is ⅔ times the standard voice speed, but the pitch is the pitch at the standard speed reproduction.

When the reproduction speed magnification n is, for example, 1, the time axis compression / expansion means 14 uses, for example, a compression ratio of 3.
At / 2, the input audio signal is expanded. That is, the waveform is inserted so that 2 pitches of the input audio signal become 3 pitches.

The signal subjected to the time axis compression / expansion processing by the time axis compression / expansion means 14 is accumulated in the buffer 16 via the output switching section 15, and then the audio signal output section 17 is provided.
Is output via.

When the periodicity determining unit 12 determines that the input signal has no periodicity, the second switching signal is output from the periodicity determining unit 12 and each switching unit 13,
Sent to 15. Therefore, when it is determined that the input audio signal has no periodicity, the input audio signal is accumulated in the buffer 16 without being subjected to the time axis compression / expansion process. Then, it is output through the audio signal output unit 17.

The time axis compression / expansion processing means 14 performs processing such as thinning out or inserting the input voice in pitch cycle units. Therefore, for a voice signal without periodicity,
When the time-base compression / decompression process is performed, the sound is deteriorated. In the first embodiment, when the input voice has no periodicity, the time-base compression / decompression means 14 does not perform the time-base compression / decompression process, so that the voice is prevented from being deteriorated.

(2) Description of the second embodiment

A second embodiment of the present invention will be described with reference to FIG. FIG. 2 shows the structure of the speech speed conversion device. 2, the same parts as those in FIG. 1 are designated by the same reference numerals and the description thereof will be omitted.

In this speech speed converting apparatus, the periodicity judging section 12
The continuation length calculation unit 21 is provided in the subsequent stage. The continuation length calculation unit 21 calculates the continuation length when the periodicity determination unit 12 determines that there is periodicity. Then, when the continuation length exceeds a predetermined value, the first
Output a switching signal.

That is, the continuation length calculation unit 21 outputs the first switching signal when the periodicity determination unit 12 determines that there is periodicity and the continuation length is equal to or greater than the predetermined value.
When the periodicity determination unit 12 determines that there is no periodicity, or when the periodicity determination unit 12 determines that there is periodicity and the duration is less than the predetermined value, the duration calculation unit 21 Output a switching signal.

The operation of each switching unit 13, 15 when the first switching signal is output and the operation of each switching unit 13, 15 when the second switching signal is output are as follows:
Same as the embodiment. Therefore, in the second embodiment, the time axis compression / expansion means 14 performs the time axis compression / expansion processing on the input audio signal only when the periodicity judgment section 12 judges that there is periodicity and the duration is equal to or more than the predetermined value. It

(3) Description of the third embodiment

A third embodiment of the present invention will be described with reference to FIG. FIG. 3 shows the structure of the speech speed conversion device. In FIG. 3, the same thing as FIG. 1 or FIG.
The same reference numerals are given and the description thereof is omitted.

This speech speed converting apparatus is for outputting the input voice signal by reducing the voice speed when the input voice signal is a voice reproduced at a speed higher than the standard reproduction speed. For this reason, since the time axis expansion is not performed, the time axis compression means 23 is provided instead of the time axis compression and expansion means.

In this speech speed converter, the time axis compression means 2
A compression rate adjusting means 22 for adjusting the compression rate of No. 3 by a control signal from the continuation length calculation unit 21 is provided.

The continuation length calculation unit 21 calculates the continuation length when the periodicity determination unit 12 determines that there is periodicity. The duration of the voice signal having periodicity is first
Is less than the reference value T1 of the continuation length calculation unit 21
Output a switching signal. The duration of the periodic audio signal is greater than or equal to the first reference value T1 and the second reference value T2 (T1 <T
If less than 2), the continuation length calculation unit 21 outputs the second switching signal. The duration of the audio signal with periodicity is the second
When the reference value T2 is equal to or larger than the reference value T2 of
Output a switching signal.

The switching signal output from the continuation length calculation unit 21 is sent to the switching units 12 and 15 and
It is also sent to the compression rate adjusting means 22. Input switching unit 12
Sends the input audio signal to the output switching unit 14 when the first switching signal is sent, and sends the input audio signal to the time axis compression means 23 when the second or third switching signal is sent.

The output switching unit 15 sends the output of the input switching unit 12 to the buffer 16 when the first switching signal is sent, and by the time axis compression means 23 when the second or third switching signal is sent. The audio signal subjected to the time base compression processing is sent to the buffer 16.

The compression ratio adjusting means 22 does not operate when the first switching signal is sent. When the second switching signal is sent to the compression rate adjusting means 22, the compression rate adjusting means 22 sets the compression rate to a relatively large value. That is, do not thin out the input signal. When the third switching signal is sent to the compression rate adjusting means 22, the compression rate adjusting means 22 sets the compression rate to a relatively small value. That is, the input signals are thinned out a lot.

For example, the operation of the speech speed conversion device when the input voice speed is a voice reproduced at a speed twice the standard reproduction speed, that is, when the reproduction speed magnification n of the input signal is 2 will be described. . When it is determined that the input audio signal has no periodicity, or even when it is determined that there is periodicity, the duration is less than the first reference value T1 (when the first switching signal is output), the input The audio signal is sent to the buffer 16 without time axis compression processing. The speed of the output sound with respect to such an input sound signal becomes the standard speed.

When it is determined that the input audio signal has periodicity and its duration is equal to or more than the first reference value T1 and less than the second reference value T2 (when the second switching signal is output). The compression rate adjusting means 22 sets the compression rate to, for example, 4/5. The speed of the output sound with respect to the input sound signal is 5/4 times the standard speed.

When it is determined that the input audio signal has periodicity and its duration is equal to or greater than the second reference value T2 (third value).
When the switching signal is being output), the compression rate adjusting means 22 sets the compression rate to, for example, 1/2. The speed of the output sound with respect to the input sound signal is twice the standard speed.

The reason why the input audio signal is thinned out a lot when the duration of the periodicity is equal to or greater than the second reference value T2 is as follows. That is, in the portion where the vowel is extended, such as “ar”, the continuity of the periodicity is long. Therefore, in such a section, since the output voice can be heard even if the degree of thinning is large, the value α of the compression rate is made small (the thinning rate is made large).

In the second or third embodiment, means for detecting the duration of the non-periodic voice signal is added, and the duration of the non-periodic voice signal is equal to or more than a predetermined value. At this time, a means for deleting the input voice may be provided. The reason why the input voice is deleted when the duration of the voice signal having no periodicity is equal to or larger than a predetermined value is as follows. That is, when the duration of the voice signal having no periodicity is equal to or greater than a predetermined value, it can be determined that the section is not a human voice.

In the second or third embodiment, means for detecting the duration of the non-periodic voice signal is added, and the duration of the non-periodic voice signal is equal to or more than a predetermined value. At that time, the input voice may be thinned out to a relatively large degree.

[0053]

According to the present invention, it is possible to prevent the voice from being deteriorated when the voice speed conversion is performed.

[Brief description of drawings]

FIG. 1 is a configuration diagram showing a schematic configuration of a speech speed conversion device according to a first embodiment of the present invention.

FIG. 2 is a configuration diagram showing a schematic configuration of a speech speed conversion device which is a second embodiment of the present invention.

FIG. 3 is a configuration diagram showing a schematic configuration of a speech speed conversion device which is a third embodiment of the present invention.

FIG. 4 is a schematic diagram for explaining a time axis compression / expansion method using PICOLA.

[Explanation of symbols]

 11 audio signal input unit 12 periodicity determination unit 13 input switching unit 14 time axis compression / expansion unit 15 output switching unit 16 buffer 17 audio signal output unit 21 continuation length calculation unit 22 compression ratio adjustment unit 23 time axis compression unit

Claims (3)

[Claims] 1. A speech speed conversion device, comprising a time-axis compression / expansion means for performing time-axis compression / expansion processing of an input voice, wherein the voice speed of the input voice is output at a low speed, whether the input voice has periodicity or not. And a unit that causes the time-axis compression / expansion unit to perform the time-axis compression / expansion process only on the section of the input signal that is determined to have periodicity. Speak rate converter. 2. A speech speed conversion device, comprising a time axis compression / expansion means for performing time axis compression / expansion processing of input speech, and outputting the speech at a reduced speech speed, whether or not the input speech has periodicity. Means for determining whether or not there is periodicity, means for calculating the duration when it is determined that there is periodicity, and time-axis compression / expansion only for sections of the input signal where the duration of periodicity is greater than or equal to a predetermined value. A speech speed conversion device comprising: means for performing time-axis compression / expansion processing by means. 3. A speech speed conversion device, comprising a time axis compression means for time axis compression processing of input voice, and outputting the voice at a low voice speed of the input voice, wherein whether or not the input voice has periodicity is determined. Determining means, means for calculating the duration when it is determined that there is periodicity, time-axis compression / expansion for the section of the input signal where the duration of the periodicity is less than the first predetermined length Means for outputting without performing time-axis compression processing by means, in the section of the input signal in which the duration of the periodicity is greater than or equal to the first predetermined length and less than the second predetermined length that is greater than the first predetermined length. On the other hand, for the means for performing the time-axis compression processing by the time-axis compression means using a relatively large compression rate, and for the section of the input signal in which the continuity length of the periodicity is the second predetermined length or more, , Using a relatively small compression ratio, Speech speed converting device comprising: means for performing time base compression processing.