JP5141033B2 - Time axis companding device, time axis companding method and program - Google Patents

Description

  The present invention relates to a technique for changing a speed at which sound is reproduced.

  A technique of time-axis companding that changes the playback speed while maintaining the pitch and formant of the input voice has been proposed (for example, Non-Patent Document 1). FIG. 7 is a conceptual diagram for explaining a procedure for changing the reproduction speed of the output sound to s times the input sound. First, as shown in part (a) of FIG. 7, a section A (A1, A2,...) And a section B (B1, B2,. Detected from column PIN. The basic period LW is a section length (number of samples) in which the similarity between waveforms in the section A and the section B is maximum. Secondly, each sample in the output section SOUT composed of the companding section SA and the duplication section SB is sequentially output as the output audio sample string POUT.

When the reproduction speed s is lower than “1” (that is, when the output sound is reproduced at a lower speed than the input sound), the waveform of the section A and the waveform of the section B are synthesized as shown in the part (b) of FIG. The companding section SA is generated by inserting the waveform having the section length LW into the gap between the section A and the section B. On the other hand, when the reproduction speed s exceeds “1” (that is, when the output sound is reproduced at a higher speed than the input sound), as shown in the part (c) of FIG. The companding section SA is generated by replacing the waveform and the section B waveform with the section X waveform. The duplication section SB is a section in which each sample of the input speech is duplicated. The section length L of the output section SOUT is set according to the basic period LW and the playback speed s. By repeating the process of generating the output section SOUT according to the above procedure for each section SIN of the input sound, an output sound having a playback speed that is s times the input sound is generated.
Naotaka Morita and Fumitada Itakura, “Elongation and compression of speech using time-based overlap addition method (PICOLA) and its evaluation”, Proc. Of the Acoustical Society of Japan, October 1986, 1-4 -14 (p.149-p.150)

  In the above technique, as shown in FIG. 7, even if the playback speed s is changed at the time T during the output of the sample in the output section SOUT, the sample output continues until the end of the output section SOUT. Is done. That is, the change in the reproduction speed s is reflected in the output sound after the sample output is completed for the output section SOUT set before the reproduction speed s is changed. Therefore, there is a problem that the change in the playback speed of the output sound is delayed according to the section length L of the output section SOUT. Since the section length L of the output section SOUT becomes longer as the playback speed s is closer to “1”, the problem that the change in the playback speed is delayed becomes particularly significant. Against the background of the above circumstances, one object of the present invention is to solve the problem of quickly changing the audio playback speed.

In order to solve the above-described problems, a time-axis companding device according to the present invention includes a storage unit that stores a sample string indicating an audio waveform, and a speed index value (reproduction speed s and companding) that determines an audio reproduction speed. Based on the setting means for setting the rate a), the basic period of the voice, and the speed index value set by the setting means, the section length of the output section after companding (for example, the section length L in FIGS. 3 and 4) is determined. If the setting means changes the speed index value within the period in which the output means outputs each sample in the output section, the output means for sequentially outputting each sample in the output section, Of these, the remaining section length (for example, section length r1 in FIGS. 3 and 4) for which sample output has not been completed when the speed index value is changed, is the speed index value before the change and the speed index value after the change. According to the section length (for example, section length r2 in FIGS. 3 and 4) ; And a new to interval length updating means.

  According to the above configuration, when the speed index value is changed during the output of the sample in the output section, the section length of the remaining section after the change of the speed index value in the output section becomes the speed index value before the change. The section length is updated according to the changed speed index value. Therefore, compared to the configuration in which the change of the speed index value is reflected after the output of the sample is completed for the output section set before the change of the speed index value, the speed index value can be quickly changed with respect to the output sound. Can be reflected.

In a preferred aspect of the present invention, the section length determining means includes an output composed of a companding section in which a sample sequence of the voice is companded and a sample section in which the samples of the sample array are arranged in units of the basic period of the speech. The section length of the section is determined based on the basic period of the voice and the speed index value set by the setting means, and the section length update means is configured so that the setting means within the period in which the output means outputs each sample in the duplicate section. When the speed index value is changed, the section length of the remaining section after the change of the speed index value in the duplicate section is updated to the section length corresponding to the speed index value before the change and the speed index value after the change. According to the above aspect, when the speed index value is changed during the output of each sample in the duplication section, the remaining section of the duplication section in which the output of the sample is not completed when the speed index value is changed . The section length is updated to the section length according to the speed index value before the change and the speed index value after the change. Therefore, even if the duplication section extends for a long period according to the speed index value, the change of the speed index value in the duplication section can be quickly reflected on the output voice.

  In a preferred aspect of the present invention, the section length update means has a playback speed (for example, playback speed s1) corresponding to the speed index value before the change with respect to the playback speed (for example, playback speed s2) corresponding to the speed index value after the change. The section length is updated so that the lower the section length is, the lower the section length is. In a more specific aspect, the section length updating means determines the remaining section, the ratio of the playback speed corresponding to the speed index value before the change to the playback speed corresponding to the speed index value after the change, and the remaining section before the update. The section length is updated according to the multiplication value with the section length. According to the above aspect, it is possible to specify the updated section length by a common method regardless of the direction in which the playback speed is changed.

The time axis companding device according to the present invention is realized by hardware (electronic circuit) such as a DSP (Digital Signal Processor) dedicated to specific signal processing, or a general-purpose operation such as a CPU (Central Processing Unit). This is also realized by cooperation between the processing device and the program. The program according to the present invention is based on a setting process for setting a speed index value for determining the playback speed of the sound of the waveform indicated by the sample sequence stored in the storage means , and a speed index value set in the basic period of the sound and the setting process. Section length determination processing for determining the section length of the output section after companding, output processing for acquiring each sample in the output section from the storage means and outputting it sequentially, and processing each sample in the output section for output processing If the speed index value is changed during the output period, the length of the remaining section of the output section where the sample output has not been completed when the speed index value is changed is indicated as the speed index value before the change. And a section length update process for updating the section length to a section length corresponding to the changed speed index value. By the above program, the same operation and effect as the time-axis companding device of the present invention are exhibited. The program of the present invention is provided to a user in a form stored in a portable recording medium such as a CD-ROM and installed in a computer, or provided from a server device in a form of distribution via a network. Installed on the computer.

The present invention is also specified as a method of companding in the direction of the audio time axis. The time-axis companding method according to the present invention sets a speed index value that determines the playback speed of the sound of the waveform indicated by the sample sequence stored in the storage means , and sets the compression based on the basic period of the sound and the set speed index value. The section length of the output section after extension is determined, each sample in the output section is acquired from the storage means and sequentially output, while the speed index value is changed within the period in which each sample in the output section is output And the length of the remaining section of the output section where the output of the sample is not completed when the speed index value is changed to the section length according to the speed index value before the change and the speed index value after the change. Update. Also by the above time axis companding method, the same operation and effect as the time axis companding device of the present invention are exhibited.

<A: Time axis companding device>
FIG. 1 is a block diagram showing a configuration of a time axis companding device according to one embodiment of the present invention. The time axis companding apparatus 100 is an apparatus that generates an output voice by expanding or compressing a waveform of an input voice (voice signal) in the direction of the time axis. As shown in FIG. 1, the time axis companding device 100 includes a storage circuit 10, a control circuit 20, an output circuit 40, and a setting unit 50.

  A sample string PIN is sequentially supplied to the storage circuit 10 from the outside. The sample string PIN is a data string in which a large number of samples extracted from the input speech are arranged in time series. The storage circuit 10 functions as an input buffer circuit that sequentially stores each sample.

  The control circuit 20 is an arithmetic processing unit (CPU) that generates a sample sequence POUT of output speech from a sample sequence PIN of input speech by executing a program. More specifically, as shown in FIG. 2, the control circuit 20 determines the output audio sample sequence POUT based on the samples of the sections (hereinafter referred to as “input intervals”) SIN obtained by dividing the input audio sample sequence PIN. A sample of a specific section (hereinafter referred to as “output section”) SOUT is generated.

  The output circuit 40 is an output buffer circuit that accumulates each sample of the sample sequence POUT generated by the control circuit 20 and sequentially outputs each sample at a predetermined cycle. The sample train POUT output from the output circuit 40 is converted into an analog signal and then supplied to an output device such as a speaker or a headphone to reproduce output sound.

  The setting unit 50 is a means for setting the output audio reproduction speed s. The setting unit 50 according to this embodiment controls the reproduction speed s based on a command given from the host device. The playback speed s is defined as a numerical value that defines the ratio (s = s_in / s_out) of the number of samples s_in in the input section SIN of the input sound to the number of samples s_out in the output section SOUT of the output sound (0.5 ≦ s ≦ 2). , S ≠ 1). That is, the number of samples s_out in the output section SOUT with respect to the number of samples s_in in the input section SIN decreases as the reproduction speed s increases. Accordingly, when the reproduction speed s exceeds “1”, the output sound is reproduced at a higher speed than the input sound (hereinafter referred to as “high-speed conversion”), and when the reproduction speed s is less than “1”, the output sound is reproduced. Is reproduced at a lower speed than the input voice (hereinafter referred to as “low speed conversion”). Note that a configuration in which the setting unit 50 sets the playback speed s in accordance with a user operation on an input device (not shown) is also employed.

  As shown in FIG. 1, the control circuit 20 operates as a plurality of functional bodies (basic cycle specifying unit 32, section length determining unit 34, output unit 36, section length updating unit 38) by executing a program. Note that the function of each unit of the control circuit 20 may be realized by a hardware circuit such as a DSP.

  3 and 4 are conceptual diagrams for explaining the operation of each part of the control circuit 20. FIG. 3 shows the operation during low speed conversion (s <1), and FIG. 4 shows the operation during high speed conversion (s> 1). As shown in part (a) of FIG. 3 and part (a) of FIG. 4, the basic period specifying unit 32 specifies a basic period LW that is a time length of a section in which similar waveforms repeat in the input speech. . For example, the basic period specifying unit 32 performs a plurality of processes for calculating the similarity of the waveforms for the sections A and B adjacent to each other with the same section length in the input speech while changing the section lengths of the sections A and B. The section length in which the degree of similarity is maximized is identified as the basic period LW.

The section length determining unit 34 determines the section length L of the output section SOUT (the total number of samples belonging to the output section SOUT) based on the basic period LW specified by the basic period specifying unit 32 and the playback speed s set by the setting unit 50. It is a means to calculate. In the case of low speed conversion in which the reproduction speed s is lower than “1” (0.5 ≦ s <1), the section length determination unit 34 determines the section length L based on the following equation (1).
L = LW / (1-s) (1)
In the case of high-speed conversion in which the reproduction speed s exceeds “1” (1 <s ≦ 2), the section length determination unit 34 determines the section length L based on the following equation (2).
L = LW / (s-1) (2)
That is, the section length L of the output section SOUT becomes longer as the reproduction speed s is closer to “1”.

  The output unit 36 generates an output section SOUT of the sample sequence POUT from each sample of the sample sequence PIN stored in the storage circuit 10 and sequentially outputs each sample. As shown in part (b) of FIG. 3 and part (b) of FIG. 4, each output section SOUT is composed of a companding section SA and a duplication section SB.

  The companding section SA is a section in which the sample string PIN is companded using the basic period LW of the input voice as a unit. As shown in FIG. 1, the output unit 36 includes a companding processing unit 361 for generating a companding section SA. When the reproduction speed s is maintained at a value lower than “1”, the companding processing unit 361 uses the waveform of the section A (A1, A2,...) And the section B as shown in the part (b) of FIG. A companding section SA is generated by inserting a section X (X1, X2,...) Having a section length LW obtained by combining the waveforms (B1, B2,...) Into the gap between the section A and the section B. . Therefore, the section length of the companding section SA at the time of low speed conversion is “3 × Lw”. On the other hand, when the regeneration speed s is maintained at a value higher than “1”, the companding processing unit 361 performs the section A (A1, A2,...) And the section B as shown in a part (b) of FIG. The companding section SA is generated by replacing (B1, B2,...) With the section X (X1, X2,...) Obtained by synthesizing the waveform of the section A and the waveform of the section B. Therefore, the section length of the companding section SA at the time of high-speed conversion is “Lw”. The waveform of the section X is generated, for example, by crossfading the waveform of the section A and the waveform of the section B.

  The output unit 36 sequentially outputs the samples of the sections A and B acquired from the storage circuit 10 and the samples of the section X synthesized from the samples of both sections to the output circuit 40 as the samples in the companding section SA. . Further, the output unit 36 acquires each sample following the section B in the sample string PIN from the storage circuit 10 as each sample in the replication section SB, and sequentially outputs it to the output circuit 40. The number of samples acquired from the storage circuit 10 as each sample in the replication section SB is adjusted so that the output section SOUT becomes the section length L. For example, since the section length of the companding section SA at the time of low speed conversion is “3 × LW”, the output unit 36 starts immediately after section B in the sample string PIN as shown in part (b) of FIG. Samples over the section length “L-3 × LW” are sequentially output as the duplicate section SB. Further, since the section length of the companding section SA at the time of high-speed conversion is “LW”, the output unit 36 selects the section length immediately after section B in the sample string PIN as shown in part (b) of FIG. Samples over "L-LW" are sequentially output as the duplication section SB.

  In the above, it is assumed that the playback speed s is not changed. However, the playback speed s set by the setting unit 50 may be changed during the period in which the output unit 36 outputs each sample in the output section SOUT. is there. 1 is a means for updating the section length L of the output section SOUT to a new section length La based on the changed playback speed s when the setting section 50 changes the playback speed s. is there. In part (c) of FIG. 3 and part (c) of FIG. 4, the playback speed s1 is changed to the playback speed s2 at the time T during the output of the output section 36 for each sample in the replication section SB. The case is illustrated.

As shown in the part (c) of FIG. 3 and the part (c) of FIG. 4, the section length update unit 38 completes the output of the sample at the time T when the reproduction speed s is instructed in the duplication section SB. The section length r1 of the section that has not been changed (that is, the section from the time T when the playback speed s is changed to the end point of the output section SOUT, hereinafter referred to as “residual section”) is changed to the playback speed s1 before the change and the post-change. The section length r2 is updated according to the reproduction speed s2. More specifically, first, the section length update unit 38 sets the section length r2 so that the section length r2 of the remaining section SR after the update becomes shorter as the playback speed s1 before the change with respect to the playback speed s2 after the change is lower. Is calculated. Second, the section length updating unit 38 calculates the section length r2 so that the section length r2 becomes longer as the section length r1 of the remaining section SR before the change of the playback speed s is longer. For example, as shown in the following equation (3), the ratio (s1 / s2) of the playback speed s1 before the change to the playback speed s2 after the change and the section length r1 of the remaining section SR before the change of the playback speed s The multiplication value is calculated as the section length r2.
r2 = r1 x (s1 / s2) (3)
The output unit 36 sequentially acquires from the storage circuit 10 each sample over the section length r2 from the first sample that has not been output at the time T when the playback speed s is changed in the sample sequence PIN as the remaining section SR. Output to the output circuit 40.

  Next, the flow of processing performed by each unit of the control circuit 20 will be described with reference to FIG. First, the basic period specifying unit 32 specifies the basic period LW of the input voice (step S1). Next, the section length determination unit 34 calculates the section length L of the output section SOUT from the basic period LW specified in step S1 and the playback speed s set by the setting unit 50 based on the formula (1) or the formula (2). Determine (step S2). The output unit 36 sequentially outputs each sample in the companding section SA specified by the companding processing unit 361 to the output circuit 40 (step S3).

  When the output of all the samples in the companding section SA is completed, the output unit 36 outputs one sample in the duplication section SB to the output circuit 40 (step S4), and in the output section SOUT (in the duplication section SB). It is determined whether or not the output of all the samples has been completed (step S5). If the result of step S5 is affirmative, the basic period specifying unit 32 specifies a new basic period LW using the sample immediately after the sample output in the immediately preceding step S4 in the sample string PIN as the starting point of the input interval SIN. (Step S1). As described above, the processing from step S1 to step S4 is repeated a plurality of times, so that the output sound sample sequence POUT reproduced at the reproduction speed of s times the input sound is sequentially stored in the output circuit 40.

  On the other hand, if the result of step S5 is negative, the section length update unit 38 determines whether or not the setting unit 50 has changed the playback speed s (step S6). If the result of step S6 is negative, the process proceeds to step S4 and the output of the sample in the duplication section SB is continued. On the other hand, when the result of step S6 is affirmative (that is, when the playback speed s is changed), the section length update unit 38 sets the section length r1 of the remaining section SR at the current stage to the playback speed s1 before the change. Are updated to the section length r2 corresponding to the relative ratio of the playback speed s2 after the change (step S7), and the output of each sample belonging to the remaining section SR after the update is executed in step S4. After the process of step S7, the result of step S5 is affirmed when the output of all the samples within the section length r2 is completed. That is, the sample immediately after the updated residual section SR (section length r2) is used as the starting point of the input section SIN (step S1) and the output of each sample in the output section SOUT (step S2 to step S2). S4) is executed.

  Next, specific examples of the relationship between the section length r2 of the remaining section SR after the update and the playback speed s before and after the change will be listed. First, when the playback speed s is changed from “0.8 (s1)” to “0.9 (s2)” when the section length r1 of the remaining section SR is “90” (that is, the playback speed s is increased within the low speed conversion range). ), The section length r2 of the remaining section SR after the update is “80”. Also, when the playback speed s is changed from “0.9 (s1)” to “0.8 (s2)” when the section length r1 is “400” (when the playback speed s is reduced within the range of the low speed conversion), the section The length r2 is “450”.

  When the playback speed s is changed from “1.2 (s1)” to “1.6 (s2)” when the section length r1 is “80” (that is, when the playback speed s is increased within the range of high-speed conversion), the section length r2 is "60". Also, when the playback speed s is changed from “1.5 (s1)” to “1.2 (s2)” when the section length r1 is “100” (that is, when the playback speed s is reduced within the range of high-speed conversion), The section length r2 is “125”.

  Further, when the playback speed s is changed from “0.8 (s1)” to “1.2 (s2)” when the section length r1 is “240” (that is, when the low speed conversion is changed to the high speed conversion), the section length r2 is “160”. Further, when the playback speed s is changed from “1.2 (s1)” to “0.8 (s2)” when the section length r1 is “200” (that is, when the high speed conversion is changed to the low speed conversion), the section length r2 is “300”.

  As illustrated above, regardless of the direction of change of the playback speed s and whether or not there is a change across “1”, the expression (3) is used in common, and the playback speed s1 is shorter as the playback speed s2 is lower. Further, the section length r2 is calculated such that the longer the section length r1, the longer the section length r1.

  As described above, when the reproduction speed s increases, the duplication section SB (output section SOUT) is shortened. That is, without waiting for the output of all the samples belonging to the output section SOUT before the update, the output of samples (calculation of the basic period LW and the section length L) according to the reproduction speed s after the change is started. Therefore, the change in the reproduction speed s is reflected in the output sound more quickly than the conventional configuration in which the change in the reproduction speed s is reflected after the completion of the initial output section SOUT (hereinafter referred to as “proportional 1”). It becomes possible to make it. As understood from the equations (1) and (2), when the reproduction speed s before the change is close to “1”, the time length of the duplication section SB becomes long. The change delay becomes noticeable. In light of the above circumstances, the present embodiment in which the change in the reproduction speed s is quickly reflected by the update of the remaining section SR is particularly effective when the reproduction speed s is increased from a value close to “1”.

  Further, when the playback speed s increases, the remaining section SR is shortened (r2 <r1), and when the playback speed s decreases, the remaining section SR is extended (r2> r1). That is, the relationship between the length of the remaining section SR and the change direction of the playback speed s matches the relationship that the section length L of the output section SOUT becomes shorter as the playback speed s is higher. Therefore, from another point of view, it can be understood that the change in the reproduction speed s is reflected in the output sound when the remaining section SR is updated to the section length r2. Even from the above viewpoint, the expected effect of the present embodiment that the change of the reproduction speed s is reflected promptly is certainly achieved.

  It should be noted that the problem that the change in the playback speed s is delayed under the conventional technique is that the playback speed s is set near “1” and the section length L of the output section SOUT is set to a long period. To be particularly serious. In other words, if a configuration in which the playback speed s is not set near “1” (for example, a configuration in which the change rate of the playback speed s is increased, hereinafter referred to as “comparative 2”), the playback speed s is changed. It is possible to suppress the delay formally. However, in contrast 2, there is a problem in that convenience is lost due to a significant decrease in the resolution of the reproduction speed s. On the other hand, in the present embodiment, even when the playback speed s is set near “1” (that is, when the variation range of the playback speed s is sufficiently reduced), the change in the playback speed s is reflected quickly. There is an advantage of being.

  By the way, from the viewpoint of reflecting the change in the reproduction speed s promptly, the output section SOUT corresponding to the changed reproduction speed s is newly started immediately after the change in the reproduction speed s is instructed. (Hereinafter referred to as “Comparison 3”) may also be employed. That is, in the comparative example 3, as shown in FIG. 6, when the change of the playback speed s is instructed at the time T in the middle of the duplication section SB, the sample immediately after the time T in the sample string PIN is input to the input section. A new output section SOUT is set as the head of SIN (that is, identification of the basic period LW and section length L and output of each sample are started).

  However, in contrast 3, since the output section SOUT is shortened each time the reproduction speed s is changed, for example, when the reproduction speed s is frequently changed in a short time, the substantial reproduction speed of the output sound is There is a possibility of deviation from the actual reproduction speed s. That is, since the output section SOUT is shortened, the number of samples in the output section SOUT with respect to the number of samples in the input section SIN (that is, the playback speed s) is reduced, so the output sound is compared with the playback speed s set by the setting unit 50. As a result, the reproduction speed becomes substantially low. When the playback speed s is near “1”, the section length L of the output section SOUT is long as understood from the formulas (1) and (2), so the playback speed s of the output section SOUT is changed. As a result, the section length (section after the time point T) shortened is inevitably increased. Therefore, the difference between the substantial reproduction speed of the output sound and the reproduction speed s set by the setting unit 50 is particularly significant.

  On the other hand, in this embodiment, after the reproduction speed s is changed, the output of the sample over the section length r2 is completed, and then the next output section SOUT is started. Moreover, the section length r2 increases or decreases according to the playback speed before and after the change. Therefore, even when the reproduction speed s is frequently changed in a short time, the frequency at which the output section SOUT is newly started is reduced from the comparative example 3. That is, according to the present embodiment, the deviation (error) between the substantial reproduction speed of the output sound and the reproduction speed s set by the setting unit 50 is suppressed as compared with the proportionality 3 (the output sound reproduction speed is high). Can be controlled with precision).

<B: Modification>
Various modifications can be made to each of the above embodiments. An example of a specific modification is as follows. In addition, you may combine each following aspect suitably.

(1) Modification 1
In the above embodiment, the setting unit 50 sets the output sound playback speed s. However, the setting target by the setting unit 50 is not limited to the playback speed s. For example, a configuration in which the setting unit 50 sets a companding rate a (a = s_out / s_in) that is a ratio of the number of samples s_out in the output section SOUT to the number of samples s_in in the input section SIN is also employed. In the configuration in which the drawing ratio a is set, the regeneration speed s in each of the above forms may be replaced with the reciprocal of the drawing ratio a. As described above, it is sufficient that the numerical value set by the setting unit 50 is an index value (speed index value) that determines the audio playback speed, and the playback speed s and the companding rate a are merely examples of speed index values.

(2) Modification 2
The method of calculating the section length L of the output section SOUT and the section length r2 of the remaining section SR (formula (1) to formula (3)) is appropriately changed. Further, in the above embodiment, the configuration in which the section length L and the section length r2 are calculated by the calculation based on the formulas (1) to (3) is illustrated, but the section length L based on a table created in advance. A configuration in which the section length r2 is specified is also adopted. For example, a table in which the section length L is associated with each combination of the reproduction speed s and the basic period LW is created, and the reproduction speed s set by the setting unit 50 and the basic period specified by the basic period specifying unit 32 are created. A configuration in which the section length determination unit 34 searches for the section length L corresponding to the period LW is also employed. Similarly, a table in which the section length r2 is associated with each combination of the playback speed s1 before change, the playback speed s2 after change, and the section length r1 of the remaining section SR is prepared, and the setting unit 50 A configuration is also employed in which the section length update unit 38 searches for the section length r2 corresponding to the playback speed s1 before change, the playback speed s2 after change, and the section length r1.

(3) Modification 3
In the above embodiment, the configuration in which the section length r1 of the remaining section SR of the duplication section SB is updated when the reproduction speed s is changed during the output of the sample in the duplication section SB is exemplified. A configuration is also adopted in which the section length r1 of the remaining section SR is updated when the playback speed s is changed during the output of the samples. For example, when the playback speed s is changed during the period in which the output unit 36 outputs the sample in the section B1 (in the companding section SA) shown in the part (b) of FIG. In the section B1, the section length r1 of the remaining section SR obtained by adding the section after the change of the playback speed s and the immediately following duplicate section SB is updated to the section length r2 based on the playback speed s before and after the change. As described above, in one embodiment of the present invention, the section length r1 of the remaining section SR after the change of the playback speed s in the output section SOUT is updated according to the playback speed s before and after the change. In other words, it is not always necessary to distinguish the companding section SA and the duplicating section SB with respect to the update of the remaining section SR.

(4) Modification 4
In the above embodiment, the low speed conversion is realized by inserting the waveform (section X) obtained by synthesizing the waveform of section A and the waveform of section B into the gap between section A and section B. The method is appropriately changed. For example, a configuration in which the waveform of the section A is inserted redundantly in the gap between the section A and the section B is also employed. In the above embodiment, high-speed conversion is realized by replacing the waveform (section X) obtained by synthesizing the waveform of section A and the waveform of section B with section A and section B. The method is also arbitrary. For example, a configuration that deletes the waveform of the section A or the section B is also employed. As described above, the scope to which the present invention is applied is not limited to the time-axis companding device adopting PICOLA (Pointer Interval Controlled OverLap and Add) of Non-Patent Document 1.

It is a block diagram which shows the structure of the time-axis companding apparatus which concerns on embodiment of this invention. It is a conceptual diagram which shows the relationship between an input area and an output area. It is a conceptual diagram for demonstrating operation | movement of the control circuit at the time of low speed conversion. It is a conceptual diagram for demonstrating operation | movement of the control circuit at the time of high-speed conversion. It is a flowchart which shows operation | movement of a control circuit. It is a conceptual diagram for demonstrating comparative operation | movement. It is a conceptual diagram explaining the procedure of time-axis companding.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 ... Time-axis companding device, 10 ... Memory circuit, 20 ... Control circuit, 32 ... Basic period specific part, 34 ... Section length determination part, 36 ... Output part, 38 ... Section length update part , 40... Output circuit, 50... Setting unit, PIN... Input audio sample sequence, POUT... Output audio sample sequence, SIN... Input segment, SOUT. ... Section length of remaining section before update, r2 ... Section length of remaining section after update.

Claims (6)

Storage means for storing a sample string indicating the waveform of the sound;
Setting means for setting a speed index value for determining the playback speed of the sound;
Section length determining means for determining the section length of the output section after companding based on the basic period of the voice and the speed index value set by the setting means;
Output means for sequentially outputting each sample in the output section;
If the setting means changes the speed index value within the period in which the output means outputs each sample in the output section , sample output is not completed when the speed index value is changed in the output section A time-axis companding device comprising: section length updating means for updating the section length of the remaining section to a section length according to the speed index value before the change and the speed index value after the change. The section length determination means is a section length of the output section composed of a companding section in which a sample sequence of the voice is companded in units of the basic period of the voice and a duplicate section in which samples of the sample series are arranged. Is determined based on the basic period of the voice and the speed index value set by the setting means,
The section length update means samples each sample in the duplication section when the speed index value is changed in the duplication section when the setting means changes the speed index value within a period in which the output means outputs. The time-axis companding device according to claim 1, wherein the section length of the remaining section in which the output is not completed is updated to a section length according to the speed index value before the change and the speed index value after the change. The section length update means updates the section length so that the section length of the remaining section becomes shorter as the playback speed corresponding to the speed index value before the change with respect to the playback speed corresponding to the speed index value after the change is lower. The time-axis companding device according to claim 1 or 2. The section length update unit is configured to calculate a ratio of the playback speed corresponding to the speed index value before the change to the playback speed corresponding to the speed index value after the change, and the section length of the remaining section before the update. The time axis companding device according to claim 3, wherein the time length is updated to a section length according to the multiplication value. On the computer,
A setting process for setting a speed index value that determines the playback speed of the sound of the waveform indicated by the sample sequence stored in the storage means ;
A section length determination process for determining a section length of the output section after companding based on the basic period of the voice and the speed index value set in the setting process;
An output process of outputting the sample in the output section sequentially obtains from the storage unit,
If the speed index value is changed within a period in which each sample in the output section is output in the output process, sample output is not completed when the speed index value is changed in the output section. A program for executing a section length update process for updating the section length of the remaining section to a section length corresponding to the speed index value before the change and the speed index value after the change. Set a speed index value that determines the playback speed of the sound of the waveform indicated by the sample sequence stored in the storage means ,
Determine the section length of the output section after companding based on the basic period of the voice and the set speed index value,
While outputs each sample in the output section sequentially to retrieve from said memory means,
If the speed index value is changed within a period for outputting each sample in the output section, the section length of the remaining section in which output of the sample is not completed when the speed index value is changed in the output section The time axis companding method is used to update the section length according to the speed index value before the change and the speed index value after the change.

Images