JP2002297200A - Speaking speed converting device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a speaking speed converting device which can perform an adequate speaking speed conversion processing for aural signals of multiple channels while keeping a reproduced aural signal synchronous. SOLUTION: This speaking speed converting device which performs the speaking-speed conversion processing for input aural signals of multiple channels inputted from a voice reproducing device according to pitch cycles obtained from those input aural signals is equipped with a pitch cycle calculation part 124 which calculates a single pitch cycle for each processing section from the input aural signals of the multiple channels and a time-base compression and expansion part 14 which time-base-compresses the input aural signals of the multiple channels in the respective processing sections according to the obtained pitch cycles.

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 changing the speech speed of an audio signal, and for example, a voice for performing a fast or slow listening of a voice of a television, a laser disk, a VTR, a hard disk recorder, etc. accompanied by video. Playback device,
For the hearing impaired and the elderly, hearing aids and devices such as telephones equipped with the hearing aids, which convert broadcast audio signals into slow, easy-to-hear sound, as well as English voice spoken at native speed slowly The present invention relates to a speech speed conversion device that can be used in various devices, such as an English learning device that converts the voice into an easy-to-hear voice.

The speech speed conversion means that the time axis of an audio signal is compressed to make its reproduction speed faster than the original speed, or conversely, the time axis of the audio signal is expanded to make its reproduction speed lower. Or slower than your speed.

[0003]

2. Description of the Related Art Conventionally, as disclosed in, for example, Japanese Patent Application Laid-Open No. 7-192392, at the time of high-speed reproduction of a VTR, of audio signals read from a video tape,
Speech speed at which the voice signal of the silent section is deleted, the voice signal of the voice section is extended on the time axis based on the pitch period, and the voice of the voice section is reproduced more slowly than the VTR reproduction speed set by the user. Conversion devices are known,
Also, a VTR equipped with such a speech speed conversion device has been put to practical use.

[0004] In such a speech speed conversion device, in recent years, there has been a strong demand for higher quality audio output after processing.

However, in the above-mentioned conventional apparatus, when the reproduced audio signal is a monaural signal, the reproduced signal is a monaural signal, and when the reproduced audio signal is a stereo signal, the audio signal of one of the channels is targeted. Speech rate conversion processing was performed, and speech rate conversion processing for stereo and multi-channel audio signals was not performed, resulting in poor audio output quality.

[0006] On the other hand, for example, it is conceivable to sequentially process voice signals of respective channels or to use a plurality of conventional voice processors to process voice signals of respective channels in parallel. Can be

[0007]

However, in each case, since the audio signal of each channel is independently processed, if the audio signal of each channel is different, the audio signal after the speech speed conversion processing is processed. A difference occurs in the amount of data. As a result, there is a problem that the audio signal to be reproduced cannot be synchronized, giving a sense of incongruity to the user.

The present invention has been made in view of such a problem, and it is an object of the present invention to perform appropriate speech speed conversion processing on multi-channel audio signals while maintaining synchronization of reproduced audio signals. It is an object of the present invention to provide a speech speed conversion device capable of performing the conversion.

[0009]

According to a first aspect of the present invention, there is provided a speech speed converting apparatus for converting input speech signals of a plurality of channels inputted from a speech reproducing apparatus based on a pitch period obtained from the input speech signals. A speech speed conversion device for performing a conversion process, comprising: a pitch cycle calculating means for calculating a pitch cycle common to each channel for each processing section from input audio signals of a plurality of channels; and And a time axis compression / expansion means for time axis compression of the input audio signal of the channel.

According to a second aspect of the present invention, in the speech speed conversion device according to the first aspect, the processing section changes according to the pitch cycle calculated by the pitch cycle calculating means.

According to a third aspect of the present invention, in the speech speed conversion device according to the first or second aspect, the pitch period calculating means includes an adding means for adding input audio signals of a plurality of channels for each processing section, and the adding means includes: The pitch period is calculated from the added input audio signal obtained as described above.

According to a fourth aspect of the present invention, in the speech speed conversion device according to the first or second aspect, the pitch period calculating means detects a channel having a maximum signal strength for each processing section from the input audio signals of the plurality of channels. And a pitch period is calculated from the input audio signal of the channel detected by the maximum signal strength detecting means. The signal strength means, for example, the power of the signal, the average amplitude value, or the cumulative amplitude value.

According to a fourth aspect of the present invention, in the speech speed conversion device according to the first or second aspect, the pitch period calculating means determines a channel having the largest autocorrelation value for each processing section from the input audio signals of the plurality of channels. It is provided with a maximum autocorrelation value detecting means for detecting, and calculates a pitch period from an input audio signal of a channel detected by the maximum autocorrelation value detecting means.

According to a sixth aspect of the present invention, there is provided a speech speed conversion device for performing a speech speed conversion process on input audio signals of a plurality of channels input from an audio reproducing device based on a pitch period obtained from the input audio signals. An adding means for adding input audio signals of a plurality of channels for each first processing section, and determining whether the processing section is a sound section based on the added input audio signal obtained by the adding means. A section determining means for determining whether or not the section is a silent section; a silent section deleting means for deleting an input audio signal of each channel in a processing section determined to be a silent section by the section determining means; Pitch period calculating means for calculating a single pitch period for each of the second processing sections from the added input audio signal determined to be a sound section; Based on the pitch period in which and a time axis compression and expansion means for compressing the time axis an input audio signal of each channel in the processing section.

A speech speed conversion apparatus according to a seventh aspect of the present invention is a speech speed conversion processing means for performing speech rate conversion processing on input audio signals of a plurality of channels input from an audio reproduction apparatus, and the speech speed conversion processing means. A speech speed conversion device comprising an audio memory into which a processed audio signal is written and reading means for reading out the audio signal from the audio memory. Maximum signal strength detection means for detecting the channel having the highest signal strength for each processing section, and the processing section having a sound section based on the input audio signal of the channel detected by the maximum signal strength detection means. Section determining means for determining whether the signal is a silent section or not, and removing the input audio signal of each channel in the processing section determined to be a silent section by the section determining means. Silence section deletion means, pitch cycle calculation means for calculating a single pitch cycle for each second processing section from an input audio signal of a channel determined to be a sound section by the section determination means, and the pitch And a time axis compression / expansion means for time axis compressing the input audio signal of each channel in the processing section based on the pitch period obtained by the period calculation means.

According to an eighth aspect of the present invention, there is provided a speech speed conversion device, comprising: speech speed conversion processing means for performing a speech speed conversion process on a plurality of channels of input audio signals input from an audio reproduction device; A speech speed conversion device comprising an audio memory into which a processed audio signal is written and reading means for reading out the audio signal from the audio memory. A maximum autocorrelation value detecting means for detecting a channel having a maximum autocorrelation value for each processing section, and a processing section having a sound based on an input audio signal of the channel detected by the maximum autocorrelation value detecting means. A section determining means for determining whether the section is a section or a silent section, and an input sound signal of each channel in the processing section determined to be a silent section by the section determining means. Silence section deletion means for removing, pitch cycle calculation means for calculating a single pitch cycle for each second processing section from an input audio signal of a channel determined to be a sound section by the section determination means, And a time axis compression / expansion means for time axis compressing the input audio signal of each channel in the processing section based on the pitch cycle obtained by the pitch cycle calculation means.

According to a ninth aspect of the present invention, in the speech speed conversion device according to any one of the sixth to eighth aspects, the second processing section changes in accordance with the pitch cycle calculated by the pitch cycle calculating means. is there.

[0018]

Embodiments of the present invention will be described below with reference to the drawings. [1] First Embodiment FIG. 1 is a schematic block diagram showing a configuration of a speech speed conversion device according to the present embodiment. In the following description,
The compression ratio is defined as P, the time length (data amount) of an input signal to the time axis compression / expansion unit 14, and the time length (data amount) of an output signal output from the time axis compression / expansion unit 14 with respect to the input signal. Is defined as P / Q. The storage rate of the unread audio signal in the audio memory 15 is the ratio [%] of the storage amount of the unread audio signal to the total data amount of the audio signal that can be stored in the audio memory 15.
Hereinafter, it is simply referred to as an accumulation rate.

In FIG. 1, reference numeral 10 denotes a hard disk drive, which converts a video signal and left and right channel audio signals required for stereo sound reproduction into
It is assumed that the information is recorded together with header information indicating the type of each signal. Note that the left channel and the subsequent right channel audio signals temporally correspond to each other.

For example, when a digital television broadcast encoded by the MPEG system is targeted, a video signal is transmitted by referring to identification information of each packet with respect to a TS (Transport Stream) of a received broadcast signal. Separately extract the video packet and the audio packet having the left channel and right channel audio signals required for stereo reproduction, subject each of the obtained signals to MPEG decoding processing, and perform processing according to the type of each signal. What is necessary is just to record on the hard disk drive 10 with header information. At this time, in the audio signal, a right channel audio signal and a left channel audio signal are alternately recorded one by one.

The signal read from the hard disk drive 10 is stored in the frame memory 11. The signal stored in the frame memory 11 is sent to a signal separating unit 121 provided in the signal analyzing unit 12, where it is separated into a video signal and an audio signal based on the header information. The separated video signal is sent to a monitor (not shown) and displayed there.

On the other hand, the left channel and right channel audio signals separated by the signal separation section 121 are sent to a signal addition section 122.

The signal adding unit 122 sequentially adds the left channel and right channel audio signals input from the signal separating unit 121, and when the audio signals for one frame of both channels are added, the added audio signal Is output.

The section determination section 123 obtains signal strengths such as power, amplitude average value, amplitude accumulation value, and the like of the added audio signal for one frame inputted from the signal addition section 122, based on the obtained values. It is determined whether the input added audio signal is a sound section or a silent section. Here, the power of the signal is used as the signal strength.

The power P is the amplitude value of the addition sound signal in one frame i _0, i _1, ... When i _N-1, obtained by the following equation number 1.

[0026]

(Equation 1)

The average value P of the power obtained by the above equation (1)
Is compared with a preset threshold Th, and it is determined whether the average value P of the power is equal to or larger than the threshold Th (P ≧ Th) or smaller than the threshold Th (P <Th). Then, the section determination unit 1
23, the average value P of the power is equal to or larger than the threshold Th (P ≧ T
In the case of h), a signal indicating that the current frame is a voiced section is output, and when the average power value P is smaller than the threshold Th (P <Th), a signal indicating that the current frame is a voiceless section is output. You.

The silent section deleting section 13 includes a section determining section 123
Based on the input signal, both the left-channel audio signal and the subsequent right-channel audio signal from which the added audio signal of the section determined to be a silent section is generated are deleted, and the sound section Then, the left channel audio signal from which the added audio signal of the section determined to be generated and the subsequent right channel audio signal are output.

As a result, the frame to be deleted as a silent section is a frame in which the audio signals of the left channel and the right channel temporally correspond to each other. Are kept synchronized.

The silent section in the added voice signal is
Since the signal strength of both the left and right channel audio signals is a weak frame, the silent section is appropriately determined for the audio signals of both channels.

On the other hand, as a method for performing the time axis compression / expansion processing with high quality of reproduced sound, there is known a method of time axis compression / expansion processing of an audio signal in units of its pitch cycle.
Here, that method is used.

For this reason, the speech speed conversion device according to the present embodiment includes a pitch cycle calculation unit 124. The pitch cycle calculation section 124 calculates a pitch cycle from the added voice signal input from the signal addition section 122 and outputs the pitch cycle. Here, an autocorrelation is used as a pitch period calculation method.

In the method of calculating the pitch period using the autocorrelation, it is assumed that the signal is time-limited, and the signal exists only within the section of the time length Ts, and the signal is outside the section of the time length Ts. There is a method of using short-time autocorrelation that always obtains autocorrelation as zero. This is a digital signal processing of voice issued by Corona (above)-LRRabiner & R.W.Schaf
As described in er, by Kuki Suzuki -p152-p152, if the audio waveform is represented by a digital audio signal x (n), the short-time autocorrelation value Rn (k) by the above-described method is as follows. become that way.

[0034]

(Equation 2)

Here, Ts is a time section in which a voice signal is assumed to exist, k is a delay time for delaying a voice waveform when calculating a short-time autocorrelation value Rn (k), and T
There is a relationship of s≫k. Then, when the value of k that maximizes the short-time autocorrelation value Rn (k) is obtained in Expression 2, the value becomes the pitch period. The obtained pitch period is sent to the time axis compression / expansion unit 14.

The time axis compression / expansion unit 14 controls the pitch input from the pitch period calculation unit 124 for the voice signal (voice signal of the voiced section) after the voice signal of the voiceless section has been deleted by the voiceless section deletion unit 13. The time axis compression / expansion processing is performed by setting the cycle as a pitch cycle common to the left channel audio signal and the subsequent right channel audio signal. The time axis compression / expansion processing is performed by the pitch period calculation unit 1
The processing is performed in units of processing that change in accordance with the pitch cycle obtained in 24.

In the time axis compression processing in the time axis compression / expansion unit 14, for example, as shown in FIG. 2, a speech waveform of two pitch periods to be processed is cut out, and then a waveform A of one pitch period is extracted. Is multiplied by a weighting coefficient that changes linearly from 1 to 0 to generate a waveform A ′.
Is multiplied by a weighting coefficient that changes linearly from 0 to 1 and the waveform B
Are generated and added together to obtain a waveform C for one pitch period.
Specifically, when the compression ratio is 1.5 times, as shown in FIG.
The waveform of the first pitch cycle and the waveform of the second pitch cycle are compressed into one waveform to be the first output waveform, and then the third pitch cycle is directly used as the second output waveform, and further the fourth pitch cycle is obtained. And the waveform of the fifth pitch cycle may be compressed into one waveform to obtain a third output waveform.

In the time axis compression / expansion processing in the time axis compression / expansion section 14, for example, as shown in FIG.
A waveform A 'is generated by multiplying a waveform A for two pitch periods by a weight coefficient that linearly changes from 0 to 1 and a waveform B for two pitch periods linearly changes from 1 to 0, for example. This is performed by generating a waveform B 'by multiplying by a weight coefficient and adding the waveforms B' to each other to obtain a waveform D and a waveform E for one pitch period.

In the time axis compression / expansion processing based on such a pitch cycle, the same pitch cycle is used for the audio signals of the left channel and the right channel, so that the audio signal after the time axis compression / expansion processing can be used.
Synchronization of the left and right channels is maintained.

The pitch period obtained based on the added audio signal is determined by the difference between the power of the audio signal of the left channel and that of the audio signal of the right channel. Since the pitch becomes closer to the pitch period of the signal, the quality of the reproduced sound of the channel with the higher power is improved. At this time, the quality of the reproduced sound of the channel with the lower power is low, but the effect is small, and the quality of the reproduced sound is generally high.
Even if there is no difference in power between the left channel and right channel audio signals, the pitch period obtained based on the added audio signal is an intermediate value between the pitch periods in the audio signals of both channels. The audio signals of the left and right channels are reproduced with substantially the same quality, and the quality of the reproduced audio is not significantly reduced in one of the channels.

Further, since the arithmetic processing in the pitch cycle calculating section 124 is reduced to the processing amount of one channel for the audio signals of the left and right two channels, the load on the arithmetic processing is reduced.

The audio signal subjected to the time axis compression / expansion processing by the time axis compression / expansion unit 14 is temporarily stored in the audio memory 15 and then stored therefrom at the standard reproduction speed regardless of the compression rate in the time axis compression / expansion processing. Is read.

The compression rate in the time axis compression processing is determined according to the reproduction mode specified by the user and the accumulation rate of the audio memory 15. For this reason, the speech speed conversion device according to the present embodiment includes a storage rate calculation unit 16 that calculates the storage rate of the audio memory 15. The accumulation rate calculated by the accumulation rate calculation unit 16 is sent to the speech speed control unit 17.

The speech speed control unit 17 controls the compression ratio used by the time axis compression / expansion unit 14 based on the reproduction mode and the accumulation ratio set by the user.

For example, assume that the user has set the high-speed playback mode. In this case, the playback speed of the video read from the hard disk drive 10 and output to the monitor (hereinafter, referred to as the set playback speed) is 1.5 times, but the sound is equal to or slower than the set playback speed. Output at speed.

Table 1 shows the relationship between the accumulation ratio and the compression ratio in the high-speed reproduction mode.

[0047]

[Table 1]

The speech speed control unit 17 has a storage ratio / compression ratio table storing the relationship between the storage ratio and the compression ratio shown in Table 1 above, and when the storage ratio is sent from the storage ratio calculation unit 16. Based on this storage rate / compression rate table, the compression rate corresponding to the storage rate sent from the storage rate calculation unit 16 is read and set in the time axis compression / expansion unit 14. (1) When the accumulation rate is 0 to 30% (0 or more and less than 30%) When the accumulation rate is 0 to 30%, the compression rate is set to 1. In this case, the audio signal read from the hard disk drive 10 is temporarily stored in the frame memory 11 and read therefrom at a playback speed (48 kHz) corresponding to the set playback speed magnification factor 1.5.

After the audio signal read from the frame memory 11 is deleted in the silent section by the silent section deletion section 13, the time axis compression / expansion section 14 does not perform the time axis compression / expansion processing, 15 are stored. The audio signal stored in the audio memory 15 is separated into left and right channel audio signals by the L / R separation unit 18 and then D / A conversion units 191 and 1, respectively.
The sampled data is output at 92 at a standard sampling frequency (32 kHz). Therefore, the speech speed of the output sound becomes equal to the speech speed of the output sound when reproduced at the standard reproduction speed (reproduction speed at 1 × speed reproduction).

When there are few signals in the silent section in the audio signal,
Since the writing speed to the audio memory 15 is faster than the data reading speed from the audio memory 15,
, The amount of storage of unread data increases. The speed at which the accumulation amount of the unread audio signal increases is
The faster the sound signal is, the less the signal in the silent section is.
On the other hand, when the number of signals in the silent section increases in the audio signal, the storage amount of unread data in the audio memory 15 is reduced depending on the data amount in the silent section.

Although not described in Table 1, when the accumulation rate of the voice memory 15 is less than 10%, the speech speed control unit 17 prohibits the silent section deleting unit 13 from deleting the silent section. Is output, and then the audio memory 15
When the accumulation rate exceeds 20%, the speech speed control unit 17 outputs a control signal to the silent interval deleting unit 13 to restart the deletion of the silent interval. (2) When the accumulation rate is 30 to 60% (30 or more and less than 60%) When the accumulation rate is 30 to 60%, the compression rate is 1.2.
Is set to In this case, the time axis compression / expansion unit 14
Performs a time axis compression / expansion process on the audio signal such that the ratio of the time length P of the input signal to the time length Q of the output signal becomes 1.2: 1. As a result, the voice speed of the output voice is slightly higher than the voice speed of the voice output at the time of reproduction at the standard reproduction speed (reproduction speed at 1 × speed reproduction). On the other hand, since the amount of voice data in the voiced section input to the voice memory 15 is reduced, the speed at which the amount of unread data stored in the voice memory 15 increases compared to the case of (1) above. And the amount of unread data stored in the audio memory 15 decreases depending on the amount of data in the silent section. (3) When the accumulation rate is 60 to 90% (60 or more and less than 90%) When the accumulation rate is 60 to 90%, the compression rate is 1.4.
Is set to In this case, the time axis compression / expansion unit 14
Performs a time axis compression / expansion process on the input signal so that the ratio of the time length P of the input signal to the time length Q of the output signal is 1.4: 1. As a result, the speech speed of the output voice is higher than in the case (2).

On the other hand, since the data amount of the sound section inputted to the audio memory 15 is further reduced as compared with the case of the above (2), the storage amount of the unread data in the audio memory 15 increases. On the other hand, depending on the amount of data in the silent section, the amount of unread data stored in the audio memory 15 decreases. (4) The accumulation rate is 90 to 100% (90 or more and 100%
When the accumulation rate is 90 to 100%, the compression rate is 1.
Set to 5. In this case, the time axis compression / expansion unit 14
Performs a time axis compression / expansion process on the input signal such that the ratio of the time length P of the input signal to the time length Q of the output signal becomes 1.5: 1. As a result, the speech speed of the output voice is higher than in the case (3).

On the other hand, since the writing speed to the audio memory 15 is equal to the data reading speed from the audio memory 15, the unread data in the audio memory 15 corresponds to the data amount of the silent section deleted by the silent section deleting unit 13. Decrease the amount of accumulation. The speed at which the accumulated amount of unread audio signals in the audio memory 15 decreases increases as the data amount in the silent section increases.

Next, a description will be given of the operation in the slow listening mode in which the sound is reproduced at a speed equal to or lower than the reproduction speed when the reproduction is performed at the standard reproduction speed.

Table 2 shows the relationship between the accumulation rate and the compression rate in the slow listening playback mode.

[0056]

[Table 2]

The speech speed control unit 17 has a storage ratio / compression ratio table storing the relationship between the storage ratio and the compression ratio shown in Table 2 above. Based on this storage rate / compression rate table, the compression rate corresponding to the storage rate sent from the storage rate calculation unit 16 is read and set in the time axis compression / expansion unit 14. Then, similarly to the high-speed playback mode described above, the time axis compression / decompression unit 14
, The time axis compression / expansion processing is performed at a compression rate corresponding to the accumulation rate. [2] Second Embodiment In the first embodiment, a case has been described in which the section determination and the pitch cycle calculation are performed using the added audio signal obtained by adding the audio signals of the left channel and the right channel. In the embodiment, a case will be described in which the section determination and the pitch period calculation are performed using one of the audio signals selected based on the signal strength of each of the audio signals of the left channel and the right channel.

FIG. 5 is a schematic block diagram showing another configuration (signal analysis unit 22) of signal analysis unit 12 in the first embodiment. The same components as those in the first embodiment are denoted by the same reference numerals.

In the figure, reference numeral 221 denotes a signal strength calculation unit which calculates the signal strength of one frame for the left channel and right channel audio signals input from the frame memory 11 via the signal separation unit 121. As the signal strength, as described in the first embodiment, the power of the signal, the average amplitude value, the cumulative amplitude value, and the like can be used.

The signal strength judging section 222 compares the signal strengths of the left and right channel audio signals input from the signal strength calculating section 221 with respect to each other in time, and determines the channel having the stronger signal strength. Is output.

The audio signal selection section 223 deletes the audio signal on the channel side where the signal intensity is determined to be low based on the signal input from the signal intensity determination section 222 and deletes the channel signal where the signal intensity is determined to be high. Output the audio signal of the side. As a result, the output audio signal has a configuration in which the audio signals of the channel having the stronger signal strength among the audio signals of the left channel and the right channel that temporally correspond to each other are connected in frame units.

[0062] The section determination section 123 includes the audio signal selection section 22.
3 is referred to the value of the signal strength previously calculated by the signal strength calculator 221 to determine whether the input voice signal is a voiced section or a silent section. I do.

The signal strength calculated by the signal strength calculation unit 221 is compared with a preset threshold value. When the signal strength value is equal to or greater than the threshold value, a signal indicating that the current frame is a voiced section is output. When the value of the signal strength is smaller than the threshold value, a signal indicating that the current frame is a silent section is output to the silent section deleting unit 13.

As in the first embodiment, based on the signal input from the section determining section 123, the silent section deleting section 13 outputs the audio signal of the left channel of the section determined to be the silent section and the The audio signal of the subsequent right channel is deleted together, and the audio signal of the left channel and the audio signal of the subsequent right channel in the section determined to be a sound section are output.

As a result, the frame deleted as a silent section is a frame in which the audio signals of the left channel and the right channel temporally correspond to each other. Are kept synchronized.

A silent section of an audio signal obtained by selecting a higher signal strength of the left and right channel audio signals in frame units is also a silent section of the audio signal having a lower signal strength. From that
A silent section is appropriately determined for the audio signals of both channels.

On the other hand, the pitch cycle calculation section 124 performs the first processing on the audio signal input from the audio signal selection section 223.
Similarly to the embodiment, the pitch period is calculated using the autocorrelation, and is output to the time axis compression / expansion unit 14.

The time axis compression / expansion unit 14 performs the following based on a single pitch period input from the pitch period calculation unit 124.
For the left channel audio signal and the subsequent right channel audio signal of the section determined to be a sound section,
Perform time axis compression / expansion processing and output.

Thus, the audio signals of the left channel and the right channel are both based on the common pitch period calculated from either the audio signal of the left channel or the audio signal of the right channel selected by the audio signal selector 223. Since the time axis compression / expansion processing is performed, the synchronization of both channels is maintained.

Further, since the pitch period is calculated based on the audio signal of the left channel and the right channel having the higher signal strength, it differs from the pitch period of the voice signal of the lower signal strength, for example. Even if there is, the effect at the time of reproduction is small, and the quality of the reproduced sound as a whole is high.

Further, since the calculation processing in the pitch cycle calculation unit 14 is reduced to the processing amount for one channel for the audio signals of two channels on the left and right, the load on the calculation processing is reduced.

The audio signal which has been subjected to the time axis compression / expansion processing by the time axis compression / expansion unit 14 is temporarily stored in the audio memory 15 and read out therefrom. [3] Third Embodiment In the second embodiment, using one of the audio signals selected based on the signal strength of each of the audio signals of the left channel and the right channel, the section determination and the pitch cycle are performed. Although the case of performing the calculation has been described, in the present embodiment, the section determination and the pitch period calculation are performed using one of the audio signals selected based on the autocorrelation coefficient of each of the left channel and the right channel. The case of calculating is described.

FIG. 6 is a schematic block diagram showing another configuration (signal analysis unit 23) of signal analysis unit 12 in the first embodiment. The same components as those in the first embodiment are denoted by the same reference numerals.

In the figure, reference numeral 231 denotes an auto-correlation coefficient calculation unit, which converts the left channel and right channel audio signals input from the frame memory 11 via the signal separation unit 121 in the first embodiment. The maximum value of the autocorrelation coefficient is calculated by dividing the autocorrelation value Rn (k) obtained by the same method as the pitch period calculation by the autocorrelation value R (0).

The auto-correlation determining section 232 compares the maximum values of the auto-correlation coefficients of the left and right channel audio signals input from the auto-correlation coefficient calculating section 231 with each other in time. As a result, a signal indicating the channel with the largest autocorrelation coefficient is output.

The audio signal selection section 233 deletes the audio signal on the channel side determined to have low correlation (the maximum value of the autocorrelation coefficient is small) based on the signal input from the autocorrelation determination section 232. , And outputs a channel-side audio signal determined to have a high correlation (the maximum value of the autocorrelation coefficient is large). As a result, the output audio signal has a configuration in which the audio signals of the channel having the higher correlation among the audio signals of the left channel and the right channel that temporally correspond to each other are connected in frame units.

On the other hand, the signal adding section 234
The audio signals of the left channel and the right channel input from 21 are sequentially added, and when data of one frame of both channels is added, the added audio signal is output.

The section determination section 123 obtains the signal strength of the voice signal input from the signal addition section 234, and based on the obtained value, the input voice signal is a voiced section or a silent section. And outputs the signal to the silent section deletion unit 13.

As in the first embodiment, based on the signal input from the section determining section 123, the silent section deleting section 13 outputs the audio signal of the left channel of the section determined to be a silent section and the The audio signal of the subsequent right channel is deleted together, and the audio signal of the left channel and the audio signal of the subsequent right channel in the section determined to be a sound section are output.

As a result, the frame to be deleted as a silent section is a frame in which the audio signals of the left channel and the right channel temporally correspond to each other. Are kept synchronized.

The silent section in the added voice signal is
Since the signal strength of both the left and right channel audio signals is a weak frame, the silent section is appropriately determined for the audio signals of both channels.

On the other hand, pitch cycle calculating section 124 obtains a pitch cycle corresponding to the maximum value of the auto-correlation coefficient previously calculated by auto-correlation coefficient calculating section 231 and outputs it to time axis compression / expansion section 14. .

The time axis compression / expansion unit 14 performs the following based on a single pitch period input from the pitch period calculation unit 124.
A time axis compression / expansion process is performed on the audio signal of the left channel and the audio data of the subsequent right channel in the section determined to be a sound section, and the result is output.

Thus, the audio signals of the left channel and the right channel are both based on the common pitch period calculated from the audio signal of either the left channel or the right channel selected by the audio signal selection section 233. Since the time axis compression / expansion processing is performed, the synchronization of both channels is maintained.

Since the pitch period is calculated based on the audio signal having the higher correlation between the left channel and the right channel, it differs from the pitch period of the audio signal having the lower correlation, for example. Even in this case, the effect at the time of reproduction is small, and the quality of the reproduced sound is generally high.

The audio signal subjected to the time axis compression / expansion processing by the time axis compression / expansion unit 14 is temporarily stored in the audio memory 15 and read out therefrom.

In each of the above embodiments,
Silence periods are deleted and time axis compression / expansion processing is performed for all the input left and right channel audio signals, but the left and right channel audio signals temporally corresponding to each other are subtracted in frame units. Alternatively, a configuration may be provided in which it is determined whether or not the silence section is to be deleted and the time axis compression / expansion processing is performed in accordance with the obtained difference value. That is, when the difference value is equal to or greater than the predetermined threshold, the two audio signals are different, so that the silent section is not deleted and the time axis compression / expansion processing is not performed. Since they are similar, a silent section is deleted and a time axis compression / expansion process is performed.

In this case, since the silence section is deleted and the time base compression / expansion processing is performed only on the frame in which the audio signals of the left and right channels corresponding to each other in time are similar, the common processing is performed on the left and right channels. Even if the pitch period is used, the deterioration of the sound quality due to this is extremely small.

In each of the above embodiments,
The case where the audio data read from the hard disk drive 10 has two channels has been described, but the same effect can be obtained in the case of three channels or more.

[0090]

According to the present invention, since the time axis compression / expansion processing is performed on a plurality of input audio signals at a single pitch period, the speech speed conversion is performed while maintaining the synchronization of the reproduced audio signals. Processing can be performed.

[Brief description of the drawings]

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

FIG. 2 is an explanatory diagram illustrating the principle of time axis compression processing in the speech speed conversion device of FIG.

FIG. 3 is an explanatory diagram illustrating an example of a time axis compression process using the principle of the time axis compression process of FIG. 2;

FIG. 4 is an explanatory diagram illustrating the principle of time axis compression / expansion processing in the speech speed conversion device of FIG. 1;

FIG. 5 is a schematic block diagram illustrating a configuration of a signal analysis unit of a speech speed conversion device according to a second embodiment of the present invention.

FIG. 6 is a schematic block diagram illustrating a configuration of a signal analysis unit of a speech speed conversion device according to a third embodiment of the present invention.

[Explanation of symbols]

 10: Hard disk drive 11: Frame memory 12: Signal analysis unit 121: Signal separation unit 122: Signal addition unit 123: Section determination unit 124: Pitch cycle calculation unit 13: Silence section deletion unit 14: Time axis compression / decompression unit 15: Voice Memory 16: Accumulation rate calculation unit 17: Speech speed conversion unit 18: L / R separation unit 191: D / A converter 192: D / A converter

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) G09B 19/06 G10L 9/00 D 21/04 9/08 B

Claims (9)

[Claims] 1. A speech speed conversion device for performing a speech speed conversion process on an input speech signal of a plurality of channels input from an audio reproduction device based on a pitch period obtained from the input speech signal, wherein the input speech of the plurality of channels is provided. Pitch period calculating means for calculating a pitch period common to each channel from the signal for each processing section, and time axis compression / expansion means for time axis compressing the input audio signal of each channel in the processing section based on the obtained pitch period. A speech speed conversion device comprising: 2. The speech speed conversion device according to claim 1, wherein said processing section changes according to a pitch period calculated by said pitch period calculation means. 3. The pitch cycle calculation means includes an addition means for adding the input audio signals of the plurality of channels for each processing section, and calculates a pitch cycle from the added input audio signal obtained by the addition means. 2. The method according to claim 1, wherein
Or the speech speed converter according to 2. 4. A maximum signal strength detecting means for detecting a channel having a maximum signal strength in each of the processing sections from the input audio signals of the plurality of channels, wherein the pitch cycle calculating means comprises: 3. The speech speed conversion device according to claim 1, wherein a pitch period is calculated from an input voice signal of the channel detected by the means. 5. A maximum autocorrelation value detecting means for detecting a channel having a maximum autocorrelation value for each processing section from the input audio signals of the plurality of channels, wherein the pitch period calculating means comprises: 3. The speech speed conversion device according to claim 1, wherein a pitch period is calculated from an input audio signal of the channel detected by the correlation value detection means. 6. A speech speed conversion device for performing speech speed conversion processing on input speech signals of a plurality of channels input from an audio reproduction device based on a pitch period obtained from the input speech signals, wherein the input speech of the plurality of channels is provided. Adding means for adding a signal for each first processing section, and section determination for determining whether the processing section is a sound section or a silent section based on the added input audio signal obtained by the addition means Means, a silent section deleting means for deleting an input audio signal of each channel in a processing section determined to be a silent section by the section determining means, and a section having a sound is determined by the section determining means. Pitch period calculating means for calculating a single pitch period for each second processing section from the added input audio signal; and Speech speed conversion apparatus characterized by an input audio signal for each channel and a time axis compression and expansion means for compressing the time axis that. 7. A speech rate conversion processing means for performing speech rate conversion processing on input audio signals of a plurality of channels input from an audio reproduction apparatus, and an audio memory in which the audio signal processed by the speech rate conversion processing means is written. And a reading means for reading an audio signal from the audio memory, wherein the speech speed conversion processing means outputs a maximum signal intensity for each of the first processing sections from the input audio signals of the plurality of channels. Maximum signal strength detection means for detecting the channel having one, and whether the processing section is a sound section or a silent section based on the input audio signal of the channel detected by the maximum signal strength detection means. A section determining means for determining; a silent section deleting means for deleting an input audio signal of each channel in a processing section determined to be a silent section by the section determining means; Pitch period calculating means for calculating a single pitch period for each second processing section from an input audio signal of a channel determined to be a sound section in a stage, and a pitch obtained by the pitch period calculating means A speech speed conversion device comprising: a time axis compression / expansion means for time axis compression of an input audio signal of each channel in the processing section based on a cycle. 8. A speech speed conversion processing means for performing speech speed conversion processing of input audio signals of a plurality of channels input from an audio reproduction device, and an audio memory in which the audio signal processed by the speech speed conversion processing means is written. Reading means for reading out a voice signal from the voice memory, wherein the voice speed conversion processing means generates a maximum autocorrelation for each first processing section from the input voice signals of the plurality of channels. Maximum autocorrelation value detecting means for detecting a channel having a value, and whether the processing section is a sound section or a silent section based on the input audio signal of the channel detected by the maximum autocorrelation value detecting means. A section determining means for determining whether or not an input audio signal of each channel in a processing section determined to be a silent section by the section determining means; Means for calculating a single pitch cycle for each second processing section from an input audio signal of a channel determined to be a sound section by means, and a pitch obtained by the pitch cycle calculating means. A speech speed conversion device, comprising: time axis compression / expansion means for time axis compression of an input audio signal of each channel in the processing section based on a cycle. 9. The speech speed conversion device according to claim 6, wherein the second processing section changes according to a pitch cycle calculated by the pitch cycle calculation means. .