JP2008107381A - Speaking speed converting device and speaking speed converting control method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a speaking speed converting device in which a speaking speed conversion magnification rate of a voice signal is set to more than 1, only for a voice part which is hard to hear in audibility, that is, voice is hardly delayed from an image by promptly reproducing the voice. <P>SOLUTION: Sonant/silence of an input voice signal is discriminated in a sonant/silence discrimination section, and from current input voice signal total time and output voice signal total time, it is determined whether or not, a silence signal is rejected by a signal rejection determination section. The silence signal which is determined to be rejected, is rejected in a silence signal rejection section, and extension processing is performed on the other signals with the speaking speed conversion magnification rate determined by previous processing in a speed conversion section, without changing a voice pitch. The signals are accumulated in an output voice control section and sequentially forwarded to an output signal terminal, and a conversion speed control section calculates the speaking speed conversion magnification rate from difference value between a voice accumulation amount of current processing time and the voice accumulation amount of previous processing time and output to the speed conversion section. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a speech speed conversion technique that makes it easy for a listener to hear a sound of a television device, a telephone, a mobile phone, or the like by slowly reproducing the sound without changing the sound pitch.

  As a technique for reproducing a voice signal slowly without changing the voice pitch of the speaker and facilitating listening to the listener, there is a voice signal conversion technique. The principle of operation of a conventional speech speed conversion device to which this technology is applied is to determine whether a voice signal is voiced or silent and to change the voice pitch only for a voice signal determined to be voiced. The time delay of the playback sound generated due to the time extension process is absorbed by discarding the audio signal determined to be silent ( For example, see Patent Documents 1 and 2).

JP 2002-297200 A JP 2001-109499 A

However, if the audio signal input to the speech rate conversion device includes background sound in addition to speaker audio, such as on-site relay audio or street interview recorded audio in a TV broadcast news program, CM (Commercial Message When the background sound is always sounding, such as), it is difficult to determine that the audio signal is silent, and it is substantially difficult to execute the discarding process of the silent signal. In such a case, due to the effect of the time expansion processing on the audio signal, the audio delay increases unilaterally with respect to the video, and the TV viewer feels uncomfortable.
Further, the audio signal corresponding to the time delay needs to be reproduced while being held in a separate memory, resulting in a disadvantage that the scale and cost of the apparatus are increased.

  As one method for solving this problem, Patent Document 2 discloses that when the storage amount of a memory for storing a speech signal subjected to compression / decompression processing is 0% or more and less than 20%, the speech speed of the speech signal is as follows. Conversion magnification (value obtained by dividing the time length of the input audio signal by the time length of the output audio signal) is 0.6 times, 0.7 times when 20% or more and less than 40%, and 40% or more and less than 60% 0.8 times, 0.9 times when 60% or more and less than 80%, 1.0 times when 80% or more and 100% or less (no compression / decompression processing), etc. A method for changing the speech speed conversion magnification of an audio signal is described.

  However, the above method is mainly intended to prevent the failure of the memory, and there is a problem that it becomes difficult to slowly reproduce the speaker voice as the storage amount of the memory increases.

  Some sports news programs, for example, have background music (BGM) in addition to the announcer's speaker voice. In this case, the voice signal without any silence is continuously converted to the speech speed. It will be input to the device. When the above method is applied to such a program, the increase in the delay time of the sound is reduced as the amount of accumulated memory increases, but there is no silence at all. Will reach 80% or more. This means that, for example, if the memory for storing the audio signal can store the audio signal for 2 seconds, the audio will be delayed by 1.6 seconds (= 2 seconds × 80%) with respect to the video. However, there is a problem that deviates from the original purpose of the speech speed converting apparatus, which is to provide a voice that is easy to listen to the listener by slowly converting the voice.

  The present invention has been made to solve such a problem, and calculates an increase in speech delay time per unit time, and a plurality of speech rate conversion magnifications prepared in advance according to the calculation result. To provide a speech speed conversion device that suppresses a delay in audio with respect to a video and reduces a sense of discomfort that a TV viewer feels due to a difference between the video and the audio by performing compression / decompression processing at any time in units of time. It is an object.

  Also, with respect to the latter problem, only the audio part that is difficult to hear due to the audibility, the speech speed conversion magnification of the audio signal is made larger than 1.0, that is, the process of reproducing the sound quickly is performed. It is an object of the present invention to provide a speech speed conversion device that hardly causes a delay time of voice.

The speech speed conversion device according to the present invention includes a voice input terminal, a voice / silence discrimination section, a signal discard judgment section, a silence signal discard section, a speed conversion section, an output voice control section, a voice output terminal, and a conversion speed control section. ,
The voice / silence discrimination unit discriminates whether the voice signal input from the voice input terminal is voiced or silent,
The signal discard judgment unit does not discard the voice signal by the discrimination signal from the voice / silence discrimination unit, and the silence signal calculates the total time of the input voice signal from the voice input terminal and outputs it from the output voice control unit Calculate the total time of the audio signal, subtract the total audio input time from the total audio output time, and determine whether to discard the silence signal based on the result,
The silent signal discarding unit also inputs the audio signal, discards the silent signal with the discard signal from the signal discard determination unit, and passes the other signals to the speed conversion unit,
The speed conversion unit performs a voice signal expansion process on the voice signal passed from the silence signal discarding unit without changing the voice pitch at the conversion rate of the voice speed from the conversion speed control unit,
The output audio control unit accumulates the audio signal subjected to the time compression / decompression processing from the speed conversion unit in the audio accumulation memory, and discards the signal as the time length of the output audio signal every time the audio signal is input. Output to the audio output terminal the audio signals waiting to be output that have been passed to the discriminator and stored in the audio storage memory, in order from the oldest to the audio output terminal, and count the number of output audio samples. To the conversion speed control unit
The conversion speed control unit calculates a difference value between the current voice accumulation amount received from the output voice control unit and the voice accumulation amount at the previous processing, and divides this difference value by unit time to obtain a voice delay increase degree. , And the speech speed conversion magnification corresponding to the degree of increase in voice delay is output to the speed conversion unit.

Further, the speech speed conversion control method according to the present invention includes:
A voice / silence discrimination step for discriminating whether the input voice signal from the voice input terminal is voiced or silent, and outputting a judgment signal;
Whether or not the sound signal is discarded by the determination signal from the sound / silence determination step, and the sound signal subtracts the total time of sound input from the total sound output time at the present time, and whether to discard the sound signal according to the result A signal discarding determination step for determining whether or not
The audio signal is input, the silence signal is discarded by the discard signal from the signal discard determination step, and other signals are sent to the silence signal discard step and the silence signal discard step that are passed to the speed conversion step of the next processing. On the other hand, a speed conversion step for performing speech signal expansion processing at a speech speed conversion magnification determined in the previous processing without changing the voice pitch,
The audio signal subjected to the time compression / decompression processing from the speed conversion step is accumulated in the audio accumulation memory, and every time the audio signal is input, the time length is set as the time length of the output audio signal, and the audio used in the signal discard determination step Output as the original data for the total number of output times, and output the standby audio signals stored in the audio storage memory to the audio output terminal in order from the oldest one, and count the number of output audio samples And an output audio control step for outputting the current audio accumulation amount at regular intervals,
Calculate the difference between the current audio accumulation amount output from the output audio control step and the audio accumulation amount at the previous processing, and divide this difference value by unit time to calculate the audio delay increase degree. It has a conversion speed control step for outputting a speech speed conversion magnification corresponding to the degree of increase in delay as a speech speed conversion magnification used in the speed conversion step.

  According to the speech speed conversion device of the present invention, the speech speed conversion magnification in the time compression / decompression process can be switched and operated at any time per unit time. For example, when applied to a television receiver, This has the effect of suppressing voice delays.

The best mode for carrying out the present invention will be described below.
Embodiment 1 FIG.
A speech speed converting apparatus according to Embodiment 1 of the present invention will be described with reference to the drawings. FIG. 1 is a functional block diagram showing the configuration of the speech rate conversion apparatus according to Embodiment 1 of the present invention. In the figure, 1 is an audio input terminal, 2 is a sound / silence discrimination unit, 3 is a signal discard determination unit, 4 is a silence signal discard unit, 5 is a speed conversion unit, 6 is an output audio control unit, and 7 is an audio output terminal. , 8 is a conversion speed control unit.

Next, the operation will be described with reference to FIG.
The voice signal is input from the voice input terminal 1 to the speech rate conversion device, and passed to the voice / silence determination unit 2, the signal discard determination unit 3, and the silence signal discard unit 4. The voice / silence determination unit 2 determines whether the input voice signal is voiced or silent, and passes the determination result to the signal discard determination unit 3.

When the determination result passed from the sound / silence determination unit 2 is sound, the signal discard determination unit 3 passes an instruction signal not to discard the signal to the silence signal discard unit 4.
On the other hand, if the determination result received from the voice / silence determination unit 2 is silent, it is determined whether or not the silent signal can be discarded, and if it is determined to be possible, an instruction to discard the signal is given. When it is determined that the signal is not possible, an instruction signal that does not discard the signal is passed to the silent signal discarding unit 4.
Here, the operation of the signal discard determination unit 3 will be described with reference to the flowchart of FIG.

FIG. 2 shows a processing flowchart of the signal discard determination unit 3. Reference numerals 3a, 3b, 3c, 3e, and 3f are processing blocks, and reference numeral 3d is a determination block.
As for the number of voice input hours used in the processing block 3b described later and the number of voice output hours used in the processing block 3c, both the values at the start of the operation of the speech speed converting apparatus, that is, the initial values are both 0. It is. Next, the operation will be described.

In the processing block 3a, the time length of the input voice signal received from the voice input terminal 1 is added to the number of voice input hours to calculate the total time of the input voice signal.
In the processing block 3b, the time length of the output audio signal received from the output audio control unit 6 to be described later is added to the number of audio output hours to calculate the total time of the output audio signal.

In the processing block 3c, the audio input time is subtracted from the audio output time.
In the determination block 3d, when the subtraction result is a positive value, in the processing block 3e, an instruction signal capable of discarding the signal is sent to the silent signal discarding unit 4 and the processing is ended.
On the other hand, when the subtraction result is a negative value or 0, in the processing block 3f, an instruction signal indicating that the signal cannot be discarded is sent to the silent signal discarding unit 4, and the processing is terminated.

A series of operations in the signal discard determination unit 3 is performed by repeatedly discarding the silent part when the voice signal input to the speech speed conversion device is a silent signal or a voice signal including many silent parts. This is executed in order to avoid losing the voice signal output from the speech speed conversion device.
The operation of the signal discard determination unit 3 has been described above. Subsequently, the description returns to a series of operations of the speech speed conversion apparatus of FIG.

  When the content of the instruction signal received from the signal discard determination unit 3 is an instruction content that does not discard the signal, the silent signal discard unit 4 directly converts the voice signal input from the voice input terminal 1 into the speed conversion unit 5. On the other hand, if the instruction content is to discard the signal, the voice signal input from the voice input terminal 1 is discarded and not sent to the speed conversion unit 5.

  The speed conversion unit 5 performs time compression / decompression processing on the audio signal received from the silence signal discard unit 4 without changing the audio pitch, and passes the audio signal generated as a result of the processing to the output audio control unit 6. In addition, as the speech speed conversion magnification of the time compression / decompression process in the speed conversion unit 5, a plurality of patterns of 1.0 times or less, such as 1.0 times, 0.9 times, and 0.8 times, are prepared in advance. The speech speed conversion magnification of the time compression / decompression processing at the start of the operation of the speech speed conversion device is the lowest conversion magnification, in this example, 0.8 times. Which speech speed conversion magnification is selected in the subsequent processing? The conversion speed control unit 8 decides. The method for determining the speech speed conversion magnification will be described in detail in the operation description of the conversion speed control unit 8 described later.

The output voice control unit 6 passes the voice signal received from the speed conversion unit 5 to the voice output signal terminal 7, and the voice signal is output from the speech speed conversion device.
Here, the operation of the output voice control unit 6 will be described with reference to FIG. FIG. 3 is a functional block diagram showing the configuration of the output sound control unit 6. Reference numeral 6a denotes an output sound number calculation unit, 6b a converted sound storage unit, and 6c a delay time calculation unit.
Next, the operation will be described.

  Each time an audio signal subjected to time compression / decompression processing is input from the speed converter 5, the output audio number calculator 6 a passes the time length to the signal discard determination unit 3 as the time length of the output audio signal. The converted voice storage unit 6b stores the voice signal that has been subjected to the time compression / decompression processing received from the speed conversion unit 5 in a voice storage memory prepared in advance. Then, the audio signals stored in the audio storage memory and waiting for output are output to the audio output signal terminal 7 in order from the oldest one, and finally the audio signal is output from the speech speed conversion device.

The delay time calculation unit 6c counts the number of audio samples output from the converted audio accumulating unit 6b to the audio output signal terminal 7, and passes it to the conversion speed control unit 8 as a current audio accumulation amount at a certain period. Here, the fixed period is, for example, 1 second, and in the case of an audio signal with a sampling frequency of 48 kHz, every time 48000 samples of audio samples are output from the converted audio storage unit 6b to the audio output signal terminal 7, Become.
The above is the description of the operation of the output sound control unit 6.

  Next, the operation of the conversion speed control unit 8 that determines the speech speed conversion magnification of the voice signal in the speed conversion unit 5 will be described with reference to the flowchart of FIG.

  FIG. 4 shows a processing flow chart of the conversion speed control unit 8, wherein 8a, 8b, 8e, 8f, 8g, and 8h are processing blocks, and 8c and 8d are determination blocks. Next, the operation will be described.

  In the processing block 8a, a difference value between the current sound accumulation amount received from the output sound control unit 6 and the sound accumulation amount recorded in the previous process is calculated.

In the processing block 8b, the difference value is divided by the unit time (in the following description, this division result is referred to as a voice delay increase degree). Here, the unit time is the same as the period of receiving the current voice accumulation amount from the output voice control unit 6, and is every 1.0 seconds in the example shown in the operation explanation of the output voice control unit 6.
As an example of calculating the audio delay increase,
For example, when the previous voice accumulation amount is 0.1 seconds, the current voice accumulation amount is 0.3 seconds, and the unit time is 1.0 second,
Increase in voice delay = (0.3 sec-0.1 sec) /1.0 sec = 0.2
When the previous voice accumulation amount is 0.3 seconds, the current voice accumulation amount is 0.1 seconds, and the unit time is 1.0 second,
Increase in voice delay = (0.1 sec-0.3 sec) /1.0 sec = -0.2
It becomes.

In the decision block 8c, when the degree of increase in the voice delay is 0.1 or more, in the processing block 8f, an instruction signal for setting the speech speed conversion magnification of the time compression / decompression process to 1.0 is sent to the speed converter 5. Then, the process proceeds to the processing block 8h.
On the other hand, if the voice delay increase degree is less than 0.1 in the decision block 8c, the process proceeds to the decision block 8d.

In the decision block 8d, if the degree of increase in voice delay is 0.05 or more, the processing block 8g sends an instruction signal for setting the speech speed conversion magnification of the time compression / decompression process to 0.9 times to the speed converter 5. Then, the process proceeds to the processing block 8h.
On the other hand, if the voice delay increase degree is less than 0.05 in the decision block 8d, the speed conversion unit 5 sends an instruction signal for setting the speech speed conversion magnification of the time compression / decompression process to 0.8 in the processing block 8e. To process block 8h.

  Finally, in the processing block 8h, the current voice accumulation amount received from the output voice control unit 6 is recorded, and the process is terminated.

The above is the description of a series of operations of the speech speed converting apparatus according to the first embodiment of the present invention.
Incidentally, during the operation description, an example of the sound / silence determination unit 2 for determining whether sound is present or not is disclosed in detail in Japanese Patent Application Laid-Open No. 2002-297200 and the like as described in the background art. . An example of a method for expanding an audio signal without changing the audio pitch in the speed conversion unit 5 is disclosed in detail in Japanese Patent Application Laid-Open No. 2001-109499.

  As described above in detail, according to the first embodiment of the present invention, it is possible to switch the speech speed conversion magnification in the time compression / decompression processing of the speech speed conversion apparatus at every unit time and to operate the video. This has the effect of suppressing voice delays.

Other Embodiments
As described above, in the detailed description of the first embodiment of the present invention, the conversion speed control unit 8 that switches the speech speed conversion ratio according to the degree of increase in the voice delay is used as the method for determining the speech speed conversion ratio of the audio signal. When the voice delay increase degree satisfies a certain condition, the conversion speed control unit A81 that does not change the speech speed conversion magnification from the previous magnification may be used.
Next, the operation of the conversion speed control unit A81 that determines the speech speed conversion magnification of the voice signal in the speed conversion unit 5 will be described with reference to the flowchart of FIG.

  FIG. 5 shows a processing flowchart of the conversion speed control unit A81, in which 81a, 81b, 81f, 81g, 81h, 81i, and 81j are processing blocks, and 81c, 81d, and 81e are determination blocks. Next, the operation will be described.

In the processing block 81a, a difference value between the current voice accumulation amount received from the output voice control unit 6 and the voice accumulation amount recorded in the previous process is calculated.
In the processing block 81b, the difference value is divided by the unit time to calculate the audio delay increase degree.

In the decision block 81c, when the degree of increase in voice delay is 0.1 or more, in the processing block 81i, an instruction signal for setting the speech speed conversion magnification of the time compression / decompression process to 1.0 is sent to the speed conversion unit 5. Then, the process proceeds to the processing block 81j.
On the other hand, when the audio delay increase degree is less than 0.1 in the decision block 81c, the process proceeds to the decision block 81d.

In the decision block 81d, when the degree of increase in voice delay is 0.05 or more, in the processing block 81h, an instruction signal for setting the speech speed conversion magnification of the time compression / decompression process to 0.9 is sent to the speed converter 5. Then, the process proceeds to the processing block 81j.
On the other hand, in the determination block 81d, when the audio delay increase degree is less than 0.05, the process proceeds to the determination block 81e.

In the decision block 81e, when the degree of increase in the voice delay is −0.1 or more, in the processing block 81g, an instruction signal that does not change the speech speed conversion magnification of the time compression / expansion processing from the previous magnification is sent to the speed conversion unit 5. To process block 81j.
On the other hand, when the voice delay increase degree is less than −0.1 in the decision block 81e, in the processing block 81f, an instruction signal for setting the speech speed conversion magnification of the time compression / decompression process to 0.8 is sent to the speed conversion unit. 5 and proceeds to processing block 81j.

  Finally, in the processing block 81j, the current voice accumulation amount received from the output voice control unit 6 is recorded, and the process is terminated.

As a feature of the conversion speed control unit A81, in the conversion speed control unit 8 of the first embodiment shown in FIG. 4, the speech speed conversion magnification is set to 0.8 when the voice delay increase is less than 0.05. If the threshold is lowered to less than -0.10, and the voice delay increase takes a value between -0.10 and less than 0.05, it is set as a dead zone, and the speech rate conversion magnification is not changed. This is the point. The reason for this is that, for example, in the case where a plurality of speakers are in discussion, the voice accumulation amount may decrease for a moment in the conversation and start to increase again. If the conversion magnification is returned to 0.8 times, which is the slowest speed, the audio delay with respect to the video is prevented from increasing again.
The above is the description of the operation of the conversion speed control unit A81.

  As described above in detail, according to another embodiment 1 in which the conversion speed control unit A81 is used instead of the conversion speed control unit 8, the speech speed conversion magnification of the time compression / decompression process is switched at any time per unit time. In addition to the effect of suppressing the audio delay with respect to the video, it is possible to maintain the current speed against a minute increase / decrease in the amount of accumulated audio, further suppressing the audio delay with respect to the video. There is an effect that can be done.

Embodiment 2. FIG.
A speech speed converting apparatus according to the best mode 2 for carrying out the present invention will be described with reference to the drawings. FIG. 6 is a functional block diagram showing the configuration of the speech rate conversion apparatus according to Embodiment 2 of the present invention, in which 1 is a voice input terminal, 21 is a voice / silence discrimination unit A, 3 is a signal discard judgment unit, 4 is a silent signal discarding unit, 51 is a speed conversion unit A, 6 is an output audio control unit, 7 is an audio output terminal, and 82 is a conversion speed control unit B.

Next, the operation will be described with reference to FIG.
The voice signal is input from the voice input terminal 1 to the speech rate conversion device, and passed to the voice / silence discrimination unit A21, the signal discard determination unit 3, and the silence signal discard unit 4. The voice / silence determination unit A21 determines whether the input voice signal is voiced or silent, and passes the determination result to the signal discard determination unit 3. Also, the level of the input audio signal determined to be sound is calculated, and the result is passed to the conversion speed control unit B82.

The signal discard determination unit 3 passes an instruction signal not to discard the signal to the silence signal discard unit 4 when the determination result delivered from the sound / silence determination unit A21 is sound. On the other hand, if the determination result received from the voice / silence determination unit A21 is silent, it is determined whether or not the silent signal can be discarded, and if it is determined to be possible, an instruction to discard the signal is given. When it is determined that the signal is not possible, an instruction signal that does not discard the signal is passed to the silent signal discarding unit 4.
The signal discard determination unit 3 is the same as that described in detail with reference to FIG. 2 in the description of the first embodiment of the present invention, and thus the description thereof is omitted.

  If the content of the instruction signal received from the signal discard determination unit 3 is an instruction content that does not discard the signal, the silent signal discard unit 4 passes the input voice signal as it is to the speed conversion unit A51, If the instruction content is to be discarded, the voice signal input from the voice input terminal 1 is discarded and not sent to the speed conversion unit A51.

  The speed conversion unit A51 performs time compression / decompression processing on the audio signal received from the silence signal discarding unit 4 without changing the audio pitch, and passes the audio signal generated as a result of the processing to the output audio control unit 6. In addition, a plurality of patterns such as 1.1 times, 1.0 times, 0.9 times, 0.8 times, etc. are prepared in advance as the speaking speed conversion magnification of the time compression / decompression process in the speed converting unit A51. The speech speed conversion magnification of the time compression / decompression processing at the start of the operation of the speed converter is the lowest conversion magnification, 0.8 in this example, and it is converted which speech speed conversion magnification is selected in the subsequent processing. The speed controller B82 determines. The method for determining the speech speed conversion magnification will be described in detail in the operation description of the conversion speed control unit B82 described later.

The output voice controller 6 passes the voice signal received from the speed converter A51 to the voice output signal terminal 7, and the voice signal is output from the speech speed converter.
The output audio control unit 6 is the same as that described in detail with reference to FIG. 3 in the description of the first embodiment of the present invention, and its configuration and operation description are omitted.

  Next, the operation of the conversion speed control unit B82 for determining the speech speed conversion magnification of the voice signal in the speed conversion unit A51 will be described with reference to the flowchart of FIG.

  FIG. 7 shows a processing flowchart of the conversion speed control unit B82, in which 82a, 82b, 82e, 82g, 82h, 82i, 82j, and 82k are processing blocks, and 82c, 82d, and 82f are determination blocks. Next, the operation will be described.

In the processing block 82a, a difference value between the current voice accumulation amount received from the output voice control unit 6 and the voice accumulation amount recorded in the previous process is calculated.
In the processing block 82b, the difference value is divided by the unit time to calculate the audio delay increase degree.

  In the decision block 82c, when the audio delay increase degree is 0.1 or more, the process proceeds to the processing block 82e. On the other hand, if the voice delay increase degree is less than 0.1 in the decision block 82c, the process proceeds to the decision block 82d.

In the decision block 82d, if the degree of increase in voice delay is 0.05 or more, in the processing block 82h, an instruction signal for setting the speech speed conversion magnification of the time compression / decompression process to 0.9 times is sent to the speed converter A51. Then, the process proceeds to the processing block 82k.
On the other hand, if the voice delay increase degree is less than 0.05 in the decision block 82d, the processing block 82g sends an instruction signal for setting the speech speed conversion magnification of the time compression / decompression process to 0.8 times the speed conversion unit A51. To process block 82k.

In the processing block 82e, a difference value between the level of the input voice signal determined to be voice received from the voice / silence discrimination unit A21 and a predetermined silence threshold level is calculated.
The silence threshold level here is a value of a level at which the audio signal is determined to be silent. For example, when the maximum value of the signal level is 0 dB in decibel notation, the level value to be determined as silence is −45 dB. . When the signal level is used for the sound / silence determination unit in the sound / silence determination unit A21, the value for determining the level of silence is changed to be received from the sound / silence determination unit A21, The received value may be substituted for the silence threshold level.

In the decision block 82f, when the level difference value calculated in the processing block 82e is 5 dB or less, in the processing block 82j, an instruction signal for setting the speech speed conversion magnification of the time compression / decompression process to 1.1 times is sent to the speed conversion unit. Send to A51 and proceed to process block 82k.
On the other hand, if the level difference value is larger than 5 dB in the decision block 82f, the processing block 82i sends an instruction signal for setting the speech speed conversion magnification of the time compression / expansion process to 1.0 to the speed converter A51. Proceed to process block 82k.

  Finally, in the processing block 82k, the current voice accumulation amount received from the output voice control unit 6 is recorded, and the process is terminated.

As a feature of the conversion speed control unit B82, in the conversion speed control unit 8 of the first embodiment shown in FIG. 1, the speech speed conversion magnification is set to 1.0 when the voice delay increase degree is 0.1 or more. In other words, the speech speed conversion magnification is changed to 1.1 only when the difference between the input speech signal level and the silence threshold level is 5 dB or less when the speech delay increase degree is 0.1 or more. The reason is that, in order to reduce the audio delay with respect to the video, only an audio signal that is difficult to hear and is slightly higher than the silence threshold level is quickly reproduced.
The above is the description of the operation of the conversion speed control unit B82.

  The above is the description of a series of operations of the speech speed converting apparatus according to the second embodiment of the present invention. Incidentally, during the operation description, an example of the sound / silence determination unit 2 for determining whether sound is present or not is disclosed in detail in Japanese Patent Application Laid-Open No. 2002-297200 and the like as described in the background art. . An example of a method for expanding an audio signal without changing the audio pitch in the speed converter A51 is disclosed in detail in Japanese Patent Application Laid-Open No. 2001-109499.

  As described above in detail, according to the second embodiment of the present invention, it is possible to switch the speech speed conversion magnification of the time compression / decompression process at every unit time, and to suppress the audio delay with respect to the video. effective.

  Further, according to the second embodiment of the present invention, it is possible to increase the speech speed conversion magnification of the time compression / expansion processing of the audio signal which is difficult to hear from the viewpoint of hearing from 1.0 times to 1.1 times of the conventional method. It is possible to further suppress the voice delay with respect to.

Other Embodiment 2
As described above, in the detailed description of the second embodiment of the present invention, the conversion speed control unit B82 that switches the speech speed conversion ratio according to the degree of increase in the voice delay is used as the method for determining the speech speed conversion ratio of the audio signal. Only when the voice delay increase degree satisfies a certain condition, the conversion speed control unit C83 that does not change the speech speed conversion magnification from the previous magnification may be used.
Next, the operation of the conversion speed control unit C83 that determines the speech speed conversion magnification of the voice signal in the speed conversion unit A51 will be described with reference to the flowchart of FIG.

FIG. 8 shows a processing flowchart of the conversion speed control unit C83, in which 83a, 83b, 83f, 83h, 83i, 83j, 83k, 83l, and 83m are processing blocks, and 83c, 83d, 83e, and 83g are determination blocks. It is.
Next, the operation will be described.

In the processing block 83a, a difference value between the current voice accumulation amount received from the output voice control unit 6 and the voice accumulation amount recorded in the previous process is calculated.
In the processing block 83b, the difference value is divided by the unit time to calculate the audio delay increase degree.

  In the decision block 83c, when the audio delay increase degree is 0.1 or more, the process proceeds to the processing block 83f. On the other hand, if the voice delay increase degree is less than 0.1 in the decision block 83c, the process proceeds to the decision block 83d.

  In the decision block 83d, when the degree of increase in voice delay is 0.05 or more, in the processing block 83j, an instruction signal for setting the speech speed conversion magnification of the time compression / decompression process to 0.9 times is sent to the speed converter A51. Then, the process proceeds to the processing block 83m. On the other hand, if it is determined at decision block 83d that the increase in audio delay is less than 0.05, the flow proceeds to decision block 83e.

  In the decision block 83e, if the degree of increase in the voice delay is −0.1 or more, in the processing block 83i, an instruction signal that does not change the speech speed conversion magnification of the time compression / expansion processing from the previous magnification is sent to the speed converter A51. And proceed to processing block 83m. On the other hand, when the voice delay increase degree is less than −0.1 in the decision block 83e, in the processing block 83h, an instruction signal for setting the speech speed conversion magnification of the time compression / decompression process to 0.8 is sent to the speed conversion unit. The data is sent to A51, and the process proceeds to process block 83m.

In the processing block 83f, a difference value between the level of the input voice signal determined to be voice received from the voice / silence discrimination unit A21 and a predetermined silence threshold level is calculated.
The silence threshold level here is a value of a level at which the audio signal is determined to be silent. For example, when the maximum value of the signal level is set to 0 dB in decibel notation, the level value to be determined as silence is −45 dB. When the signal level is used for the sound / silence determination unit in the sound / silence determination unit A21, the value for determining the level of silence is changed to be received from the sound / silence determination unit A21, The received value may be substituted for the silence threshold level.

  In the decision block 83g, if the level difference value calculated in the processing block 83f is 5 dB or less, the processing block 83l sends an instruction signal for setting the speech speed conversion magnification of the time compression / decompression process to 1.1 times as a speed conversion unit. The data is sent to A51, and the process proceeds to process block 83m. On the other hand, if the level difference value is larger than 5 dB in the decision block 83g, the processing block 83k sends an instruction signal for setting the speech speed conversion magnification of the time compression / decompression process to 1.0, to the speed converter A51. Proceed to process block 83m.

  Finally, in the processing block 83m, the current voice accumulation amount received from the output voice control unit 6 is recorded, and the process is terminated.

As a feature of the conversion speed control unit C83, in the conversion speed control unit B82 of the second embodiment shown in FIG. 6, the speech speed conversion magnification is set to 0.8 when the voice delay increase is less than 0.05. If the threshold is lowered to less than -0.10, and the voice delay increase takes a value between -0.10 and less than 0.05, it is set as a dead zone, and the speech rate conversion magnification is not changed. This is the point. The reason for this is that, for example, in the case where a plurality of speakers are in discussion, the voice accumulation amount may decrease for a moment in the conversation and start to increase again. If the conversion magnification is returned to 0.8 times, which is the slowest speed, the audio delay with respect to the video is prevented from increasing again.
The above is the description of the operation of the conversion speed control unit C83.

  As described above in detail, according to the second embodiment in which the conversion speed control unit C83 is used instead of the conversion speed control unit B82, the speech speed conversion magnification of the time compression / decompression process is switched as needed every unit time. In addition to the effect of suppressing the audio delay with respect to the video, it is possible to maintain the current speed against a minute increase / decrease in the amount of accumulated audio, further suppressing the audio delay with respect to the video. There is an effect that can be done.

  In the above embodiments, the relationship between video and audio of television broadcasting has been described as an example. However, the present invention is also applied to a telephone or the like so that a response is delayed in a conversation with the other party. You can prevent things in between.

  The present invention is applied to a television device, a telephone set, a mobile phone, and the like, so that the voice of these devices can be reproduced slowly without changing the voice pitch, and a spoken voice that is easy to hear for the listener can be obtained. It can be provided.

It is a functional block diagram which shows the structure of the speech speed conversion apparatus which concerns on Embodiment 1 of this invention. 5 is a process flowchart of a signal discard determination unit in the first embodiment. 3 is a functional block diagram illustrating a configuration of an output audio control unit according to Embodiment 1. FIG. 3 is a flowchart of the operation of a conversion speed control unit in the first embodiment. 7 is a process flowchart of a conversion speed control unit A of another example according to the first embodiment. It is a functional block diagram which shows the structure of the speech-speed converter based on Embodiment 2 of this invention. It is a process flowchart of the conversion speed control part B which concerns on this Embodiment 2 for this invention. 10 is a process flowchart of a conversion speed control unit C of another example according to the second embodiment.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1; Audio | voice input terminal, 2; Sound / silence discrimination | determination part, 3; Signal discard judgment part, 4; Silent signal discard part, 5; Speed conversion part, 6; Output voice control part, 6a; 6b; Conversion voice accumulating section, 6c; Delay time calculation section, 7; Voice output terminal, 8; Conversion speed control section, 21; Sound / silence discrimination section A, 51; Speed conversion section A, 81; A, 82; Conversion speed control unit B, 83; Conversion speed control unit C.

Claims (4)

A voice input terminal, a voice / silence determination unit, a signal discard determination unit, a silence signal discard unit, a speed conversion unit, an output voice control unit, a voice output terminal, and a conversion speed control unit;
The voice / silence discrimination unit discriminates whether the voice signal input from the voice input terminal is voiced or silent,
The signal discard judgment unit does not discard the voice signal by the discrimination signal from the voice / silence discrimination unit, and the silence signal calculates the total time of the input voice signal from the voice input terminal and outputs it from the output voice control unit Calculate the total time of the audio signal, subtract the total audio input time from the total audio output time, and determine whether to discard the silence signal based on the result,
The silent signal discarding unit also inputs the audio signal, discards the silent signal with the discard signal from the signal discard determination unit, and passes the other signals to the speed conversion unit,
The speed conversion unit performs a voice signal expansion process on the voice signal passed from the silence signal discarding unit without changing the voice pitch at the conversion rate of the voice speed from the conversion speed control unit,
The output audio control unit accumulates the audio signal subjected to the time compression / decompression processing from the speed conversion unit in the audio accumulation memory, and discards the signal as the time length of the output audio signal every time the audio signal is input. Output to the audio output terminal the audio signals waiting to be output that have been passed to the discriminator and stored in the audio storage memory, in order from the oldest to the audio output terminal, and count the number of output audio samples. To the conversion speed control unit
The conversion speed control unit calculates a difference value between the current voice accumulation amount received from the output voice control unit and the voice accumulation amount at the previous processing, and divides this difference value by unit time to obtain a voice delay increase degree. And a speech speed conversion device configured to output a speech speed conversion magnification corresponding to the degree of increase in voice delay to the speed converter.   The conversion speed control unit does not perform the time compression / expansion processing based on the voice delay increase degree and the amplitude level of the input voice signal is smaller than a predetermined value. The speech speed converting apparatus according to claim 1, wherein the apparatus is configured to output to the speed converting unit.   3. The conversion speed control unit according to claim 1, wherein the conversion speed control unit is configured to output a speech speed conversion magnification of the previous process to the speed conversion unit when the degree of increase in voice delay is within a specific range. Speaking speed converter. A voice / silence discrimination step for discriminating whether the input voice signal from the voice input terminal is voiced or silent, and outputting a judgment signal;
Whether or not the sound signal is discarded by the determination signal from the sound / silence determination step, and the sound signal subtracts the total time of sound input from the total sound output time at the present time, and whether to discard the sound signal according to the result A signal discarding determination step for determining whether or not
The audio signal is input, the silence signal is discarded by the discard signal from the signal discard determination step, and other signals are sent to the silence signal discard step and the silence signal discard step that are passed to the speed conversion step of the next processing. On the other hand, a speed conversion step for performing speech signal expansion processing at a speech speed conversion magnification determined in the previous processing without changing the voice pitch,
The audio signal subjected to the time compression / decompression processing from the speed conversion step is accumulated in the audio accumulation memory, and every time the audio signal is input, the time length is set as the time length of the output audio signal, and the audio used in the signal discard determination step Output as the original data for the total number of output times, and output the standby audio signals stored in the audio storage memory to the audio output terminal in order from the oldest one, and count the number of output audio samples And an output audio control step for outputting the current audio accumulation amount at regular intervals,
Calculate the difference between the current audio accumulation amount output from the output audio control step and the audio accumulation amount at the previous processing, and divide this difference value by unit time to calculate the audio delay increase degree. A speech speed conversion control method, comprising: a conversion speed control step for outputting a speech speed conversion magnification according to a delay increase degree as a speech speed conversion magnification used in the speed conversion step.

Images