JP2008262140A - Musical pitch conversion device and musical pitch conversion method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To keep voice quality approximately to the same level as that before conversion, and to suppress noise generated at the time of the conversion, even when a musical pitch of voice is converted. <P>SOLUTION: In a band pass filter output switching section, on the basis of output of an output amplitude comparison section, selection of only one of outputs BPF15 to BPF 21 is repeated. A pitch period detecting section obtains a period of a voice signal on the basis of output from the band pass filter output switching section. A storage section for temporarily storing period information stores the period information detected by the pitch period detecting section. A musical pitch control section performs musical pitch conversion processing of voice data which are stored in a storage section for temporarily storing voice data, with a pitch period of the voice signal as a unit, on the basis of information of the storage section for temporarily storing period information, and outputs the result to the storage section for temporarily storing the voice data, and updates the storage content. The obtained musical pitch conversion processing result is controlled to output from the storage section for temporarily storing the voice data, via a voice data output section. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a pitch conversion device and a pitch conversion method for converting a pitch of an input audio signal into a desired pitch and outputting the pitch.

  2. Description of the Related Art Conventionally, an apparatus for converting a pitch (pitch period) by making a ratio of a sampling frequency of an input audio signal and a sampling frequency of an output signal equal to a ratio of a pitch to be converted has been proposed (for example, Patent Documents). 1). Also, after extracting the period of the audio signal and cutting out the waveform for each period of the audio, the pitch (pitch period) is converted by connecting the waveforms for each period at time intervals according to the ratio of pitch conversion. An apparatus has been proposed (see, for example, Patent Document 2).

Japanese Patent Publication No. 7-62800 JP 2001-265400 A

  The pitch conversion process in the prior art represented by the above-mentioned Patent Document 1 is theoretically equivalent to changing the running speed of the tape when reproducing the audio signal recorded on the magnetic tape. This technique can easily change the pitch (fundamental frequency) of the speech, but at the same time, the formant (peak due to resonance of the vocal tract) also changes, so the speaker-specific characteristics such as voice quality change are lost. However, there is a problem that the sound becomes unnatural due to the pitch change.

The pitch conversion process in the prior art represented by the above-mentioned Patent Document 2 is performed in order to eliminate the unnaturalness of the voice due to the above-described voice conversion, and the waveform is changed by changing the time interval for each cycle of the voice. Since the formant frequency remains constant even when the pitch of the voice changes, the voice quality is unlikely to deteriorate during the pitch conversion process. On the other hand, when joining with a wide time interval, a silent period is inserted between the periodic waveforms of the voice, so that the waveform of one period of the voice is isolated and a new harmonic is generated, resulting in deterioration of sound quality. There is a risk. Further, the waveform processing is performed by the window function to extract one period of the voice, but the deterioration of the sound quality at that time is not taken into consideration.
Accordingly, an object of the present invention is to maintain the converted voice quality substantially the same as that before the conversion even if the pitch of the voice uttered by a person is converted in the pitch conversion device, and occurs at the time of the conversion. It is to be able to suppress noise.

  A pitch converter according to the first aspect of the present invention covers pitch change rate setting means for changing the pitch of an input audio signal to a desired pitch, a frequency range of a human voice, and a band. A plurality of band pass filters that set each pass frequency band so as to divide the inside, harmonic component suppression means for obtaining a fundamental frequency signal that suppresses the harmonic component using each of these band pass filters; and The pitch period detecting means for detecting the period of the actual waveform of the audio signal based on the waveform of the fundamental frequency signal, and the pitch period detecting means obtained by the pitch period detecting means to control the pitch changed by the pitch change rate setting means. A pitch conversion device comprising pitch conversion means for performing pitch conversion by processing the actual waveform of the audio signal for each pitch period interval, wherein the harmonic component suppression means By comparing the threshold value that is changed based on the output value of each bandpass filter and the output value of each bandpass filter, the amplitude is equal to or greater than a certain amount, and a low frequency band is emphasized. Output amplitude comparison means for generating an output for selecting a filter is provided, and the pitch conversion means performs waveform connection processing for each pitch period interval obtained by the pitch period detection means.

  In a preferred embodiment according to the first aspect of the present invention, the pitch period detecting means sequentially detects the peak position of the waveform of the fundamental frequency signal or the vicinity thereof, and the detected order is continuous. It is a pitch period detection means for detecting an interval between peak positions or the vicinity of the peak positions.

    In another embodiment different from the above, the pitch period detection means sequentially detects the zero-cross position of the waveform of the fundamental frequency signal or the vicinity thereof, and the zero-cross position or the zero-cross position in which the order of detection is continuous. It is a pitch period detection means for detecting the vicinity.

    In another embodiment different from the above, the interval of the peak positions detected by the pitch period detecting means so that the pitch converting means controls the reproduction of the audio signal to the pitch set by the pitch change rate setting means. And means for converting the pitch by processing the actual waveform of the audio signal for the peak position or an intermediate point between the peak positions with the coincidence point on the time axis of the actual waveform of the audio signal as a processing position. It is.

    In an embodiment different from the above, a filter that selects and outputs only one of the outputs from each band-pass filter based on the output of the output amplitude comparison means at the subsequent stage of the output amplitude comparison means Output switching means is provided.

    A pitch conversion method according to the second aspect of the present invention includes a step for changing the pitch of an input audio signal to a desired pitch, a frequency range of a human voice, and a band division. As described above, a plurality of band-pass filters for setting each pass frequency band are prepared, and a step of obtaining a fundamental frequency signal in which harmonic components are suppressed using each of the band-pass filters and a waveform of the fundamental frequency signal are used. The step of detecting the actual waveform period (pitch period) of the audio signal and the pitch of the audio signal for each pitch period obtained in the pitch period detection step so as to control the pitch changed in the setting step. A pitch conversion method by performing waveform processing on an actual waveform as a target, and in the harmonic component suppression step, each bandpass In order to select a filter so that the amplitude is equal to or greater than a certain amount and the low frequency band is emphasized by comparing the threshold value that is changed based on the output value of the filter with the output value of each bandpass filter. In the step of generating the output for generating the output and the pitch conversion step, a waveform connection process is performed for each pitch cycle interval obtained in the pitch cycle detection step.

  According to the present invention, when performing the process of changing the pitch of the input voice in the pitch converter, it is avoided to perform the waveform processing for other than the specified periodic interval. Therefore, it is difficult to lose speaker-specific features such as formants, and high-quality pitch conversion is possible.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a functional block diagram showing an internal configuration of a pitch changing apparatus according to an embodiment of the present invention.

  As shown in FIG. 1, the pitch conversion device includes a voice data input unit 1, a pitch change rate setting unit 3, a voice data temporary storage unit 5, a periodic information temporary storage unit 7, and a voice feature. A detection unit 9, a pitch control unit 11, and an audio data output unit 13 are included.

  The audio feature detection unit 9 includes a band pass filter (hereinafter referred to as “BPF”) 15, BPF 17, BPF 19, BPF 21, output amplitude comparison unit 23, band pass filter output switching unit 25, and pitch period detection unit 27. And comprising.

  In each of these units, PCM data (PCM: Pulse Code Modulation) to be converted into pitches is input to the audio data input unit 1. The voice data input unit 1 to which the PCM data has been input outputs the PCM data to the voice data temporary storage unit 5 and the BPF 15, BPF 17, BPF 19, and BPF 21.

  The voice data temporary storage unit 5 is provided for performing pitch conversion processing on voice data, temporarily stores the input voice data, and stores the temporarily stored voice data in the pitch control unit 11. Under control, the sound data is converted into sound data suitable for the pitch conversion process and output to the sound data output unit 13.

  BPF15, BPF17, BPF19, and BPF21 set each passing frequency band so as to cover the frequency range of human voice and to divide the band.

  The output amplitude comparison unit 23 compares the outputs from the BPF 15, BPF 17, BPF 19, and BPF 21 for each predetermined section, and generates an output for selecting a filter whose amplitude is equal to or larger than a predetermined amount and places importance on a low frequency band. To do.

  The band-pass filter output switching unit 25 is provided in the subsequent stage of the output amplitude comparison unit 23, and based on the output of the output amplitude comparison unit 23, only one of the outputs of the BPF 15, BPF 17, BPF 19, and BPF 21 is output. Select and output.

  The pitch period detection unit 27 obtains the period of the audio signal based on the output from the bandpass filter output switching unit 25 and outputs the result.

  The period information temporary storage unit 7 stores the period information detected by the pitch period detection unit 27.

  The pitch control unit 11 converts the voice data stored in the voice data temporary storage unit 5 based on the information in the periodic information temporary storage unit 7 according to the change rate value set in advance by the pitch change rate setting unit 3. Then, the pitch conversion process is performed in units of the cycle of the audio signal, the result is output to the audio data temporary storage unit 5 and the stored content is updated. Control is performed to output the pitch conversion processing result thus obtained from the voice data temporary storage unit 5 via the voice data output unit 13.

  The voice feature detection unit 9 in the configuration including the above-described units will be described in detail. The voice feature detection unit 9 is provided with four band filters BPF15, BPF17, BPF19, and BPF21 in order to accurately detect a period (pitch period) determined by the pitch of a human voice. The pass band of each filter is determined so as to cover the frequency range of the speech to be converted and to divide the frequency range. Here, for example, as shown by the gain A-frequency f [Hz] characteristic in FIG. 2, the pass band of the filter of the BPF 15 is 100 to 200 [Hz] because the gain A in the pass band of the filter is 1.0. The pass band of the BPF 17 filter is 200 to 300 [Hz], the pass band of the BPF 19 filter is 300 to 400 [Hz], and the pass band of the BPF 21 is 400 to 500 [Hz].

  The outputs of BPF 15, BPF 17, BPF 19, and BPF 21 having the relationship shown in FIG. 2 are input to the output amplitude comparison unit 23. The output amplitude comparison unit 23 compares the outputs from the BPF 15, BPF 17, BPF 19, and BPF 21 for each predetermined section, and a threshold value that is set and changed at a period corresponding to the basic frequency obtained by the detection by the pitch period detecting means, and the BPF 15 , BPF17, BPF19, and BPF21 are compared with the average output value for each predetermined section, thereby generating an output for selecting a filter whose amplitude is not less than a certain amount and attaches importance to a low frequency band.

  The band pass filter output switching unit 25 repeats selecting only one of the outputs of the BPF 15, BPF 17, BPF 19, and BPF 21 based on the output of the output amplitude comparison unit 23. For this reason, when the sound signal from the sound data input unit 1 is input to the BPF 15, BPF 17, BPF 19, and BPF 21 as the waveform of the original sound as shown in FIG. 3, for example, the output of the bandpass filter output switching unit 25 is output. As shown in FIG. 4, the voice waveform includes the fundamental frequency component of the human voice and suppresses the harmonic component of the frequency, resulting in a substantially single sine wave.

  At the same time, in the pitch period detection unit 27, based on the output of the bandpass filter output switching unit 25, the processing for obtaining the peak position of the waveform and the distance (time) between two adjacent peak positions, the two points Processing for obtaining the pitch period A is performed. In FIG. 3 and FIG. 4, the positions indicated by reference numerals (1), (2), (3), (4), (5), and (6) are peak positions of the waveform. The output from the pitch cycle detection unit 27 is sequentially input to the cycle information temporary storage unit 7 and updated and stored as cycle information used by the pitch control unit 11.

  The pitch control unit 11 includes a pitch cycle detection unit 27 in the output of the bandpass filter output switching unit 25 stored in the cycle information temporary storage unit 7 in order to control the pitch changed by the pitch change rate setting unit 3. A processing reference position (position indicated by X in FIG. 3, hereinafter referred to as processing reference position X) is provided at intervals between the two detected peak positions and for each peak position interval, and the time axis in the actual waveform of the audio signal The pitch conversion is realized by performing waveform processing on the upper coincidence point. FIG. 5 shows an example of processing for lowering the pitch, and FIG. 6 shows an example of processing for raising the pitch.

  In the processing example for lowering the pitch in FIG. 5, a synthesis interval is defined between two peak positions appropriately selected according to the setting change in the pitch change rate setting unit 3, and is expanded on the time axis centering on the processing reference position X. To synthesize. At this time, in order to smoothly connect the waveforms, the known cross-fade process is performed for a certain section (composition section) centered on the processing reference position X. For other period intervals, the crossfade process is similarly performed according to the setting change in the pitch change rate setting unit 3.

  In the example of increasing the pitch in FIG. 6, a synthesis interval is defined between two peak positions appropriately selected according to the setting change in the pitch change rate setting unit 3, and narrowed on the time axis centering on the processing reference position X. To synthesize. At this time, in order to smoothly connect the waveforms, the known cross-fade process is performed for a certain section (composition section) centered on the processing reference position X. For other period intervals, the crossfade process is similarly performed according to the setting change in the pitch change rate setting unit 3.

  FIG. 7 is a detailed explanatory view schematically showing a processing example for lowering the pitch. First, in accordance with the setting change in the pitch change rate setting unit 3, as shown in FIG. 7 (a), the sound corresponding to the pitch period A between the two points of the peak position (3) and the peak position (4) is captured and processed. A processing range is provided by equally dividing the time width before and after the reference position X. Next, in this processing range, the audio amplitude after the processing reference position X is reduced from the original amplitude to 0 as shown in FIG. 7B, and the processing reference position X is shown in FIG. 7C. The voice amplitude before is increased from 0 to the original amplitude. As a result, the time width overlapped around the processing reference position X is included in the sound of FIG. 7B and the sound of FIG. 7C. Next, as shown in FIG. 7 (d), in the section of the overlapped time width, which is a fixed section (composite section) centering on the processing reference position X, the voice of FIG. 7 (b) and FIG. 7 (c). By performing a well-known cross-fade process that synthesizes and matches the voice of the voice, the period interval of the voice is increased from A to A ′. The pitch can be lowered as desired by performing the cross-fading process in the same manner in accordance with the setting change in the pitch change rate setting unit 3 for other periodic intervals.

  FIG. 8 is a detailed explanatory view schematically showing an example of processing for raising the pitch. First, in accordance with the setting change in the pitch change rate setting unit 3, as shown in FIG. 8 (a), the sound of the pitch period A between the two points of the peak position (3) and the peak position (4) is captured and processed. Centering on the reference position X, the front and back blank time widths are equally divided to provide respective processing ranges separated in the front and rear. Next, among the processing ranges separated before and after this, the speech amplitude is reduced from the original amplitude to 0 as shown in FIG. 8 (b) for the processing range before the processing reference position X, and For the processing range after the processing reference position X, the audio amplitude is increased from 0 to the original amplitude as shown in FIG. As a result, the voice of FIG. 8B and the voice of FIG. 8C include a blank time width centering on the processing reference position X. Next, as shown in FIG. 8 (d), in the interval of the blank time width, which is a fixed interval (composite interval) centered on the processing reference position X, the voice of FIG. 8 (b) and FIG. 8 (c). By performing a known cross-fading process for synthesizing with the voice of the voice, the period interval of the voice is narrowed from A to A ″. The same applies to other period intervals according to the setting change in the pitch change rate setting unit 3. By performing the crossfading process, the pitch can be raised as desired.

  As described above, in one embodiment of the present invention, the band-pass filter output switching unit 25 selects only one of the outputs of the BPF 15 to BPF 21 based on the output of the output amplitude comparison unit 23. Is repeated. The pitch period detection unit 27 obtains the period (pitch period) of the audio signal based on the output from the bandpass filter output switching unit 25. The period information temporary storage unit 7 stores the period information detected by the pitch period detection unit 27. The pitch control unit 11 controls the pitch period in the output of the bandpass filter output switching unit 25 stored in the cycle information temporary storage unit 7 so as to control the reproduction of the audio signal to the pitch set by the pitch change rate setting unit 3. The pitch conversion process is performed with reference to the period information detected by the detection unit 27, the result is output to the voice data temporary storage unit 5, and the stored content is updated. Control is performed to output the pitch conversion processing result thus obtained from the voice data temporary storage unit 5 via the voice data output unit 13.

  Therefore, according to one embodiment of the present invention, the pitch conversion is performed under the condition that the period information of the audio signal is obtained with high accuracy as described above, so that noise generated during the conversion can be effectively suppressed. It becomes possible. Also, in any process of lowering or raising the pitch, waveform processing is not performed for anything other than the specified periodic interval. This makes it possible to convert high-quality pitches without losing voice characteristics.

  The preferred embodiment of the present invention has been described above, but this is an example for explaining the present invention, and the scope of the present invention is not limited to this embodiment. The present invention can also be implemented in various other forms. For example, in the description using FIGS. 3 to 8, the processing reference position X is the center position between two adjacent peak positions. The processing reference position X may be set at any position between the two adjacent peak positions.

  Further, the processing reference position X may be set not only at one point but also at two or more points between two adjacent peak positions.

    Further, the position near the peak position is regarded as the peak position, the peak period is set, or the zero-cross position in the output of the bandpass filter output switching unit 25 or the vicinity thereof is used as a reference position for determining the peak period instead of the peak position. Of course, such applications as possible are possible.

The functional block diagram which shows the internal structure of the pitch converter which concerns on one Embodiment of this invention. The figure which shows the gain one frequency characteristic used in order to demonstrate the condition where two or more band pass filters were prepared. The signal waveform diagram which shows an example of the output waveform from the audio | voice data input part described in FIG. FIG. 2 is a signal waveform diagram related to a process of pitch conversion processing of input voice data performed in a pitch control unit described in FIG. 1. FIG. 2 is a signal waveform diagram related to a process of pitch conversion processing of input voice data performed in a pitch control unit described in FIG. 1. FIG. 2 is a signal waveform diagram related to a process of pitch conversion processing of input voice data performed in a pitch control unit described in FIG. 1. The schematic diagram regarding the process of the pitch conversion process of the input audio | voice data performed in the pitch control part described in FIG. The schematic diagram regarding the process of the pitch conversion process of the input audio | voice data performed in the pitch control part described in FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Voice data input part 3 Pitch change rate setting part 5 Voice data temporary storage part 7 Period information temporary storage part 9 Voice feature detection part 11 Pitch control part 13 Voice data output part 15 Band pass filter (BPF)
17 Bandpass filter (BPF)
19 Band pass filter (BPF)
21 Bandpass filter (BPF)
23 Output amplitude comparison unit 25 Band pass filter output switching unit 27 Pitch period detection unit

Claims (7)

A pitch change rate setting means for changing the pitch of the input audio signal to a desired pitch;
Prepare a plurality of bandpass filters that set each pass frequency band so as to cover the frequency range of human voice and divide the band, and use the output value of each bandpass filter to determine the harmonic components. Harmonic component suppression means for obtaining a fundamental frequency signal of the suppressed speech;
Pitch period detecting means for detecting the period of the audio signal (pitch period) based on the waveform of the fundamental frequency signal;
In order to control the pitch to be changed by the pitch change rate setting means, the pitch is converted by processing the actual waveform of the audio signal with a period corresponding to the pitch period obtained by the pitch period detecting means. A pitch changing device comprising pitch changing means,
In the harmonic component suppression means, the amplitude is a certain amount or more by comparing the threshold value that is set and changed with reference to the output value of each bandpass filter and the output value of each bandpass filter. And an output amplitude comparing means for generating an output for selecting a filter so that a low frequency band is emphasized,
In the pitch conversion means, a pitch conversion apparatus that performs waveform connection processing at intervals corresponding to the pitch period obtained by the pitch period detection means.   2. The pitch changing device according to claim 1, wherein the pitch period detecting means sequentially detects peak positions of the waveform of the fundamental frequency signal or the vicinity of the peak positions, and peak positions where the detected ranks are continuous or their positions. A pitch changing device which is a pitch period detecting means for detecting an interval between peak positions.   2. The pitch changing device according to claim 1, wherein the pitch period detecting means sequentially detects a zero-cross position of the waveform of the fundamental frequency signal or a vicinity thereof and a zero-cross position in which the detected order is continuous or the zero-cross position. A pitch converter which is a pitch period detecting means for detecting the vicinity of positions.   3. The pitch changing device according to claim 2, wherein the pitch changing means detects the peak position detected by the pitch period detecting means so as to control the reproduction of the audio signal to a pitch set by the pitch change rate setting means. The pitch is converted by processing the actual waveform of the audio signal as a processing position at the interval and at the peak position or for each intermediate point of the peak position using the coincidence point on the time axis in the actual waveform of the audio signal as a processing position. A pitch converter as a means.   2. The pitch changing apparatus according to claim 1, wherein the waveform connection processing is performed by the pitch change rate setting means at a processing reference position that determines one or more voice waveforms for each pitch cycle in the middle of the cycle. The time width corresponding to the pitch that has been deleted is deleted, and the end faces are joined together to shorten the pitch period and increase the pitch, or the waveform connecting process is a voice waveform for each pitch period. Are cut out so that the time widths overlap according to the pitch set and changed by the pitch change rate setting means at a processing reference position determined at one or more points in the middle of the cycle, and the end faces are joined together A pitch conversion device that is a process for extending the pitch period and lowering the pitch.   5. The pitch conversion device according to claim 1, wherein, after the output amplitude comparison unit, based on the output of the output amplitude comparison unit, of the outputs from the bandpass filters, A pitch converter comprising filter output switching means for selecting and outputting only one of them. A step for changing the pitch of the input audio signal to a desired pitch;
Basically, multiple bandpass filters with each pass frequency band set to cover the frequency range of the human voice and to divide the band, and harmonic components are suppressed using these bandpass filters. Determining a frequency signal;
The pitch obtained by the detection in the pitch period detection step so as to control the pitch set by the step of detecting the period (pitch period) of the audio signal based on the waveform of the fundamental frequency signal and the pitch setting step. A pitch conversion method by performing pitch processing by subjecting the actual waveform of the audio signal to waveform processing in a cycle according to the cycle,
In the harmonic component suppression step, an amplitude is equal to or greater than a predetermined amount by comparing a threshold value that is changed with reference to an output value of each bandpass filter and an output value of each bandpass filter; Generating an output for generating an output for selecting a filter so that a low frequency band is emphasized;
In the pitch conversion step, a pitch conversion device that performs waveform connection processing at intervals corresponding to the pitch cycle obtained in the pitch cycle detection step.

Images