JP2001255876A - Method for expanding and compressing musical sound waveform signal in time base direction - Google Patents

Abstract

PROBLEM TO BE SOLVED: To compress and expand a musical sound waveform signal in the time base direction in a natural form. SOLUTION: A frequency band separation part 12 converts time-series waveform data presenting time variation of the musical sound waveform signal stored in a waveform data storage part 10 into each frequency band component waveform data of a high frequency band, a middle frequency band, and a low frequency band, and stores them in buffer parts 13a, 13b, 13c, respectively. Time base control parts 14a, 14b, 14c are controlled by a control part 20, and divide each frequency band component waveform data into time frame data of a prescribed time length in each frequency band, and rearrange each time frame data according to an inputted expansion/compression ratio (target tempo/ original tempo) to output them, respectively. A mixer part 15 mixes the rearranged and outputted waveform data and output them.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for compressing or expanding a musical tone waveform signal in the time axis direction without changing the frequency of the audibility, and to a method for expanding and contracting the musical tone waveform signal in the time axis direction.

[0002]

2. Description of the Related Art Conventionally, as shown in FIG. 7 (A), continuous musical tone waveform signals are divided at predetermined intervals (t0-t1, t1).
−t2, t2−t3...) In time series, and before and after each of the division points t0, t1, t2, t3.
The waveform data representing the tone waveform signal in the section to which t has been added is represented by first frame data FR1 and second frame data FR.
2, when the tone waveform signal is compressed or expanded in the time axis direction and output, the frame data is rearranged in time series and the read rate is set to the sampling rate. A method of expanding and contracting a tone waveform signal in which the tone waveform signal is compressed or expanded in the time axis direction without changing its frequency by outputting the tone waveform signal while maintaining the same is well known.

In this case, a musical sound waveform signal is output while maintaining an expansion / contraction ratio defined as a ratio of a target tempo to an original tempo (target tempo / original tempo) at “1.0” (ie, the original tempo). (A case of outputting the same tone waveform signal as the tone waveform signal of the first frame data), the first frame data FR1, the second frame data FR2, and the third tone data as shown in FIG.
.. Are read out at the same rate as the sampling rate, and the waveform data for a minute time ΔT at each end (only the rear end of the first frame data FR1) is subjected to cross-fade processing and output. . When a waveform signal is output with the expansion / contraction ratio larger than “1.0” (that is, when the original musical sound waveform signal is compressed and output), as shown in FIG. Are read out at the same sampling rate as the first frame data FR1, the second frame data FR2, the third frame data FR3,... At each time interval equal to a value obtained by dividing by the expansion / contraction ratio. The waveform data for a minute time ΔT before and after the division point that defines the time interval is cross-fade processed and output. When the waveform signal is output with the expansion / contraction ratio smaller than “1.0” (that is, when the original musical sound waveform signal is expanded and output), as shown in FIG. Are read out at the same sampling rate as the first frame data FR1, the second frame data FR2, the third frame data FR3,. After reading the frame data for a time interval equal to the value obtained by dividing the frame data by the expansion / contraction ratio, the process proceeds to reading of the next frame data. Note that fade-out and fade-in processing are performed at both ends of each frame data.

[0004]

However, in the above-mentioned conventional method, when the musical sound waveform signal is expanded and output,
When the degree of expansion is small (the expansion / contraction ratio is large), there is not much problem. However, when the degree of expansion is large, a gap is left between each frame data, and the discontinuity of the musical tone waveform signal based on each frame data becomes annoying. There is a problem.

In order to solve this problem, a method of filling the gap with the preceding or succeeding frame data is considered. This method is effective for sustained tone waveform signals, but when applied to attenuated tone waveform signals such as percussion instrument sounds and piano sounds, there is a problem that the attack portion is heard twice.

As another improvement method, a rising portion (attack portion) and a steady portion (sustain portion) of a tone waveform signal are detected, and a part or all of the frame data is repeatedly used for the steady portion. It is also conceivable that the gap is not formed. However, in this method, the repeated portion of the frame data becomes monotonous and may sound unnatural, and it is difficult to repeat the frame data so as not to be unnatural.

[0007]

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems, and has as its object to provide a musical sound waveform signal in the time axis direction which allows the musical sound waveform signal to be compressed and expanded in the time axis direction without being unnatural. An object of the present invention is to provide a method for expanding and contracting a signal.

In order to achieve the above object, the present invention is characterized in that time-series waveform data representing a time change of a musical tone waveform signal is input, and a plurality of different frequency bands included in the musical tone waveform signal are input. A band component data generating step of generating a plurality of time-series band component waveform data representing a time change of a plurality of band component waveform signals respectively belonging to the same based on the input time-series waveform data; and A dividing step of independently dividing the band component waveform data into time-series frame data having a predetermined length, and dividing each frame data of the plurality of time-series band component waveform data according to a designated expansion / contraction ratio. A rearrangement step of rearranging each of the time-series band component waveform data on the time axis, and mixing a plurality of time-series band component waveform data rearranged on the time axis, respectively. Lies in the arrangement by a mixing step.

In the above method, a plurality of time-series band component waveform data representing a time change of a plurality of band component waveform signals respectively belonging to a plurality of different frequency bands are generated from the time-series waveform data representing a time change of the musical tone waveform signal. Each time-series band component waveform data is independently divided into time-series frame data having a predetermined length, and each of the divided frame data is divided into time-series band component waveform data in accordance with a designated expansion / contraction ratio. After being rearranged on the time axis each time, they are mixed and output. Therefore, according to this method, the musical sound waveform signal is compressed or expanded on the time axis for each of a plurality of frequency bands, and optimal compression and expansion processing according to the frequency band can be performed. Further, when the musical sound waveform signal is expanded, since the time-series frame data is set independently for each frequency band, a time shift occurs in a portion corresponding to the gap for each frequency band,
As a whole, it is difficult to form a gap, and there is no need to employ a method of filling the gap portion with the previous or subsequent frame data as described above and a method of repeating the frame data with respect to the stationary part, which is not monotonous and natural. A simple musical sound waveform signal can be obtained. Also, when compressing the musical tone waveform signal, a natural musical tone waveform signal can be obtained because the connection portion of each frame data is temporally shifted for each frequency band and the connection portion is set inconspicuously. it can.

In this case, the dividing step is realized, for example, by dividing the plurality of time-series band component waveform data into time-series frame data having different lengths for each of the frequency bands. Further, the dividing step includes: an envelope detecting step of detecting each amplitude envelope of the plurality of band component waveform signals based on the generated plurality of time-series band component waveform data; and the plurality of time-series band component waveforms. An envelope dividing step of dividing data into time-series frame data for each of the frequency bands based on the detected amplitude envelopes. Further, these division methods can be selected for each band component waveform signal.

In the rearrangement step, for example, the reading start timing of each frame data of the plurality of time-series band component waveform data is changed according to the instructed expansion / contraction ratio, and the respective frame data is kept constant. At a read rate of

[0012]

DETAILED DESCRIPTION OF THE INVENTION a. First Embodiment Hereinafter, a first embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a functional block diagram of a tone signal generating device according to the first embodiment. In addition, each part described later,
It is composed of a microcomputer device and its peripheral devices, and most of them exhibit functions by program processing.

This musical tone signal generator comprises a waveform data storage unit 10, a readout unit 11, a band separation unit 12, and a buffer unit 1.
3a, 13b, 13c, time axis control units 14a, 14b,
14c and a mixing unit 15 are provided.

The waveform data storage unit 10 stores time-series waveform data representing a time change of a musical tone waveform signal and sampled at a predetermined sampling rate. In this case, the waveform data has a length of one song even for one song.
It may be a phrase or a single sound, and the type may be waveform data for each part such as a rhythm part, a piano, a guitar, or a mixture of a plurality of parts.

The reading unit 11 is controlled by the control unit 20 to read the waveform data stored in the waveform data storage unit 10 and supply the read waveform data to the band separation unit 12. The band separation unit 12 performs high-pass filter processing such as normal FIR type or IIR type, band-pass filter processing, and low-pass filter processing to perform a high-frequency component waveform signal included in a musical sound waveform signal represented by input waveform data. , Generating time-series low-frequency component waveform data, mid-frequency component waveform data, and high-frequency component waveform data representing respective time changes of the mid-frequency component waveform signal and the low-frequency component waveform signal based on the input waveform data. These waveform data are stored in the buffer unit 13.
a, 13b and 13c. Buffer unit 1
3a, 13b, and 13c are controlled by the control unit 20 to convert the generated low-frequency component waveform data, middle-frequency component waveform data, and high-frequency component waveform data into the time axis control units 14a, 14a.
An amount necessary for the control of 14b and 14c is temporarily stored.

The time axis control units 14a, 14b, 14c
Controlled by the control unit 20, the generated high-band component waveform data, middle-band component waveform data, and low-band component waveform data are each independently divided into time-series frame data having a predetermined length, and The data is rearranged and output on the time axis in accordance with the designated expansion ratio. In this division, the time length of each frame data of the high-frequency component waveform data, the middle-frequency component waveform data, and the low-frequency component waveform data is set to a predetermined length that is longer in this order. This rearrangement is performed at the same read rate as the sampling rate of the waveform data stored in the waveform data storage unit 10 at each time interval obtained by dividing the time length of each frame data by the expansion / contraction ratio (target tempo / original tempo). This is performed by reading out the respective waveform data temporarily stored in the buffers 13a, 13b, 13c.

The mixing section 15 includes time axis control sections 14a and 14
The high-frequency component waveform data, the mid-frequency component waveform data, and the low-frequency component waveform data output from b and 14c are mixed (added) and output to the D / A conversion unit 16. The D / A converter 16 converts the mixed waveform data output from the mixer 15 from digital to analog, and outputs the converted data to the sound system 17. The sound system 17 includes an amplifier, a speaker, and the like, and generates the analog-converted signal as a musical tone.

A virtual address counter 21, a panel operator 22, a panel display 23, and a communication interface 24 are connected to the controller 20.

The virtual address counter unit 21 determines the read time interval of each frame data by using the same expansion / contraction ratio (target tempo / original tempo) as the sampling rate of the waveform data stored in the waveform data storage unit 10. Is output to the control unit 20.
The original tempo is a performance tempo of a musical tone of the waveform data (tone waveform signal) stored in the waveform data storage unit 10, that is, without controlling the waveform data on a time axis.
When a musical tone corresponding to the reproduced waveform data is emitted by reproducing at a predetermined sampling frequency, it indicates a performance tempo felt by a person who has heard the musical sound. This performance tempo may be specified by the user or may be obtained by analyzing the waveform data. The target tempo is a performance tempo of a tone at the time of reproduction of the musical tone waveform signal, that is, a waveform reproduced by reproducing the waveform data while controlling the waveform data on a time axis to be described later. When a musical tone corresponding to the data is emitted, it indicates a performance tempo felt by a person who has heard the musical tone. The panel operator 22 is constituted by a plurality of operators operated by the user, and designates one of a plurality of waveform data stored in the waveform data storage unit 10 or specifies the same designated waveform data. It is used for inputting a tempo (original tempo) and a target tempo.

The panel display 23 displays characters, figures, and the like, and displays data necessary when the panel operator 22 is operated. The communication interface 24
It is composed of MIDI, USB, IEEE 1394, the Internet, etc., and externally inputs waveform data stored in the waveform data storage unit 10, and externally inputs data representing the type of waveform data, original tempo, target tempo, and the like. Are used to output these various data to the outside.

Next, the operation of the musical tone signal generating device configured as described above will be described. The user operates the panel operator 22 or fetches data from the outside via the communication interface 24 to specify the type of waveform data to be read from the waveform data storage unit 10 and to specify the original tempo and the target tempo. Is specified. In response to this, the control unit 20 outputs the address of the specified waveform data stored in the waveform data storage unit 10 to the reading unit 11, and based on the specified original tempo and target tempo. The expansion / contraction ratio is calculated and output to the virtual address counter unit 21. When the expansion ratio is directly input to the control unit 20 via the panel operator 22 or the communication interface 24, the calculation of the expansion ratio can be omitted.

Next, when an instruction to start reading waveform data is issued through the panel operator 22 or the communication interface 24, the control unit 20 causes the reading unit 11 and the buffer unit 13 to read.
a, 13b, 13c and time axis control units 14a, 14b,
A control signal is output to 14c to control output of the waveform data stored in the waveform data storage unit 10,
The virtual address counter 21 is instructed to start counting virtual addresses.

The reading section 11 sequentially reads out the waveform data stored in the waveform data storage section 10 at a rate higher than its sampling rate, and supplies it to the band separation section 12. The band separation unit 12 generates high-frequency component waveform data, middle-frequency waveform data, and high-frequency component waveform data based on the read waveform data, and
a, 13b and 13c. Buffer units 13a, 1
3b and 13c store the supplied waveform data, respectively. In this case, the reading unit 11 is the time axis control unit 1
Waveform data is read out intermittently for each time required for the processing of 4a, 14b, and 14c, and the buffer units 13a, 13b, and 13c store at least the waveform data for the time required for the same processing. For example, the buffer unit 1
3a, 13b, 13c include time axis control units 14a, 14
b and 14c store the frame data currently being read and the frame data before and after the data, and the reading unit 11 converts the waveform data corresponding to each waveform data to be stored in the buffers 13a, 13b and 13c into waveforms. Read from the data storage unit 10. The frame data separated by the band by the band separation unit 12 may be stored in advance in the buffers 13a, 13b, and 13c.

Here, the frame data will be described. The high-frequency component waveform data, the mid-frequency waveform data, and the high-frequency component waveform data have predetermined time intervals (t0-t10, t10-t20, t2) on the time axis, as in the case where the expansion ratio = 1 in FIG.
0-t30...), And these divided frame data are referred to as first frame data FR1, second frame data FR2, third frame data FR3. As shown in FIG. 3 (A), the time interval of this division gradually increases from the high frequency to the low frequency.

Returning to the description of the operation again, in the following description, when the set expansion / contraction ratio is "1.0", when it is larger than "1.0", and when it is smaller than "1.0" The description will be made in each case. First, the expansion ratio is "1.0"
In the case of (2), the virtual address counter unit 2
1 accumulates the expansion / contraction ratio “1.0” from the virtual address count start instruction at the same rate as the sampling rate of the waveform data stored in the waveform data storage unit 10, and sets the accumulation result as a virtual address value. It is sequentially supplied to the control unit 20. As shown in FIG. 2, the control unit 20 sets the division position address 1 corresponding to the division time position t10, t20, t30... The virtual address value is determined by each of the time axis control units 14a, 14b, 14c. .., And outputs a signal instructing the start of reading of the next frame data to the time axis control units 14a, 14b, and 14c, respectively.
As described above, each predetermined time t0-t10, t10-t20, t2
.. Are set so as to become longer in the order of the high-frequency component waveform data, the middle-frequency component waveform data, and the low-frequency component waveform data, so that the time axis control units 14a, 14b, 14
The read start instruction signal is supplied to c at every predetermined time which becomes longer in this order.

On the other hand, the time axis control units 14a, 14b, 14
c stores the first frame data FR1 stored in the buffer units 13a, 13b, 13c in the waveform data storage unit 1.
The data is sequentially read out at the same rate as the sampling rate of the waveform data stored in 0 and supplied to the mixing unit 15. When the time axis controllers 14a, 14b, and 14c finish reading the first frame data FR1, they wait until the controller 20 instructs the controller 20 to start reading the second frame data FR2. However, in this case, since the expansion / contraction ratio is "1.0", the timing at which the virtual address value becomes the division position address value corresponding to the division time position t10 determined by each of the time axis control units 14a, 14b, and 14c. And the time axis control units 14a, 14b, 14c
The timing of ending the reading of the frame data FR1 coincides with the timing. Therefore, the time axis control units 14a, 14
b and 14c start reading the second frame data FR2 at the same time as the reading of the first frame data FR1 ends. The time axis control units 14a, 1
As shown in FIG. 3A, the read start timings of the respective second frame data FR2 by 4b and 14c are different from each other, and are delayed in the order of the control units 14a, 14b and 14c.

The time axis control units 14a, 14b, 1
4c completes the reading of the second frame data FR2, at the time of the completion, the control units 14a, 14b,
14c is the third frame data F
The reading of R3 is started. In this way, the time axis control units 14a, 14b, 14c sequentially read the first frame data FR1, the second frame data FR2, the third frame data FR3,.
The mixture is supplied to the mixing unit 15. Strictly speaking, the time axis control units 14a, 14b, 14c sequentially connect and output the first frame data FR1, the second frame data FR2, the third frame data FR3,. In this case, each frame data before and after each connection part is cross-fade processed and connected.

As described above, the mixing section 15 outputs the first frame data FR1 and the second frame data FR supplied from the time axis control sections 14a, 14b and 14c, respectively.
2, the third frame data FR3,... Are added and synthesized and supplied to the D / A converter 16. Since the high-band component waveform data, the mid-band component waveform data, and the low-band component waveform data are added and synthesized by this addition synthesis, the waveform data obtained by the addition synthesis is equal to the waveform data read from the waveform data storage unit 10. Become something. The D / A converter 16 performs digital-to-analog conversion of the added and combined waveform data and supplies the converted data to the sound system 17, so that the system 17 emits a tone signal corresponding to the added and combined waveform data. As a result, in this case, the waveform data stored in the waveform data storage unit 10 is reproduced as it is without changing the tempo.

Next, a case where the expansion / contraction ratio is larger than "1.0" will be described. In this case, since the expansion / contraction ratio is set to a value larger than “1.0”, the virtual address counter 21 increases the virtual address value by accumulating the expansion / contraction ratio. , The virtual address value is faster than the divided time positions t10, t20, t30,... As shown in FIG. Reach the division position addresses 1, 2, 3,.... FIG. 2 shows an expansion / contraction ratio of 4 /
3 is shown. Therefore, the control unit 20 determines the time positions t11, t21, t
.., And outputs a read start instruction signal to the time axis control units 14a, 14b, 14c. Also in this case, the output time interval of the read start instruction signal to the time axis control units 14a, 14b, and 14c is also the same as the control unit 14a, 14b.
It becomes longer in the order of 14b and 14c.

On the other hand, the time axis control units 14a, 14b, 14
c indicates that the buffer units 13a and 13
The reading of the first frame data FR1, the second frame data FR2, the third frame data FR3,... stored in b and 13c, respectively, is started in order. Similarly, the sampling rate is the same as the sampling rate of the waveform data stored in the waveform data storage unit 10. Therefore, the time axis control units 14a, 14
b, 14c have not completed reading of the first frame data FR1, the second frame data FR2, the third frame data FR3,... at the time positions t11, t21, t31. Part of the latter half of each frame data is not read, but the next frame data is sequentially read and output. That is, in response to the instruction to read the next frame data, the reading of the frame data that has been read so far is terminated.

The time axis control unit 14
The frame data of the high-frequency component waveform data, the mid-frequency component waveform data, and the low-frequency component waveform data read out by a, 14b, and 14c are supplied to the mixing unit 15 where they are added and synthesized, and the D / A converter 16 and The sound is supplied to the sound system 17 and is emitted as a tone signal. As a result, in this case, the tone signal is compressed by a ratio corresponding to the expansion / contraction ratio in the time axis direction while the frequency of the tone signal to be sounded is kept in the state when it was stored in the waveform data storage unit 10. Thus, a tone signal is generated according to the target tempo set at a high speed. FIG. 3B shows the connection and compression state of each frame data of the high-band component waveform data, the middle-band component waveform data, and the low-band component waveform data. In this case as well, the connection section is subjected to cross-fade processing. . By this crossfade processing, even if waveform data that is not actually continuous is connected,
The discomfort caused by the connection can be minimized.

Next, a case where the expansion / contraction ratio is smaller than "1.0" will be described. In this case, since the expansion / contraction ratio is set to a value smaller than “1.0”, the increasing speed of the virtual address value due to the accumulation of the expansion / contraction ratio in the virtual address counter 21 is such that the expansion / contraction ratio is “1.0”. , The virtual address value is later than the divided time positions t10, t20, t30,... As shown in FIG. Reach the division position addresses 1, 2, 3,.... FIG. 2 shows an expansion / contraction ratio of 4 /
5 is shown. Therefore, the control unit 20 sets the time axis control unit 14a, 1 for each of the time positions t12, t22, t32,.
A read start instruction signal is output to each of 4b and 14c. In this case, also in this case, the time axis control units 14a, 14b,
The output time interval of the read start instruction signal with respect to 14c is
It becomes longer in the order of the control units 14a, 14b, 14c.

On the other hand, the time axis control units 14a, 14b, 14
c is the buffer unit 13a, 1 for each read instruction.
The reading of the first frame data FR1, the second frame data FR2, the third frame data FR3,... Stored in the memory cells 3b and 13c is sequentially started.
The reading rate in this case is also the same as the sampling rate of the waveform data stored in the waveform data storage unit 10, as in the case where the expansion / contraction ratio is "1.0". Therefore, the time axis control units 14a, 14b, 14c determine the time positions t12,
Since the reading of each of the first frame data FR1, each of the second frame data FR2, each of the third frame data FR3... has already been completed at a time earlier than t22, t32. There will be a gap between the next frame data. Although the gap may be left as it is, a part or all of each frame data may be used once or repeatedly to fill the gap.

The time axis control unit 14
The respective frame data of the high-frequency component waveform data, the mid-frequency component waveform data, and the low-frequency component waveform data read by a, 14b, and 14c are also supplied to the mixing unit 15 and added and synthesized, and the D / A converter 16 and The sound is supplied to the sound system 17 and is emitted as a tone signal. As a result, in this case, the tone signal is expanded by a ratio corresponding to the expansion / contraction ratio in the time axis direction, while maintaining the frequency of the tone signal to be generated in the waveform data storage unit 10 as it is. The tone signal is generated according to the target tempo set late. FIG. 3C shows an expanded state of each frame data of the high-frequency component waveform data, the middle-frequency component waveform data, and the low-frequency component waveform data. In this case, the end of each frame data is fade-out and fade-out. In processing is performed respectively.

As described above, according to the first embodiment, the processing of the band separation unit 12 allows the waveform data storage unit 1
From the time-series waveform data that is stored in 0 and represents the time change of the tone waveform signal read by the reading unit 11,
Time-series high-, middle-, and low-frequency band component waveform data are separately generated, and the buffer units 13a, 13b, 1
3c. Then, the time axis control unit 14
a, 14b, and 14c, the buffer units 13a,
Each of the band component waveform data stored in 13b and 13c is divided into time-series frame data of a predetermined length, which is sequentially increased in the order of high band, middle band and low band. Each band component waveform data is rearranged on the time axis in accordance with the instructed expansion / contraction ratio. In this rearrangement, for each band component waveform data, the read start timing of each frame data is determined according to the predetermined length and the designated expansion / contraction ratio for each band component waveform data, respectively. The reading is started from the determined reading start timing,
Each frame data is read out at the same sampling rate as when the waveform data in the waveform data storage unit 10 is stored. The high-band, middle-band, and low-band component waveform data are mixed by the processing of the mixing unit 15.

As a result, according to the first embodiment, the tone waveform signal is compressed or expanded on the time axis for each of a plurality of frequency bands, and compression and expansion processing suitable for the frequency band can be performed. Also, when expanding the musical tone waveform signal,
Since the time-series frame data is set independently for each frequency band, a time shift occurs in a portion corresponding to the gap for each frequency band, and it becomes difficult to form a gap as a whole, and Natural tone waveform signals can be obtained.
Further, when the gap is filled with frame data and when a musical tone waveform signal is compressed, a connection portion of each frame data is temporally shifted for each frequency band,
Since the connection portion is set inconspicuously, a natural musical sound waveform signal can be obtained.

In the description of the first embodiment,
The length of the frame data is different for each frequency band and only increases as the frequency band decreases, and the length and the division position of the frame data are not described in detail. Regarding the length of the frame data, it is preferable that the length of each frame data for each frequency band does not become an integral multiple so that the gaps and connection portions between the frame data for each frequency band do not overlap. May be an integral multiple depending on the system configuration. Also, regarding the division position of the frame data, it is preferable that the division positions of the frame data of all the bands are different from each other. However, some division positions may overlap due to the configuration of the system. Furthermore, if the processing of frame data is made different for each frequency band, such as cross-fade processing, fade-in processing, and fade-out processing for each frequency band, even if the division positions of the frame data for each frequency band overlap, ,
The gaps or joints between frame data become inconspicuous,
More preferred.

B. Second Embodiment Next, a second embodiment of the present invention will be described with reference to the drawings. FIG. 4 is a functional block diagram of a tone signal generating device according to the second embodiment. It should be noted that the second embodiment also includes a microcomputer device and its peripheral devices similarly to the first embodiment, and most of them exhibit functions by program processing. Since it has a common part, only the part different from the first embodiment will be described in the following description. The tone signal generating apparatus according to the second embodiment is characterized in that the frame data is divided based on the envelope of each band component waveform data, and the analysis for detecting the envelope of each band component waveform data is performed. Parts 30a, 30b, and 30c are provided.

Each of the analyzers 30a, 30b, 30c is arranged as shown in FIG.
And an envelope extraction unit 31 for extracting an amplitude envelope of each band component waveform data (that is, an amplitude envelope of a musical tone waveform signal represented by each band component waveform data). The envelope extracting unit 31 calculates, for example, the absolute value of the input waveform data, peak-holds the calculated absolute value every few milliseconds, and outputs the peak-held value by low-pass filtering. It consists of.

A rising detecting section 32 and a falling detecting section 33 are connected to the envelope extracting section 31. The rise detection unit 32 detects a rise time position of the amplitude envelope, and supplies data representing the time position to the time frame determination unit 34. This rising detector 3
In 2, the rise time position of the amplitude envelope is, for example, a predetermined time before (for example, several milliseconds before) the time when the amplitude envelope exceeds a predetermined threshold value or at the same time.
At a time point, when the differential value of the amplitude envelope exceeds a predetermined threshold value, or at a time point a predetermined time (for example, several milliseconds before) from the simultaneous point, when the amplitude envelope exceeds the predetermined threshold value and the differential value of the amplitude envelope is Is detected as a point in time that exceeds a predetermined threshold different from or a point in time that is a predetermined time (for example, several milliseconds before) from the simultaneous point.

The fall detecting section 33 detects the fall time position of the amplitude envelope (decay end time position of the amplitude envelope), and supplies data representing the same time position to the time frame determining section 34. In the fall detection unit 33, the fall time position of the amplitude envelope is, for example, a time point when the amplitude envelope falls below a predetermined threshold value or a time point a predetermined time (for example, several milliseconds before) from the simultaneous point, the amplitude envelope, When the differential value of the amplitude envelope falls below a predetermined threshold or at a predetermined time before (for example, several milliseconds before) the simultaneous point, the amplitude envelope falls below the predetermined threshold and the differential value of the amplitude envelope differs from the predetermined value. It is detected as a point in time below the threshold or a point in time a predetermined time (for example, several milliseconds before) from the simultaneous point. In addition,
The threshold value for detecting the rise time position may be the same as or different from the threshold value for detecting the fall time position.

The time frame determination unit 34 determines each rise time position as a start position of each frame based on the supplied data representing the rise time position and the fall time position, and also determines each rise time. The time from the position to the immediately following fall time position is determined as the length of each frame. When the next rise time position is detected after the rise time position is detected and before the fall time position corresponding to the same rise time position is detected, the time until the next rise time position is determined by the frame. Let the same rise time position be the start position of the next frame.

Thus, the analysis units 30a, 30b, 3
Data indicating the start position and the length of each frame determined by each time frame determination unit 34 of 0c is supplied to the time axis control units 14a to 14c. Time axis control units 14a, 1
4b and 14c are buffer units 13a, 13b. The high-frequency component waveform data, the mid-frequency component waveform data, and the low-frequency component waveform data stored in 13c are independently divided based on data representing the start position and length of each frame.
In this case, as shown in FIG. 6, in the high-frequency component waveform data, the time lengths corresponding to the lengths of the frames from the time positions t11, t12, t13. .. Are respectively set as first frame data FR1, second frame data FR2, third frame data FR3,. In the middle band component waveform data, the time lengths L21, L22,... Corresponding to the lengths of the frames from the time positions t21, t22.
-Each waveform data is converted to the first frame data FR
1, the second frame data FR2,...
In the low-frequency component waveform data, waveform data of time lengths L31, L32,... Corresponding to the lengths of the frames from time points t31, t32,. Are set as one frame data FR1, second frame data FR2,...

On the other hand, the control unit 20
Regarding the division of the frame data by a, 14b and 14c, a signal instructing the start of reading of each frame data is supplied to the time axis control units 14a, 14b and 14c, respectively. Also in this case, similar to the case of the first embodiment, the second
1 The control unit 20 calculates the expansion / contraction ratio (target tempo / original tempo) according to the input original tempo and target tempo, and the virtual address counter unit 21 samples the waveform data stored in the waveform data storage unit 10. The expansion / contraction ratio is accumulated at the same rate as the rate, and the accumulation result is stored in the control unit 20.
To supply. The control unit 20 determines that the virtual address values are divided time positions t11, t12, and t13 determined by the time axis control unit 14a.
Is output to the time axis control unit 14a each time the divided position address value corresponding to... Is reached, and the virtual address value is output to the time axis control unit 1
4b, a signal instructing the start of reading of the next frame data is output to the time axis control unit 14b each time the divided position address value corresponding to the divided time position t21, t22... Each time the value becomes a division position address value corresponding to the division time position t31, t32,... Determined by the time axis control unit 14c, a signal instructing the start of reading of the next frame data is sent to the time axis control unit 14c. Output each.

The time axis control units 14a, 14b, 14c
As in the first embodiment, the buffer units 13a, 13b,
The frame data stored in the waveform data storage unit 10 is sequentially read out at the same constant read rate as the sampling rate of the waveform data stored in the waveform data storage unit 10, thereby executing the rearrangement processing of each frame data. That is, as shown in FIG. 6, the time axis control unit 14a stores the virtual address value in the buffer unit 13a every time the virtual address value becomes each division position address value corresponding to the division time position t11, t12, t13. The first frame data FR1, the second frame data FR2, the third frame data FR3
.. Are sequentially read out at the constant readout rate and output to the mixing unit 15. Each time the virtual address value becomes a division position address value corresponding to the division time position t21, t22..., The time axis control unit 14b stores the first frame data FR1 and the second frame data stored in the buffer unit 13b. The data FR2... Are sequentially read at the constant read rate and output to the mixing unit 15. Each time the virtual address value becomes a division position address value corresponding to the division time positions t31, t32,..., The time axis control unit 14c stores the first frame data FR1 and the second frame data stored in the buffer unit 13c. The data FR2... Are sequentially read at the constant read rate and output to the mixing unit 15. In this case, it is not necessary to adopt a silent part (a part where the level of the waveform data is substantially “0”) as the frame data.

The mixing section 15 adds and synthesizes the respective waveform data from the time axis control sections 14a, 14b and 14c as in the first embodiment, and the added and synthesized waveform data is
The sound is generated as a tone signal via the D / A converter 16 and the sound system 17. As a result, in this case as well, the tone signal is expanded and contracted in the time axis direction by a ratio corresponding to the expansion and contraction ratio while maintaining the frequency of the tone signal to be generated in the waveform data storage unit 10. It is pronounced, and the tone signal is emitted at the target tempo.

As described above, also in the second embodiment, the processing of the band separation unit 12 allows the waveform data storage unit 1
From the time-series waveform data that is stored in 0 and represents the time change of the tone waveform signal read by the reading unit 11,
Time-series high-, middle-, and low-frequency band component waveform data are separately generated, and the buffer units 13a, 13b, 1
3c. Then, the time axis control unit 14
a, 14b, and 14c, the buffer units 13a,
Each band component waveform data stored in each of the band component waveform data is divided into a plurality of time-series frame data in accordance with each amplitude envelope of each band component waveform data. Are rearranged on the time axis for each band component waveform data. In this rearrangement, the reading start timing of each frame data is determined for each band component waveform data according to the start time position of each divided frame data for each band component waveform data and the designated expansion ratio. Then, the reading of each frame data is started from the determined read start timing, and each frame data is read at the same sampling rate as when the waveform data in the waveform data storage unit 10 is stored. The high-band, middle-band, and low-band component waveform data are mixed by the processing of the mixing unit 15.

As a result, according to the second embodiment, when the tone waveform signal is expanded, the time-series frame data is set independently for each frequency band. , A time lag occurs in the portion corresponding to, and it is difficult to form a gap as a whole, and a natural musical sound waveform signal can be obtained. When compressing the musical tone waveform signal, a natural musical tone waveform signal can be obtained because the connection of each frame data is shifted in time for each frequency band and the connection is set inconspicuously. Moreover,
Since each frame data is determined according to the rise and fall of each envelope of each band component waveform data, the gap when the musical tone waveform signal is expanded and the connection portion when the musical tone waveform signal is compressed are the same as those described above. It becomes more natural than in the case of the first embodiment, and a higher-quality tone signal can be obtained.

In the second embodiment, when time axis control is performed by dividing each band component waveform data, when the amplitude envelope is an attenuation system, the above method is used, and when the amplitude envelope is a continuous system, the amplitude envelope is a continuous system. In some cases, the above method is employed for the start position of each frame data, and a portion excluding a portion (transient portion) where the tone waveform signal changes rapidly, that is, a portion having a gradual change is deleted from each frame data. By repeating or repeating, the length of the frame data may be expanded or contracted. In this case, as shown by the broken line in FIG. 5, data indicating the amplitude envelope from the envelope extracting unit 31 is input to the time frame determining unit 34, and is represented by the waveform data input to the analyzing units 30a, 30b, and 30c. A frequency and other analysis unit 35 for analyzing changes in the musical tone elements such as the frequency and frequency characteristics of the musical tone waveform signal is provided, and analysis data of the musical tone elements from the analyzer 35 is input. Then, based on these input data, the time frame determination unit 34 determines a portion where the amplitude, frequency, frequency characteristics, etc. of the musical tone waveform signal does not change drastically, and converts the data representing the determined portion into a frame. The time axis control units 14a, 14a together with data representing the start position and the length
4b and 14c. Time axis control units 14a, 14
b and 14c delete or repeat the infrequently changing portion from the frame data defined by the frame start position and the frame length based on the supplied data, and perform buffering operations on the buffer units 13a, 13b and 13c. It is preferable to create new frame data from the waveform data stored in the.

In the second embodiment, the original tempo is inputted by the panel operator 22 or inputted from the outside through the communication interface 24. However, the original tempo is stored in the waveform data storage section 10. The original tempo may be detected based on the present waveform data. In this case, as indicated by broken lines in FIGS.
Each of the rise detectors 0a, 30b, and 30c detects the peak value of the amplitude envelope near the rise time position as the rise intensity when detecting the rise time position of the amplitude envelope of each band component waveform data. The rise detector 32 supplies the controller 20 with data representing the rise time position and the rise intensity of the amplitude envelope of the tone waveform signal for each frequency band.

The control unit 20 compares the rise time position and the rise intensity of the amplitude envelope for each frequency band with each other, thereby obtaining a series of musical tone waveform signals represented by the waveform data read from the waveform data storage unit 10. To detect beat positions and fluctuations in tempo. In this case, for example,
Based on the number of simultaneous or almost simultaneous rises of the amplitude envelope detected on the time axis and the magnitude of the rise strength, the beat position is detected, and the detected beat positions are classified into a front beat and a back beat, and the The basic tempo may be set by determining the beat position based on the beat, and the fluctuation of the tempo, that is, the swing rate (fluctuation rate) may be obtained from the deviation of the back beat from the front beat.

By using the basic tempo thus obtained as the original tempo, the original tempo can be automatically determined, and the necessity of inputting is eliminated. Also, if the swing rate is added to the basic tempo,
Precise time axis control can also be performed. Further, the basic tempo may be inputted from the outside as in the second embodiment, and the automatically determined swing rate may be added to the basic tempo.

C. Other Modifications In the above-described first and second embodiments, the original tempo is input from the outside via the panel operator 22 or the communication interface 24. The data representing the original tempo may be stored together with the waveform data. According to this, it is possible to calculate the expansion / contraction ratio of the waveform data using the stored original tempo without inputting the original tempo from the outside. In this case, the length of each frame data in the first and second embodiments is set to a quarter note, eighth note, 1
It is preferable that the length of each frame data determined by the amplitude envelope is adjusted by the time length.

In the first and second embodiments,
The target tempo is externally input through the panel operator 22 or the communication interface 24. Instead, a tempo clock signal having a cycle corresponding to the target tempo is input to the control unit 20, and the control unit 20 The target tempo may be calculated from the timing of the tempo clock signal.

In the first and second embodiments, the writing and reading of the high band, middle band, and low band component waveform data to and from the buffer units 13a, 13b, and 13c are performed by the same number of samples and the same period. To do it. However, since there is no problem even if the number of samples is reduced toward the low-frequency component, the number of samples of the low-frequency-side component waveform data separated by the band separation unit 12 is reduced and written to the buffer unit (down-sampling). )
The low frequency component side time axis control unit may read out the same sample value from the buffer unit redundantly (oversampling). For example, the sample values output from the band separation unit 12 are thinned out one by one and written to the buffer unit 13c, and the time axis control unit 14c overlaps and reads twice from the buffer unit 13c. Is also good.

Further, in the first and second embodiments, each band component waveform data of the high band, the middle band and the low band is generated based on the waveform data read from the waveform data storage unit 10. By doing so, the musical tone waveform signal is divided into three frequency bands and processed. However, the waveform data read from the waveform data storage unit 10 is divided into two or more frequency bands and processed. Is also good.
In this case, the number of buffer units and the number of time axis control units equal to the number of band separations are provided, and the band separation unit 12 uses the number of band component waveforms based on the waveform data read from the waveform data storage unit 10. Data may be generated, and these band-separated band component waveform data may be processed by the provided buffer unit and time axis control unit in the same manner as in the first and second embodiments.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an example of a tone signal generating device according to a first embodiment of the present invention.

FIG. 2 is a diagram illustrating a relationship between a division time position and a division position address and an output state of each frame data when an expansion ratio is variously changed.

FIGS. 3A to 3C are diagrams showing output states of high-, middle-, and low-band frame data when the expansion / contraction ratio is variously changed.

FIG. 4 is a block diagram showing an example of a tone signal generating device according to a second embodiment of the present invention.

FIG. 5 is a block diagram illustrating details of an analysis unit in FIG. 4;

FIG. 6 is a diagram illustrating an example of an amplitude envelope of each of high-band, middle-band, and low-band waveform data, and a division position of each frame data based on the same amplitude envelope.

FIGS. 7A to 7D are diagrams showing output states of respective frame data when the expansion ratio is variously changed by a conventional method.

[Explanation of symbols]

10: Waveform data storage unit, 12: Band separation unit, 13a,
13b, 13c: buffer section, 14a, 14b, 14c
... time axis control unit, 15 ... mixing unit, 20 ... control unit, 21 ...
Virtual address counters, 30a, 30b, 30c: analysis unit, 31: envelope extraction unit, 32: rising detection unit, 33: falling detection unit, 34: time frame determination unit.

Claims (4)

[Claims] A time series waveform data representing a time change of a musical tone waveform signal is inputted, and a plurality of time series waveform data included in the musical tone waveform signal and representing a time variation of a plurality of band component waveform signals respectively belonging to a plurality of different frequency bands. A band component data generating step of generating the time-series band component waveform data based on the time-series waveform data having the same input, converting the plurality of generated time-series band component waveform data into time-series frame data having a predetermined length. A dividing step of dividing each frame data independently, and rearranging each frame data of the plurality of time-series band component waveform data on the time axis for each of the time-series band component waveform data according to a designated expansion / contraction ratio. A time axis characterized by comprising a rearrangement step and a mixing step of mixing a plurality of time-series band component waveform data rearranged on the time axis, respectively. Stretching method tone waveform signal in direction. 2. The method of expanding and contracting a musical tone waveform signal in the time axis direction according to claim 1, wherein the dividing step includes:
A method for expanding and contracting a tone waveform signal in a time axis direction, comprising a step of dividing the plurality of time-series band component waveform data into time-series frame data having different lengths for each of the frequency bands. 3. The method of expanding and contracting a musical tone waveform signal in a time axis direction according to claim 1, wherein the dividing step includes:
An envelope detecting step of detecting each of the amplitude envelopes of the plurality of band component waveform signals based on the generated plurality of time-series band component waveform data; and An envelope dividing step of dividing the frequency band into time-series frame data for each frequency band based on an amplitude envelope. 4. The method of expanding and contracting a musical tone waveform signal in the time axis direction according to any one of claims 1 to 3, wherein the rearranging step includes the step of re-arranging each of the plurality of time-series band component waveform data. While changing the read start timing of the frame data according to the instructed expansion / contraction ratio,
A method of expanding and contracting a tone waveform signal in the time axis direction, in which each frame data is read at a constant read rate.