JP2006017900A - Time stretch processing device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To obtain a time stretch processing apparatus capable of implementing an appropriate speed change conversion process regardless of the kinds of a sound source. <P>SOLUTION: A comparison section 2 for deciding whether the pitch period T extracted by a pitch period extraction section 1 is within a permissible range or not is provided and when the result of comparison indicating that the pitch period is within the permissible range is obtained from the comparison section 2, the processing frame length FL corresponding to the pitch period T is specified. When the result of the comparison indicating that the pitch period T is deviated from the permissible range is obtained from the comparison section 2, the preset processing frame length FL is set. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a time stretch processing device that performs speed conversion processing of an audio signal in accordance with, for example, the pitch period of the audio signal.

A conventional time stretch processing device extracts a pitch period of an audio signal in order to change the sound length without changing the pitch or sound quality of an input signal (for example, an audio signal), and repeats the audio signal at the pitch period. Interpolation processing and thinning processing for thinning out audio signals are performed.
Alternatively, the audio signals are cut out for each pitch period, and each cut-out signal is windowed and then subjected to processing such as superimposing each other.

However, when the pitch period of the audio signal cannot be extracted, the length of the sound cannot be changed. Therefore, there is a method of classifying the audio signal for each section and switching the processing according to the feature.
For example, Patent Document 1 below classifies a voice signal into a voiced sound section and an unvoiced sound section where there is no pitch period. In a voiced sound section, the voice signal is compressed using the pitch period. Perform decompression to convert speed. On the other hand, in the unvoiced sound section, a method of converting the speed according to the sampling period is disclosed.

  Further, in Patent Document 2 below, there is a method for classifying a voice signal into an unvoiced section in which no pitch period exists and other sections, and in the unvoiced section, a thinning process or an interpolation process is performed with an arbitrary fixed frame length. It is disclosed.

JP-A-63-204300 (pages 9 to 13, FIG. 1) JP-A-10-91189 (paragraph numbers [0010] to [0011], FIG. 1)

  Since the conventional time stretch processing apparatus is configured as described above, the speed can be converted even in a silent section where the pitch period of the audio signal cannot be extracted. However, when the BGM overlaps the audio signal or when the input signal is an acoustic signal, the pitch period cannot be extracted even in a sound section, and an appropriate speed conversion process is performed. There were problems such as being unable to do so.

  The present invention has been made to solve the above-described problems, and an object of the present invention is to obtain a time stretch processing device capable of performing an appropriate speed conversion process regardless of the type of sound source.

  The time stretch processing device according to the present invention is provided with a determination unit that determines whether or not the pitch period extracted by the pitch period extraction unit is within an allowable range, and that the pitch period is within the allowable range from the determination unit. If a determination result indicating that the pitch period is out of the allowable range is obtained from the determination means, a determination result indicating that the pitch period is out of the permissible range is set in advance. The specified processing frame length is designated.

  According to the present invention, the determination means for determining whether or not the pitch period extracted by the pitch period extraction means is within the allowable range is provided, and the determination result indicating that the pitch period is within the allowable range from the determination means If the determination result indicating that the pitch period is out of the allowable range is obtained from the determination means, the processing frame set in advance is specified. Since the length is designated, there is an effect that an appropriate speed conversion process can be performed regardless of the type of the sound source.

Embodiment 1 FIG.
1 is a block diagram showing a time stretch processing apparatus according to Embodiment 1 of the present invention. In the figure, a pitch period extraction unit 1 extracts a pitch period of an input signal (for example, an audio signal, an acoustic signal). It constitutes an extraction means.
The comparison unit 2 compares the pitch period extracted by the pitch period extraction unit 1 with the shortest value of the allowable range, compares the pitch period with the longest value of the allowable range, and outputs the comparison result to the control unit 3. The comparison unit 2 constitutes a determination unit.

When a comparison result indicating that the pitch period is within the allowable range is obtained from the comparison unit 2, the control unit 3 designates a processing frame length corresponding to the pitch period, and the comparison unit 2 determines that the pitch period is within the allowable range. When a comparison result indicating that there is a deviation is obtained, a preset processing frame length is designated. The control unit 3 constitutes a frame length designation unit.
The speed conversion processing unit 4 constitutes a speed conversion means for performing the speed conversion process of the input signal with the processing frame length designated by the control unit 3.
FIG. 2 is a flowchart showing the processing contents of the time stretch processing apparatus according to Embodiment 1 of the present invention.

Next, the operation will be described.
First, when receiving an input signal such as an audio signal, for example, the pitch period extracting unit 1 extracts the pitch period T of the input signal as shown in FIG. 12 (step ST1).
As a method of extracting the pitch period T, for example, a method of obtaining an average pitch period by performing autocorrelation processing on the original speech waveform of the input signal, a method of performing LPC analysis of the input signal, and using a residual, FFT There are various methods, such as a method using, and a method for obtaining the maximum value from the correlation at the translation point.

However, the pitch period is originally a sensory quantity, and there is no complete periodic wave other than a mechanically generated waveform, and it is only a quasi-periodic one. Further, since there are noise effects and individual differences, it is extremely difficult to extract an accurate pitch period.
Furthermore, when the input signal is not limited to voice or the like, it is difficult to extract an accurate pitch period because various signals can be superimposed.
Therefore, the pitch period T extracted by the pitch period extraction unit 1 is not necessarily an accurate pitch period, but such values are extracted.

When the pitch period extraction unit 1 extracts the pitch period T, the comparison unit 2 determines whether the pitch period T is within an allowable range (for example, 1/3000 second to 1/50 second).
That is, since the pitch period T extracted by the pitch period extraction unit 1 is not necessarily an accurate pitch period but is a correct value, the processing frame length always used for the speed conversion process according to the pitch period T. If FL is specified, it may not be possible to perform an appropriate speed conversion process, so it is determined whether or not the pitch period T is within an allowable range.

Specifically, the comparison unit 2 compares the pitch period T extracted by the pitch period extraction unit 1 with the shortest allowable value S _ref (for example, 1/3000 seconds), and the pitch period T is the shortest. If it is shorter than the value S _ref (step ST2), a shortest value deviation signal indicating that it is shorter than the shortest value is output to the control unit 3.
If the pitch period T is not shorter than the shortest value S _ref (step ST2), the comparison unit 2 compares the pitch period T with the longest allowable value L _ref (for example, _1/50 second). , the longer the pitch period T is than the longest value L _ref (step ST3), and outputs the maximum value deviation signal indicating that longer than the maximum value to the control unit 3.
If the pitch period T is not longer than the longest value L _ref (step ST3), an allowable signal indicating that it is within the allowable range is output to the control unit 3.

When receiving the permission signal from the comparison unit 2, the control unit 3 calculates a processing frame length FL corresponding to the pitch period T extracted by the pitch period extraction unit 1 as shown below, for example (step ST4). Here, α is a constant such as 1, 2, 3,.
FL = α × T

When the control unit 3 receives the shortest deviation signal from the comparison unit 2 and specifies the processing frame length FL corresponding to the pitch period T extracted by the pitch period extraction unit 1, the processing frame length FL becomes too short. Since the pitch is disturbed and distorted, a preset processing frame length FL is designated (step ST5).
For example, if the pitch period T is 1/100 second, the processing frame length FL is as follows.
FL = α / 100

When the control unit 3 receives the longest value deviation signal from the comparison unit 2 and specifies the processing frame length FL corresponding to the pitch period T extracted by the pitch period extraction unit 1, the processing frame length FL becomes too long. As a result, clear omission and echo of the signal occur, and a preset processing frame length FL is designated (step ST6).
For example, if the pitch period T is 1/100 second, the processing frame length FL is as follows.
FL = α / 100

When the control unit 3 designates the processing frame length FL as described above, the speed conversion processing unit 4 performs speed conversion processing of the input signal with the processing frame length FL (step ST7).
For the speed conversion processing of the input signal, the method disclosed in the above-mentioned Patent Documents 1 and 2 may be applied. For example, thinning processing for thinning the processing frame length FL from the pitch width corresponding to the pitch period T is performed. carry out.

As apparent from the above, according to the first embodiment, the comparison unit 2 for determining whether or not the pitch period T extracted by the pitch period extraction unit 1 is within the allowable range is provided. When a comparison result indicating that the pitch period is within the allowable range is obtained, the processing frame length FL corresponding to the pitch period T is specified, and the pitch period T deviates from the allowable range from the comparison unit 2. If a comparison result indicating is obtained, it is configured to specify a preset processing frame length FL, so that any irregularity or distortion of the pitch, obvious missing of the signal or occurrence of echo is generated regardless of the type of sound source. There is an effect that an appropriate speed conversion process can be performed without inviting.
Even if the pitch period extraction unit 1 does not extract an accurate pitch period T, the control unit 3 can designate an appropriate processing frame length FL by extracting such a correct value. There is an effect that the amount of calculation accompanying extraction can be reduced.

In the first embodiment, when the control unit 3 receives the shortest value deviation signal or the longest value deviation signal from the comparison unit 2, the control frame 3 designates a preset processing frame length FL. The processing frame length FL corresponding to the property and the quality of the output signal may be designated.
Further, when the control unit 3 receives the shortest value deviation signal or the longest value deviation signal from the comparison unit 2, the previously specified processing frame length FL may be designated again.

Embodiment 2. FIG.
3 is a block diagram showing a time stretch processing apparatus according to Embodiment 2 of the present invention. In the figure, the same reference numerals as those in FIG.
The pitch cycle storage unit 11 is a memory that stores the pitch cycle extracted by the pitch cycle extraction unit 1, and the pitch cycle storage unit 11 constitutes pitch cycle storage means.
The comparison unit 12 obtains a difference between the current pitch period extracted by the pitch period extraction unit 1 and the previous pitch period stored in the pitch period storage unit 11, and the difference between the pitch periods and the shortest allowable range Are compared with the difference between the pitch periods and the longest value of the allowable range, and the comparison result is output to the control unit 13. The comparison unit 12 constitutes a determination unit.

When the control unit 13 obtains a comparison result indicating that the pitch period difference is within the allowable range from the comparison unit 12, the control unit 13 designates a processing frame length corresponding to the current pitch period, and the comparison unit 12 specifies the pitch period. When a comparison result indicating that the difference deviates from the allowable range is obtained, a preset processing frame length is designated. The control unit 13 constitutes a frame length designation unit.
FIG. 4 is a flowchart showing the processing contents of the time stretch processing apparatus according to the second embodiment of the present invention.

Next, the operation will be described.
First, when receiving an input signal such as an audio signal, for example, the pitch period extraction unit 1 extracts the pitch period _TN of the input signal as in the first embodiment (step ST11).
Note that the pitch period extraction unit 1 outputs the extracted pitch period _TN to the comparison unit 12 and stores the pitch period _TN in the pitch period storage unit 11.
However, the subscript N of T indicates the pitch period of the Nth frame.

When the pitch cycle extraction unit 1 extracts the current pitch cycle T _N (the pitch cycle of the Nth frame), the comparison unit 12 extracts the previous pitch cycle T _N-1 (the N−1th frame) from the pitch cycle storage unit 11. The pitch period) is read out, and a difference ΔT between the pitch period _TN of the Nth frame and the pitch period _TN-1 of the (N-1) th frame is obtained (step ST12).

Next, the comparison unit 12 compares the pitch period difference ΔT with the shortest allowable value S _ref (for example, a value of −10% of the Nth frame pitch period T _N ), and the pitch period difference. If ΔT is shorter than the shortest value S _ref (step ST13), a shortest value deviation signal indicating that the shortest value is deviated is output to the control unit 13.
If the difference ΔT in the pitch period is not shorter than the shortest value S _ref (step ST13), the comparison unit 12 and the difference ΔT in the pitch period and the longest allowable value L _ref (for example, the pitch of the Nth frame) comparing +10% of the value) and the period T _N, the longer the difference ΔT of the pitch period is than the longest value L _ref (step ST14), the longest value deviation signal indicating that deviates from the maximum value control unit 13 is output.
If the pitch period difference ΔT is not longer than the longest value L _ref (step ST14), an allowable signal indicating that it is within the allowable range is output to the control unit 13.

When receiving the permission signal from the comparison unit 12, the control unit 13 calculates the processing frame length FL corresponding to the pitch period _TN extracted by the pitch period extraction unit 1 as in the first embodiment (step ST15). ).

When the control unit 13 receives the shortest deviation signal from the comparison unit 12 and designates the processing frame length FL corresponding to the pitch period T _N of the Nth frame, the processing frame length FL becomes the (N−1) th frame. Since this is a value far from the processing frame length FL, which leads to instability of the speed conversion processing, a preset processing frame length FL is designated (step ST16).
For example, if the pitch period T is 1/100 second, the processing frame length FL is as follows.
FL = α / 100

When the control unit 13 receives the longest value deviation signal from the comparison unit 12 and designates the processing frame length FL corresponding to the pitch period _TN of the Nth frame, the processing frame length FL is changed to the (N−1) th frame. Since it becomes a value far from the processing frame length FL, which leads to instability of the speed conversion processing, a preset processing frame length FL is designated (step ST17).
For example, if the pitch period T is 1/100 second, the processing frame length FL is as follows.
FL = α / 100

  When the control unit 13 designates the processing frame length FL as described above, the speed conversion processing unit 4 performs the speed conversion processing of the input signal with the processing frame length FL as in the first embodiment (step ST18).

  As apparent from the above, according to the second embodiment, the difference ΔT between the current pitch period extracted by the pitch period extracting unit 1 and the previous pitch period stored in the pitch period storing unit 11 is allowed. When the comparison unit 12 for determining whether or not it is within the range is provided, and the comparison result indicating that the pitch period difference ΔT is within the allowable range is obtained from the comparison unit 12, processing according to the current pitch cycle The frame length FL is designated, and when a comparison result indicating that the pitch period difference ΔT is out of the allowable range is obtained from the comparison unit 12, a preset processing frame length FL is designated. Therefore, as in the first embodiment, an appropriate speed conversion process can be performed regardless of the type of the sound source, and the pitch period is temporarily accompanied with the level fluctuation of the signal superimposed on the input signal. Change to Even, to prevent large variations in processing frame length FL, offers an advantage of being able to stabilize the speed conversion process.

In the second embodiment, as mentioned above. Comparing section 12 obtains a difference ΔT between the pitch period T _N-1 of the pitch period T _N and N-1 th frame N th frame, the previous pitch period T _N-1 need not be a pitch period _{T N-1} to N-1 th frame, for example, a difference ΔT between the pitch period _{T N-2} of the pitch period _{T N} and N-2-th frame of the N th frame May be requested.
Further, in the second embodiment, when the control unit 13 receives the shortest value deviation signal or the longest value deviation signal from the comparison unit 12, it is shown that the preset processing frame length FL is designated. The processing frame length FL may be designated again.

Embodiment 3 FIG.
In the second embodiment, the shortest allowable value S _ref and the longest value L _ref are set regardless of the band of the input signal. However, the shortest value S of the allowable range is set for each band of the input signal. _ref and longest value L _ref may be set, and the comparison unit 12 may compare the shortest value S _ref and longest value L _{ref of} the allowable range corresponding to the band of the input signal with the pitch period difference ΔT.

For example, when the shortest value S _ref and the longest value L _ref of the allowable range are set as follows for each band of the input signal, for example, if the band of the input signal is 200 Hz, the comparing unit 12 is N using the 10% value of the frame of the pitch period _{T N} as the shortest value _{S ref,} so as to use a + 10% of the value of the pitch period _{T N} of the N th frame as the longest value _{L ref.}
Input signal bandwidth Minimum value Maximum value ~ 100Hz → -5% + 5%
100-300Hz → -10% + 10%
300Hz-> -20% + 20%

As apparent from the above, according to the third embodiment, since the shortest value S _ref and longest value L _{ref of} the allowable range corresponding to the band of the input signal are compared with the pitch period difference ΔT, There is an effect that the influence of erroneous extraction of the pitch period can be reduced.

Embodiment 4 FIG.
In the first to third embodiments, when the control units 3 and 13 receive the comparison results from the comparison units 2 and 12, the processing frame length FL is immediately specified in consideration of the comparison results. If the processing frame length FL is changed following the temporary change of the pitch period accompanying the level fluctuation of the signal superimposed on the signal, the speed conversion process may become unstable.

Therefore, in the fourth embodiment, even when the control units 3 and 13 receive the comparison results from the comparison units 2 and 12, the processing frame length FL is not immediately changed, and the processing frame is not changed when the same designated condition continues for a predetermined number of times. Change the length FL.
For example, when the current pitch cycle T received by the comparison unit 2 and the current pitch cycle T extracted by the pitch cycle extraction unit 1 has the same value M times (for example, M = 3) continuously, the control unit 3 Designates the processing frame length FL according to the current pitch period T.
Alternatively, the control unit 13 responds to the current pitch cycle T when receiving the allowable signal from the comparison unit 12 continuously M times (for example, M = 3) and the pitch cycle difference ΔT is the same value. The processing frame length is changed to FL.

Alternatively, when the shortest value deviation signal is received from the comparison units 2 and 12 consecutively M times (for example, M = 3), the processing frame length FL when the control units 3 and 13 deviate from the shortest value is changed. Like that.
Alternatively, when the longest value deviation signal is received from the comparison units 2 and 12 continuously M times (for example, M = 3), the control units 3 and 13 are changed to the processing frame length FL when the longest value is deviated. Like that.

  As is apparent from the above, according to the fourth embodiment, the processing frame length FL is not changed immediately even if the processing frame length designation condition changes, and the processing frame length is changed when the same designation condition continues a predetermined number of times. Since it is configured to change the FL, even if the pitch period changes temporarily due to the level fluctuation of the signal superimposed on the input signal, a large change in the processing frame length FL is prevented and the speed conversion process is performed. The effect which can aim at stabilization of is produced.

Embodiment 5. FIG.
In the first to third embodiments, when the control units 3 and 13 receive the comparison result from the comparison units 2 and 12, the processing frame length FL is immediately specified in consideration of the comparison result. When changing the designated processing frame length FL, the processing frame length FL may be changed stepwise.

  Specifically, for example, when the processing frame length FL specified last time is “1/25 seconds” and the processing frame length FL that meets the current specified conditions is “1/100 seconds”, 3 immediately specifies “1/100 second” as the processing frame length FL, but in the fifth embodiment, the processing frame length FL is not immediately changed to “1/100 second”, but “1/25”. The processing frame length FL is changed stepwise in the order of “second” → “1/50 second” → “1/100 second”.

  According to the fifth embodiment, even if the pitch period temporarily changes with the level fluctuation of the signal superimposed on the input signal, a large fluctuation in the processing frame length FL can be prevented. There is an effect that the conversion processing can be stabilized.

Embodiment 6 FIG.
5 is a block diagram showing a time stretch processing apparatus according to Embodiment 6 of the present invention. In the figure, the same reference numerals as those in FIG.
The voice / silence detector 21 detects whether the input signal is voiced or silent. The switch 22 provides an input signal to the pitch cycle extraction unit 1 when the voice / silence detection unit 21 detects the voice, while an input signal to the pitch cycle extraction unit 1 when the voice / silence detection unit 21 detects silence. Will no longer be offered.

As in the case of the control unit 3 in FIG. 1, when the control unit 23 obtains a comparison result indicating that the pitch period T is within the allowable range from the comparison unit 2, the control unit 23 sets the processing frame length FL corresponding to the pitch period T. When the comparison result indicating that the pitch period T deviates from the allowable range is obtained from the comparison unit 2, the preset processing frame length FL is designated, but the sound / silence detection unit 21 When silence is detected, a preset processing frame length FL is designated.
The voice / silence detection unit 21, the switch 22, and the control unit 23 constitute a frame length designation unit.
FIG. 6 is a flowchart showing the processing contents of the time stretch processing apparatus according to the sixth embodiment of the present invention.

Next, the operation will be described.
First, the voice / silence detector 21 detects whether the input signal is voiced or silent (step ST21).
The switch 22 gives an input signal to the pitch period extraction unit 1 (step ST23) when the sound / silence detection unit 21 detects the sound (step ST22).
Since the subsequent processing contents in this case are the same as those in the first embodiment, description thereof will be omitted.

When the sound / silence detection unit 21 detects silence (step ST22), the switch 22 stops providing the input signal to the pitch period extraction unit 1 (step ST24).
That is, when the input signal is silent, there is clearly no periodic signal in the input signal, so the input signal is provided to the pitch period extraction unit 1 in order to stop the useless pitch period extraction process. Cancel. Thereby, it is possible to reduce the amount of calculation in the pitch period extraction unit 1 and to prevent erroneous extraction of the pitch period.

When the voice / silence detection unit 21 detects voice, the control unit 23 specifies the processing frame length FL in the same manner as the control unit 3 in FIG. Is detected, a preset processing frame length FL is designated (step ST25).
For example, if the pitch period T is 1/100 second, the processing frame length FL is as follows.
FL = α / 100

  As apparent from the above, according to the sixth embodiment, when the input signal is silent, the processing frame length FL set in advance is designated, so that the amount of calculation in the pitch period extraction unit 1 is reduced. In addition to being able to achieve reduction, it is possible to prevent an erroneous extraction of the pitch period and perform an appropriate speed conversion process.

In the sixth embodiment, the voice / silence detection unit 21 and the switch 22 are applied to the time stretch processing apparatus of FIG. 1, but the voice / silence detection unit 21 and the switch 22 are shown in FIG. You may apply to a time stretch processing apparatus.
In this case, when the sound / silence detection unit 21 detects sound, the control unit 23 specifies the processing frame length FL in the same manner as the control unit 13 of FIG.

Embodiment 7 FIG.
7 is a block diagram showing a time stretch processing apparatus according to Embodiment 7 of the present invention. In the figure, the same reference numerals as those in FIG.
The extraction cycle limiter 31 constitutes an extraction range setting means for setting the extraction range of the pitch cycle of the input signal.
If the pitch period of the input signal is within the extraction range set by the extraction period limiting unit 31, the pitch period extraction unit 32 extracts the pitch period T. If not, the pitch period extraction unit 32 extracts the pitch period within the extraction range. The most likely pitch period T is extracted. Note that the pitch period extracting unit 32 constitutes a pitch period extracting unit.

The control unit 33 constitutes a frame length designating unit that designates a processing frame length FL corresponding to the pitch period T extracted by the pitch period extracting unit 32.
FIG. 8 is a flowchart showing the processing contents of the time stretch processing apparatus according to the seventh embodiment of the present invention.

Next, the operation will be described.
First, the extraction cycle limiter 31 sets the extraction range of the pitch cycle T of the input signal (step ST31).
For example, when the pitch period T of the input signal is in the range of 1/3000 sec to 1/50 sec, if the speed conversion process based on the pitch period T can be optimized, the pitch period T of the input signal is extracted. As a range, 1/3000 second to 1/50 second is set.

  When receiving an input signal such as an audio signal, the pitch period extraction unit 32 extracts the pitch period T of the input signal as in the first embodiment (step ST32). However, if the pitch period T of the input signal is within the extraction range set by the extraction period limiting unit 31 (step ST33), the pitch period T is output to the control unit 33, but is not within the extraction range (step ST33). Step ST33), the most probable pitch period T within the extraction range is extracted (step ST34), and the pitch period T is output to the control unit 33. For example, when the input signal is subjected to FFT conversion and converted to a frequency domain signal, even if the maximum signal is outside the extraction range, the signal also appears inside the extraction range. The largest signal is extracted as the most likely pitch period T.

When receiving the pitch period T from the pitch period extraction unit 32, the control unit 33 designates the processing frame length FL corresponding to the pitch period T in the same manner as in the first embodiment (step ST35).
When the control unit 3 designates the processing frame length FL, the speed conversion processing unit 4 performs the speed conversion process of the input signal with the processing frame length FL as in the first embodiment (step ST36).

  As apparent from the above, according to the seventh embodiment, if the pitch period T of the input signal is within the extraction range set by the extraction period limiting unit 31, the pitch period T is extracted while the extraction is performed. If it is not within the range, the most probable pitch period T within the extraction range is extracted, so that it is possible to improve the reliability of the speed conversion process based on the pitch period T.

Embodiment 8 FIG.
9 is a block diagram showing a time stretch processing apparatus according to Embodiment 8 of the present invention. In the figure, the same reference numerals as those in FIG.
The comparison unit 41 compares the pitch cycle T extracted by the pitch cycle extraction unit 1 with the shortest value S _ref , and outputs a Long signal indicating that to the control unit 42 if the pitch cycle T is longer than the shortest value S _ref. If the pitch period T is shorter than the shortest value S _ref , a Short signal indicating that fact is output to the control unit 42. The comparison unit 41 constitutes a determination unit.

When receiving the Long signal from the comparison unit 41, the control unit 42 designates the processing frame length FL corresponding to the pitch period T extracted by the pitch period extraction unit 1, and when receiving the Shrot signal from the comparison unit 41, the control unit 42 extracts the pitch period. A processing frame length FL corresponding to an integral multiple of the pitch period T extracted by the unit 1 is designated. The control unit 42 constitutes a frame length designation unit.
FIG. 10 is a flowchart showing the processing contents of the time stretch processing apparatus according to the eighth embodiment of the present invention.

Next, the operation will be described.
When receiving an input signal such as an audio signal, the pitch period extraction unit 1 extracts the pitch period T of the input signal as in the first embodiment (step ST41).

The comparison unit 41 compares the pitch period T extracted by the pitch period extraction unit 1 with the shortest value S _ref (for example, 1/1000 second), and if the pitch period T is longer than the shortest value S _ref (step ST42). Then, a Long signal indicating that is output to the control unit 42.
On the other hand, if the pitch period T is shorter than the shortest value S _ref (step ST42), a Short signal indicating that is output to the control unit 42.

When receiving the long level signal from comparison unit 41, control unit 42 designates a processing frame length FL corresponding to pitch period T extracted by pitch period extraction unit 1 as in the first embodiment (step ST43). ).
However, when the Short signal is received from the comparison unit 41, if the processing frame length FL corresponding to the pitch period T extracted by the pitch period extraction unit 1 is specified, the processing frame length FL becomes too short, and the pitch is disturbed. Since distortion occurs, a processing frame length FL corresponding to an integral multiple of the pitch period T extracted by the pitch period extraction unit 1 is designated (step ST44).
For example, if the pitch period T extracted by the pitch period extraction unit 1 is 1/1000 second, the processing frame length FL corresponding to 1/500 second or 1/250 second is designated.

  When the control unit 42 designates the processing frame length FL, the speed conversion processing unit 4 performs the speed conversion process of the input signal with the processing frame length FL as in the first embodiment (step ST45).

As apparent from the above, according to the eighth embodiment, if the pitch period T extracted by the pitch period extracting unit 1 is longer than the shortest value S _ref, the pitch period T extracted by the pitch period extracting unit 1 is A corresponding processing frame length FL is specified, and if the pitch period T is shorter than the shortest value S _ref , a processing frame length FL corresponding to an integer multiple of the pitch period T extracted by the pitch period extraction unit 1 is specified. Since it comprised, regardless of the kind of sound source, there exists an effect which can implement an appropriate speed conversion process, without causing disorder of a pitch or generation | occurrence | production of distortion.

Embodiment 9 FIG.
In the eighth embodiment, if the pitch period T extracted by the pitch period extraction unit 1 is shorter than the shortest value S _ref , the control unit 42 responds to an integral multiple of the pitch period T extracted by the pitch period extraction unit 1. Although the processing frame length FL is specified, if the pitch period T extracted by the pitch period extraction unit 1 is shorter than the shortest value S _ref , the control unit 42 extracts the pitch period T extracted by the pitch period extraction unit 1. It is also possible to specify a processing frame length corresponding to and set a plurality of processing times.

That is, the control unit 42 designates the processing frame length FL corresponding to the pitch period T extracted by the pitch period extraction unit 1 as in the first embodiment, but when receiving the Short signal from the comparison unit 41, Set multiple processing times. For example, the number of processings is set within a range in which the time stretch processing device can continue the real-time processing. If there is a margin for the current processing capacity, the number of processing times is set to a small value.
When the control unit 42 designates the processing frame length FL, the speed conversion processing unit 4 performs the speed conversion process of the input signal with the processing frame length FL as in the first embodiment. However, when the control unit 42 sets a plurality of processing times, the input signal speed conversion processing is performed a plurality of times with the processing frame length FL designated by the control unit 42 for the number of processing times.

  According to the ninth embodiment, it is possible to prevent the speed conversion processing from becoming unstable due to the processing frame length FL immediately changing to the processing frame length FL1 corresponding to the pitch period T. There is an effect that the problem that the frame length FL cannot follow the change of the pitch period T can be solved.

Embodiment 10 FIG.
FIG. 11 is a block diagram showing a time stretch processing apparatus according to Embodiment 10 of the present invention. In the figure, a pitch period extraction unit 51 starts from a change point detected by a pitch period change point detection unit 52. Pitch period extraction means for extracting the pitch period T of the signal is configured.
The pitch cycle change point detector 52 detects a change point at which the pitch cycle T extracted by the pitch cycle extractor 51 changes greatly. The pitch cycle change point detector 52 constitutes a change point detector.

The control unit 53 constitutes a frame length designating unit that designates a processing frame length corresponding to the pitch period T extracted by the pitch period extracting unit 51.
The speed conversion processing unit 54 performs speed conversion processing of the input signal with the processing frame length designated by the control unit 53 from the change point detected by the pitch period change point detection unit 52. The speed conversion processing unit 54 constitutes a speed conversion unit.

Next, the operation will be described.
The pitch period change point detector 52 monitors the pitch period T extracted by the pitch period extractor 51 and detects a change point at which the pitch period T changes greatly.
That is, a change point at which the pitch period T changes more than a preset value is detected. For example, a position where the input signal changes from a sound signal to an acoustic signal is detected.
The change point of the pitch period T can be detected from the amount of change in the power of the input signal, for example. It is also possible to detect by subdividing the input signal range and observing the FFT characteristics.

When the pitch cycle change point detection unit 52 detects the change point of the pitch cycle T, the pitch cycle extraction unit 51 extracts the pitch cycle T of the input signal from the change point under the instruction of the control unit 53. .
The control unit 53 causes the pitch cycle T extracted by the pitch cycle extraction unit 51 to process up to the change point, and causes the pitch cycle extraction unit 51 to resubmit the cycle after the change point. From the change point, it is processed in the resubmitted cycle. .
Note that when the pitch cycle change point detection unit 52 detects the change point of the pitch cycle T, the control unit 53 performs processing according to the pitch cycle T extracted by the pitch cycle extraction unit 51 as in the first embodiment. Specifies the frame length FL.

  Under the instruction of the control unit 53, the speed conversion processing unit 54 performs speed conversion processing of the input signal with the processing frame length FL designated by the control unit 53 from the change point detected by the pitch period change point detection unit 52. To do.

  As is apparent from the above, according to the tenth embodiment, the speed conversion process of the input signal is performed with the processing frame length designated by the control unit 53 from the change point detected by the pitch cycle change point detection unit 52. As in the first embodiment, since it is possible to perform an appropriate speed conversion process regardless of the type of sound source, and the follow-up performance with respect to fluctuations in the pitch period T is enhanced, high sound quality is achieved. This produces an effect of obtaining a stable output signal.

It is a block diagram which shows the time stretch processing apparatus by Embodiment 1 of this invention. It is a flowchart which shows the processing content of the time stretch processing apparatus by Embodiment 1 of this invention. It is a block diagram which shows the time stretch processing apparatus by Embodiment 2 of this invention. It is a flowchart which shows the processing content of the time stretch processing apparatus by Embodiment 2 of this invention. It is a block diagram which shows the time stretch processing apparatus by Embodiment 6 of this invention. It is a flowchart which shows the processing content of the time stretch processing apparatus by Embodiment 6 of this invention. It is a block diagram which shows the time stretch processing apparatus by Embodiment 7 of this invention. It is a flowchart which shows the processing content of the time stretch processing apparatus by Embodiment 7 of this invention. It is a block diagram which shows the time stretch processing apparatus by Embodiment 8 of this invention. It is a flowchart which shows the processing content of the time stretch processing apparatus by Embodiment 8 of this invention. It is a block diagram which shows the time stretch processing apparatus by Embodiment 10 of this invention. It is explanatory drawing which shows the pitch period of an input signal.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 Pitch period extraction part (pitch period extraction means), 2 Comparison part (determination means), 3 Control part (frame length designation means), 4 Speed conversion process part (speed conversion means), 11 Pitch period preservation | save part (Pitch period preservation | save) Means), 12 comparison unit (determination unit), 13 control unit (frame length designation unit), 21 voice / silence detection unit (frame length designation unit), 22 switch (frame length designation unit), 23 control unit (frame length) Designation unit), 31 extraction cycle limiting unit (extraction range setting unit), 32 pitch cycle extraction unit (pitch cycle extraction unit), 33 control unit (frame length designation unit), 41 comparison unit (determination unit), 42 control unit ( Frame length designation means), 51 pitch period extraction section (pitch period extraction means), 52 pitch period change point detection section (change point detection means), 53 control section (frame length designation means) , 54 speed conversion unit (speed converting means).

Claims (12)

  A pitch period extracting means for extracting a pitch period of the input signal; a determining means for determining whether or not the pitch period extracted by the pitch period extracting means is within an allowable range; If a determination result indicating that the pitch period is within the specified range, a processing frame length corresponding to the pitch period is specified, and a determination result indicating that the pitch period is outside the allowable range is obtained from the determination unit. In this case, a time stretch process comprising: a frame length designating unit that designates a preset processing frame length; and a speed conversion unit that performs a speed conversion process on the input signal with the processing frame length designated by the frame length designating unit. apparatus.   Pitch period extracting means for extracting the pitch period of the input signal, pitch period storing means for storing the pitch period extracted by the pitch period extracting means, the current pitch period extracted by the pitch period extracting means and the pitch Determination means for determining whether or not the difference from the previous pitch period stored in the period storage means is within the allowable range, and a determination result indicating that the difference in pitch period is within the allowable range from the determination means Is set in advance when a processing frame length corresponding to the current pitch period is designated and a determination result indicating that the difference in pitch period is out of the allowable range is obtained from the determination means. A frame length specifying means for specifying the processed frame length, and a speed change for executing the speed conversion processing of the input signal with the processing frame length specified by the frame length specifying means. Time stretching process and means.   3. The time stretch processing device according to claim 2, wherein the determination means sets a shortest value and a longest value in an allowable range in advance, and compares the shortest value and the longest value with a pitch period difference.   The determination means is characterized in that the shortest and longest values of the allowable range are set for each band of the input signal, and the shortest and longest values of the allowable range corresponding to the band of the input signal are compared with the difference between the pitch periods. The time stretch processing apparatus according to claim 2.   The time stretch processing device according to any one of claims 1 to 4, wherein the frame length specifying means has a processing frame length set in advance as a processing frame length specified last time.   2. The frame length specifying means does not immediately change the processing frame length even if the processing frame length specifying condition changes, and changes the processing frame length when the same specifying condition continues for a predetermined number of times. The time stretch processing apparatus according to claim 1.   The time stretch according to any one of claims 1 to 4, wherein the frame length designation means changes the process frame length in stages when the designated process frame length is changed. Processing equipment.   5. The time stretch processing device according to claim 1, wherein the frame length designation means designates a preset processing frame length when the input signal is silent.   Extraction range setting means for setting the extraction range of the pitch period of the input signal, and if the pitch period of the input signal is within the extraction range set by the extraction range setting means, the pitch period is extracted while the extraction range If not, the pitch period extracting means for extracting the most probable pitch period within the extraction range, and the frame length specifying means for specifying the processing frame length according to the pitch period extracted by the pitch period extracting means, A time stretch processing device comprising: speed conversion means for performing speed conversion processing of an input signal with the processing frame length designated by the frame length designation means.   A pitch period extracting means for extracting the pitch period of the input signal; a determining means for determining whether the pitch period extracted by the pitch period extracting means is shorter than the shortest value; When a determination result indicating that the pitch period is long is obtained, a processing frame length corresponding to the pitch period is specified. When a determination result indicating that the pitch period is shorter than the shortest value is obtained from the determination unit, the pitch period is determined. A time stretch process comprising: a frame length designating unit for designating a processing frame length corresponding to an integral multiple of the input signal; and a speed conversion unit for performing a speed conversion process on the input signal with the processing frame length designated by the frame length designating unit. apparatus.   The pitch period extracting means for extracting the pitch period of the input signal, the frame length specifying means for specifying the pitch period extracted by the pitch period extracting means, and the processing frame length specified by the frame length specifying means for the input signal In a time stretch processing apparatus including a speed conversion unit that performs a speed conversion process, a determination unit that determines whether the pitch period extracted by the pitch period extraction unit is shorter than a shortest value is provided. When the determination result indicating that the pitch period is shorter than the shortest value is obtained, the speed conversion means performs the speed conversion process of the input signal a plurality of times with the processing frame length specified by the frame length specifying means. A time stretch processing device.   Extracted by the pitch period extracting means for extracting the pitch period of the input signal, the changing point detecting means for detecting a changing point at which the pitch period extracted by the pitch period extracting means changes by a predetermined value or more, and the pitch period extracting means. A frame length designating unit for designating a processing frame length corresponding to the pitch period, and a speed conversion process of the input signal with the processing frame length designated by the frame length designating unit from the change point detected by the change point detecting unit A time stretch processing device comprising speed conversion means for performing

Images