JP4580297B2 - Audio reproduction device, audio recording / reproduction device, and method, recording medium, and integrated circuit - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To realize reproduced voice quality that is easier to here by performing an optimum velocity ratio control according to fluctuation of a voice content rate. <P>SOLUTION: A voice period and a non-voice period are determined in a voice/no-voice determination section 11 and a voice content rate in a frame length for calculation is calculated as voice information in a voice information calculation section 12, and an average value and a standard deviation of the voice content ratio are calculated. Then, the velocity ratio in the voice period is calculated for each frame using the voice information in a velocity ratio calculation section 14, and the velocity ratio in the non-voice period is calculated so that regeneration time may become a target regeneration time using the velocity ratio in the voice period. Based on the calculated velocity ratio in the voice period and the non-voice period, a regeneration velocity of an input voice signal is changed in a voice velocity changing section 15 and the voice signal is output. <P>COPYRIGHT: (C)2007,JPO&amp;INPIT

Description

  The present invention relates to an audio reproducing device, an audio recording / reproducing device, and a method thereof, a recording medium, and an integrated circuit. More specifically, the present invention relates to an audio reproducing device, an audio recording / reproducing device, and the like that convert and reproduce the reproducing speed. And a recording medium and an integrated circuit.

2. Description of the Related Art Conventionally, there has been known a method of reproducing at higher speed without changing the pitch of a voice in an audio reproducing apparatus that reproduces prerecorded audio (for example, see Patent Document 1). In the audio reproduction device disclosed in Patent Document 1, when the entire audio signal is reproduced at a specified speed, the reproduction speed ratio is partially reduced for the audio section. As a result, the conventional audio reproduction device disclosed in Patent Document 1 can provide reproduced audio that is easy to listen to with little loss of information.
JP 2001-222300 A

  Hereinafter, with reference to FIG. 11, the conventional audio reproduction device 9 disclosed in Patent Document 1 will be specifically described. FIG. 11 is a block diagram showing a configuration of a conventional audio reproduction device 9. In FIG. 11, the conventional audio reproduction device 9 includes an acoustic analysis unit 91, a speech speed conversion unit 92, a non-speech section length control unit 93, and a synthesis unit 94.

  The acoustic analysis unit 91 discriminates a voice segment and a non-speech segment from input voice data based on a preset power threshold. And the acoustic analysis part 91 calculates | requires the time information of a voice area and a non-voice area, respectively. In the conventional audio reproduction device 9 shown in FIG. 11, different reproduction processes are applied to the audio segment and the non-speech segment determined by the acoustic analysis unit 91. The voice data of the voice section determined by the acoustic analysis unit 91 and the time information are output to the speech speed conversion unit 92. The voice data of the non-speech section determined by the acoustic analysis unit 91 is output to the non-speech section length control unit 93.

  The speech speed conversion unit 92 first specifies a speech section sandwiched between non-speech sections having a predetermined time length or more based on the speech data of the speech section and each time information. Then, the speech speed conversion unit 92 performs speed ratio control such that the speed ratio at the beginning of the speech section is slower than the predetermined speed ratio and gradually returns to the predetermined speed ratio toward the end. The voice data of the voice section in which the speed ratio is controlled is output to the synthesis unit 94. Further, the speech speed conversion unit 92 outputs the delay time information of the voice segment generated by the waveform expansion process to the non-speech segment length control unit 93.

  On the other hand, the non-speech section length control unit 93 appropriately performs a process of deleting and compressing the speech data in the non-speech section based on the delay time information output from the speech speed conversion unit 92. That is, the non-speech interval length control unit 93 performs a process so as to meet the target designated speed ratio and eliminate the delay of the speech interval caused by the speech rate conversion unit 92. The voice data of the non-speech section processed by the non-speech section length control unit 93 is output to the synthesis unit 94.

  The synthesizer 94 synthesizes the voice data of the voice section output from the speech speed conversion unit 92 and the voice data of the non-voice section output from the non-voice section length control unit 93. The synthesizing unit 94 then outputs the final reproduced audio using the audio data obtained by synthesizing the audio section in which the speed ratio is converted and the non-audio section as converted audio data.

  In the conventional audio reproduction device 9, for example, when m-times speed (m is a positive number of 1 or more) is given as the designated speed, reproduction is performed at a speed ratio slower than m-times speed at the beginning of the audio section. Then, the conventional audio reproduction device 9 gradually increases the reproduction speed ratio toward the end of the audio section. Here, in general, there are many cases where important information is included at the beginning of the speech section. Therefore, according to the conventional audio reproducing device 9, it is possible to realize reproduction that is easy to listen to without missing important information at the beginning of the audio section. As described above, in the conventional audio reproduction device 9, processing that is easy to hear is performed for the voice section, and processing that is adapted to the designated speed ratio is performed for the non-voice section.

  Here, at the time of high-speed playback, the speech utterance speed increases, and the load for the user to understand the content increases. Furthermore, if the voice sections are concentrated and concentrated in the entire program (if the voice is continuously uttered), it becomes more difficult for the user to understand. However, the conventional audio reproducing apparatus 9 assumes only changing the reproduction speed ratio in one audio section. That is, in the conventional audio reproduction device 9, the same speed ratio control process is applied throughout the entire television program, for example. Therefore, the conventional audio reproduction device 9 has an essential problem that it is relatively difficult to listen to the audio content at a portion where the audio sections are concentrated and concentrated.

  SUMMARY OF THE INVENTION Therefore, an object of the present invention is to provide an audio reproducing device, an audio recording / reproducing device, and a method and a recording medium for realizing an easy-to-listen reproduction by performing optimum speed ratio control in consideration of the entire program such as a television. And to provide an integrated circuit.

A first aspect of the present invention is an audio reproduction apparatus for listening to an audio signal by applying an audio speed conversion process to the input audio signal to shorten the reproduction time from the normal time and listening to the audio signal. at least calculating a discrimination unit for discriminating between non-speech section without the voice section and a voice, based on the speech segment and the non-speech interval, the speech content indicating a ratio that contains the audio section by pairs in a predetermined time length, including The speed ratio of the voice information calculation unit and the speed ratio of the voice speed conversion processing is set in advance as a reference value, and the speed of the voice section in the predetermined time length when the voice content rate of the predetermined time length is relatively high change the ratios to Do value smaller than the reference value, the speed ratio to change the speed ratio of the speech interval to larger than the reference value in the predetermined time length when the sound content of the predetermined time length is relatively low and a calculation unit, sound Ru der performs adaptive speed control for adjusting the speed ratio of the speech segment relative temporal variations of the content.

In a second aspect based on the first invention, the speed ratio calculating section sets the shortened playback time in accordance with the operation by the user, based on the speed ratio change in the speech interval, reproduction of the audio signal The speed ratio of the non-speech section is calculated so that the playback time is set as the time.

  According to a third aspect, in the second aspect, the speed ratio calculation unit calculates the speed ratio of the non-speech section to be constant within the set reproduction time.

A fourth aspect of the invention is variable in the first aspect, velocity ratio calculation unit, the adaptation degree of the speed ratio of the speech intervals for the temporal variation of the sound content by selecting a set value of the predetermined time length It is characterized by doing.

In a fifth aspect based on the first aspect, the audio reproduction device includes a buffer for recording the audio signal while sequentially updating the audio signal to include at least a predetermined time length of the audio signal. A speed conversion unit that performs an audio speed conversion process on the audio signal recorded in the buffer and outputs the audio signal, and a determination unit includes an audio section for the audio signal of a predetermined time length recorded in the buffer. The voice information calculation unit further determines a non-voice section, calculates a statistical value related to the voice content rate of a predetermined time length, and sequentially updates the statistical value calculated so far for each unit time, thereby calculating a speed ratio. The unit calculates the speed ratio of the voice section according to the statistical value updated every unit time and the voice content rate set to the predetermined time length at the time of the update, and the speed conversion unit is sequentially updated in the buffer For audio signals Characterized by sequentially performing the voice speed conversion processing by using the speed ratio of the speech interval calculated for each unit time.

In a sixth aspect based on the first aspect, the voice information calculation unit further calculates a statistical value related to the voice content rate of a predetermined time length, and the speed ratio calculation unit calculates the statistical value and the voice content rate of a predetermined time length. The speed ratio of the voice section is calculated according to the above.

According to a seventh invention, in the fifth or sixth invention, the statistical value is an average value and a standard deviation of a voice content rate having a predetermined time length .

Eighth aspect of the present invention is described above in the seventh aspect, the speed ratio calculating unit, the difference value and the speed ratio coefficient calculated using the standard deviation therefore obtained value to the average value of the audio content in a predetermined time length The speed ratio of the voice section is calculated by multiplying by the reference value.

Ninth aspect of the invention related to the seventh invention, the audio information calculation unit for a predetermined length of time duration is different each respectively calculated sound content and statistics set up multiple speed ratio calculating unit, where and calculating by multiplying the reference value of the difference value and the speed ratio engagement number calculated using the values obtained for each predetermined time length by the standard deviation to the average value of the audio content in a constant time length.

A tenth aspect of the present invention is an audio reproduction method for listening to an audio signal by applying an audio speed conversion process to the input audio signal and reducing the reproduction time from the normal time. At least calculating a discrimination step of discriminating between non-speech section without the voice section and a voice, based on the speech segment and the non-speech interval, the speech content indicating a ratio that contains the audio section by pairs in a predetermined time length, including The speed ratio of the voice information calculation step to be performed and the speed ratio of the voice speed conversion process of 1 or more is set in advance as a reference value, and the speed of the voice section in the predetermined time length when the voice content rate of the predetermined time length is relatively high speed change ratio to small values than the reference value, to change the speed ratio of the speech interval in the predetermined time length when the sound content of the predetermined time length is relatively low larger than the reference value Ratio calculation And a step.

An eleventh aspect of the present invention is a computer that records a sound reproduction program that is executed by a computer for applying a sound speed conversion process to an input sound signal to shorten the reproduction time and listening to the sound signal. A readable recording medium, wherein the computer has a discrimination step for discriminating a voice section including voice and a non-voice section containing no voice with respect to the voice signal, and a predetermined time length based on the voice section and the non-voice section. audio information calculation step for at least calculating a speech content indicating a ratio that contains the audio segment for the ratio of the speech speed conversion process is preset as a reference value to 1 or more speed ratios, the audio content of a predetermined time length the predetermined time speed ratio of the speech interval in the length change in Do value smaller than the reference value, a predetermined time length audio content is relative when but a relatively high The predetermined time speed ratio of the speech interval in the length recording a program for executing the speed ratio calculation step of changing the larger value than the reference value, which is a computer-readable recording medium when low.

A twelfth aspect of the invention is an integrated circuit for listening to an audio signal by applying an audio speed conversion process to the input audio signal and reducing the playback time from the normal time, and includes the audio with respect to the audio signal. a determination unit that determines the non-speech section without the voice section and a voice, based on the speech segment and the non-speech section, at least calculated audio content indicating a ratio that contains the audio section by pairs in a predetermined time length When the ratio of the voice information calculation unit and the voice speed conversion process is 1 or more as a reference value, and the voice content rate of the predetermined time length is relatively high, the speed ratio of the voice section in the predetermined time length is the reference change in small than the value, the speed ratio calculating unit for changing the speed ratio of the speech interval to larger than the reference value in the predetermined time length when the sound content of the predetermined time length is relatively low Prepare.

A thirteenth aspect of the present invention is an audio recording and reproducing apparatus for listening to an audio signal by applying an audio speed conversion process to the input audio signal and reducing the reproduction time from the normal time. an information recording unit for recording, and a discrimination unit for discriminating between non-speech section without the voice section and audio including audio to the front of the audio signal to be recorded in the information recording unit, based on the speech segment and the non-speech section , and versus a predetermined time length set in advance as the reference value and the audio information calculation unit, one or more speed ratio ratio of the speech speed conversion processing for at least calculating a speech content indicating a ratio that contains the speech segment, the speed ratio of the speech interval in the predetermined time length when a predetermined time length audio content is relatively high change in Do value smaller than the reference value, when the speech content of the predetermined time length is relatively low In the predetermined time length The speed ratio of the voice interval and a speed ratio calculating unit for changing the larger value than the reference value.

In a fourteenth aspect based on the thirteenth aspect, the information recording unit records a result determined by the determining unit when the audio signal is recorded, and the audio information calculating unit is recorded in the information recording unit. Based on the result, a voice content rate of a predetermined time length is calculated.

According to a fifteenth aspect, in the thirteenth aspect, the information recording unit records a result determined by the determination unit and an audio content rate of a predetermined time length when the audio signal is recorded, and a speed ratio calculation unit Is characterized in that the speed ratio of the voice section is calculated using the voice content rate of a predetermined time length recorded in the information recording unit.

A sixteenth aspect of the present invention is an audio recording / reproducing method for listening to an audio signal by applying an audio speed conversion process to the input audio signal to shorten the reproduction time from the normal time. an information recording step of recording, a determination step of determining a non-speech section without the voice section and audio including audio to the front of the audio signal to be recorded on the information recording step, on the basis of the voice section and the non-speech section, audio information calculation step for at least calculating a speech content indicating a ratio that contains a predetermined length of time in pairs to the audio section, the ratio of the speech speed conversion process is preset as a reference value of 1 or more speed ratios, a predetermined to change the speed ratio of the speech interval to Do value smaller than the reference value in time length the predetermined length of time when the speech content is relatively high, when the speech content of the predetermined time length is relatively low The speed ratio of the speech section in said predetermined constant-time length and a speed ratio calculation step of changing the larger value than the reference value.

According to a seventeenth aspect of the present invention, there is provided a recording medium on which an audio recording / reproducing program executed by a computer for applying an audio speed conversion process to an input audio signal to shorten the reproduction time and listening to the audio signal. An information recording step for recording an audio signal input to the computer in the recording unit, an audio segment including audio and a non-audio segment not including audio with respect to the audio signal before being recorded in the recording unit, a discriminating step of discriminating that, based on the speech segment and the non-speech section, and the audio information calculation step of at least calculating a speech content indicating a ratio that contains the audio section by pairs in a predetermined time length, the ratio of the speech speed conversion process There one or more the speed ratio as a reference value, small than the reference value and a speed ratio of the speech interval in the predetermined time length when the sound content of the predetermined time length is relatively high Change, a program for executing the speed ratio calculation step of changing the speed ratio of the speech interval in the predetermined time length when the sound content of the predetermined time length is relatively low larger than the reference value A recorded recording medium that can be read by a computer.

  According to the first invention, by calculating the speed ratio of the voice section according to the fluctuation of the voice content rate, the reproduction sound quality after the speed conversion of the input voice signal is understood according to the fluctuation of the voice content rate. Can be excellent.

According to the second invention, so that the reproduction time set does not contain important audio information and a speed ratio of the non-speech section by separately calculating the speed ratio of the speech interval, the speech section The speed ratio can be adjusted to a speed ratio within the range that the user can hear.

  According to the third aspect of the invention, it is possible to perform playback with efficient speed conversion by setting the speed ratio of the non-voice section in which no important voice information is included to a constant speed ratio.

According to a fourth aspect of the invention, when the location constant time length even is long, the speed ratio of the speech segment to be adjusted is a high value of global and more accurate with respect to variations in speech content. In addition, for example, when the predetermined time length is short , the speed ratio of the adjusted voice section is sensitive to the fluctuation of the voice content rate and becomes a value with better followability. That is, by selecting a set value of a predetermined time length and varying the adaptation degree of the speed ratio of the voice section with respect to the fluctuation of the voice content rate , the accuracy or followability can be freely selected.

  According to the fifth aspect, by updating the statistical value every unit time, it is possible to immediately perform the speed conversion process according to the input of the audio signal and reproduce it.

  According to the sixth invention, by using the statistical value for calculating the speed ratio of the voice section, the speed ratio of the voice section can be calculated more in line with the actual fluctuation of the voice content rate. Thus, the reproduced sound quality after speed conversion can be made more natural and understandable.

  According to the seventh aspect, it is possible to calculate the speed ratio of the voice section in consideration of the degree of bias of the existence of the voice section.

  According to the eighth aspect of the invention, it is possible to calculate the speed ratio of the voice section in accordance with the degree of bias of the existence of the voice section.

According to the ninth aspect, it is possible to calculate the speed ratio of the optimum speech section corresponding to both the sensitive variation of voice content and global variations.

  According to the thirteenth aspect, by calculating the speed ratio of the voice section according to the fluctuation of the voice content rate, the reproduced sound quality after the speed conversion of the recorded voice signal is changed according to the fluctuation of the voice content rate. It can be excellent.

  According to the fourteenth aspect, it is possible to shorten the processing time from the recording of the audio signal to the time before the speed-converted reproduction is performed by the processing time in the determination unit.

  According to the fifteenth aspect, it is possible to shorten the processing time after recording an audio signal and before performing speed-converted reproduction by the processing time in the determination unit and the audio information calculation unit. , Playback with speed conversion can be performed immediately.

(First embodiment)
With reference to FIG. 1, the audio | voice reproduction apparatus which concerns on the 1st Embodiment in this invention is demonstrated. FIG. 1 is a block diagram showing a configuration of an audio reproduction device 1 according to the first embodiment of the present invention. In FIG. 1, the audio reproduction device 1 includes an audio / non-audio discrimination unit 11, an audio information calculation unit 12, an audio information recording unit 13, a speed ratio calculation unit 14, and an audio speed conversion unit 15. Note that the audio playback device 1 according to the present embodiment is a device that assumes that the entire recorded audio signal can be read once before the audio signal recorded on the recording medium is speed-converted and reproduced. is there. Here, examples of the recording target include a television and a radio program. Further, the recording medium may be a recording medium such as a DVD in which a movie or the like is recorded in advance. In the following description, as an example, it is assumed that the audio reproduction device 1 according to the first embodiment performs speed conversion processing on the audio signal of a recorded television program.

  An audio signal recorded on a recording medium or the like is read and input to the audio / non-audio discrimination unit 11. The voice / non-voice discrimination unit 11 analyzes the power envelope value and periodicity of the input voice signal. Then, the voice / non-speech discrimination unit 11 discriminates a speech segment and a non-speech segment on the time axis with respect to the input speech signal. Information on the voice section and the non-voice section discriminated on the time axis of the voice signal (hereinafter referred to as discrimination information) is output to the voice information calculation unit 12 before performing speed-converted reproduction.

The voice information calculation unit 12 calculates voice information necessary for calculating the speed ratio between the voice section and the non-voice section based on the discrimination information of the voice / non-voice section. As the audio information, an audio content rate, an average value of the audio content rate, a standard deviation, and the like are calculated . Specifically, the audio information calculation unit 12 calculates the audio content rate throughout the recorded program, and then calculates the average value and standard deviation of the audio content rate. The voice content rate calculated by the voice information calculation unit 12, the average value of the voice content rate, and the standard deviation are recorded in the voice information recording unit 13. Hereinafter, the voice content rate, the average value of the voice content rate, and the standard deviation will be described.

つまり、数式（１）によって算出される第ｋフレームの音声含有率Ｒｉｓ＿ｎ（ｋ）は、算出用フレーム長に対して音声区間が含まれる時間比率を示す。 The voice content rate indicates a time ratio in which a voice section is included with respect to a predetermined number (at least one) of frames. The voice content rate is calculated for each frame. Here, the frame is a processing unit for analyzing the input audio signal, and the time length of the frame is the frame length. The frame includes a speech segment and / or a non-speech segment. Further, at least one frame used for calculating the voice content rate is set as a calculation frame, and the time length thereof is set as a calculation frame length. In the following description, as an example, the time length of one frame (one frame length) is 1 minute. In addition, the calculation frame length for calculating the voice content rate is n (n is a positive number). That is, since the length of one frame is 1 minute, the calculation frame is composed of n frames. Further, it is assumed that the number of frames of the entire recorded program is N (N is a positive number). The frame number is k (k = 1 to N), and the frame when the frame number is k is the “kth frame”. At this time, the voice content rate Ris_n (k) of the k-th frame is expressed by Equation (1).

That is, the voice content rate Ris_n (k) of the k-th frame calculated by Expression (1) indicates a time ratio in which a voice section is included with respect to the calculation frame length.

Here, with reference to FIG. 2 to FIG. 4, a calculation example of the voice content rate Ris_n (k) will be given. In FIG. 2 to FIG. 4, as an example, when calculating the audio content rate of a TV broadcast documentary program (30 minutes), calculation is performed with three types of calculation frame lengths of 1 minute, 5 minutes, and 10 minutes. . FIG. 2 is a diagram illustrating a calculation example of the voice content rate Ris_1 (k) when the calculation frame length is 1 minute. FIG. 3 is a diagram illustrating a calculation example of the voice content rate Ris_5 (k) when the calculation frame length is 5 minutes. FIG. 4 is a diagram illustrating a calculation example of the voice content rate Ris_10 (k) when the calculation frame length is 10 minutes. 2 to 4, the horizontal axis represents the frame number (k), and the vertical axis represents the voice content rate (%) .

  In FIG. 2, the audio content rate Ris_1 (1) of the first frame (k = 1) is the audio content rate of the first frame from Equation (1) because the calculation frame length is 1 minute. In FIG. 3, the voice content rate Ris_5 (1) of the first frame calculated from the formula (1) is an average of the voice content rates of the first to fifth frames in FIG. In FIG. 4, the voice content rate Ris_10 (1) of the first frame calculated from the formula (1) is an average of the voice content rates of the first to tenth frames in FIG.

As shown in FIG. 2 to FIG. 4, it can be seen that the variation of the voice content rate is different for each calculation frame length. Specifically, when the calculated frame length is short (Fig. 2) is relatively large fluctuation between the frames of the speech content. That is, when the calculation frame length is short, the actual fluctuation of the voice content rate is sensitively reflected. In contrast, as shown in FIGS. 3 and 4, as calculated for the frame length is long, relatively small is fluctuation between the frames of the speech content. This is because, as described above, the audio content rate of each frame is averaged as the calculation frame length increases. That is, when the calculated frame length is long, a small fluctuation by averaging is absorbed, variations of the audio content is global reflected. Further, the dispersion and standard deviation for each calculation frame length also, due to differences in fluctuations of the audio content, a different value.

  Next, the average value and standard deviation of the voice content will be described. The average value of the audio content rate is a value obtained by averaging the audio content rate Ris_n (k) in the entire program. In FIG. 2 described above, it is a value obtained by averaging the voice content ratios of Ris_1 (1) to Ris_1 (30). In other words, when expressed by the calculation frame length n (n is a positive number), the average value of the voice content rate is the average of the voice content rates from Ris_n (1) to Ris_n (N). The standard deviation is a value calculated using the voice content rate Ris_n (k) and the average value of the voice content rate. Here, based on the value of the voice content ratio Ris_n (k) shown in FIG. 2 to FIG. 4, the average value and the standard deviation of the voice content ratio are obtained for each calculation frame length. It becomes a value as shown. FIG. 5 is a diagram showing the calculation results of the average value and standard deviation of the audio content rate of each calculation frame length. In FIG. 5, the average value A1 of the voice content rate when the calculation frame length is 1 minute is 0.506, and the average value A5 of the voice content rate when the calculation frame length is 5 minutes is 0.498. The average value A10 of the voice content rate when the frame length is 10 minutes is 0.488. In FIG. 5, the standard deviation S1 with respect to the average value A1 is 0.161, the standard deviation S5 with respect to the average value A5 is 0.073, and the standard deviation S10 with respect to the average value A10 is 0.028.

Thus, as shown in FIG. 5, in the standard deviation, when the calculated frame length is short, the value of the standard deviation increases to fluctuations increases (increasing dispersion). If the calculated frame length is long, the value of the standard deviation is reduced to fluctuations decreases (variations is small). That is, the standard deviation is a value that is greatly affected by the length of the calculation frame length, and can generally be considered as a value that indicates a bias in the presence of an audio section in the entire program.

  Next, in the stage of converting the speed of the input audio signal and reproducing it, the speed ratio calculation unit 14 sets the audio information (the audio content rate, the average value of the audio content rate, and the standard) recorded in the audio information recording unit 13. Deviation) is used to calculate, for each frame, the speed ratio of the speech section corresponding to the bias of the presence of the speech section. Then, the speed ratio calculation unit 14 calculates the speed ratio of the non-voice section based on the speed ratio of the voice section and the desired playback time input by the user or the like. Then, the speed ratio calculation unit 14 sets a speed ratio for each frame with respect to the discrimination information discriminated by the voice / non-speech discrimination unit 11 and outputs it to the voice speed conversion unit 15. Here, it is assumed that the calculated speed ratio of the voice section of each frame is uniformly applied to the voice section existing in the frame. Further, the speed ratio of the non-speech section is assumed to be applied to the non-speech section in the frame at a constant speed ratio, as will be described later.

数式（２）より、目標再生時間比Ｒｔおよび音声含有率の平均値Ａ０が与えられ、音声区間の速度比ＳＲｓ０および非音声区間の速度比ＳＲｎｓ０のうち、いずれか一方が決まれば残りの他方が算出されることが分かる。 Here, before explaining the method of calculating the speed ratio, the relationship between the audio reproduction speed and the ease of listening will be explained. To listen to the audio signal in a shorter time than the normal playback time, the target playback time ratio Rt (0 <Rt <1) is given a set value of the reproduction time length to the normal playback time. For example, when the user tries to listen at half the playback time with respect to the normal playback time , the target playback time ratio Rt is Rt = 0.5. Such a target reproduction time ratio Rt is expressed by Equation (2). In Equation (2), the average value of the voice content rate is A0, the speed ratio of the voice interval when the voice content rate is constant, SRs0, and the speed ratio of the non-speech interval when the voice content rate is constant Is SRns0.

From Equation (2), et al given average A0 of target playback time ratio Rt and audio content are, among the speed ratio SRns0 speed ratio SRs0 and non-speech section in a speech period, the remaining other if Kimare either one It can be seen that is calculated.

The voice section speed ratio SRs0 shown in Equation (2) is generally easier to hear as the value is closer to 1.0, which is generally normal speed (same speed). As the value of the speed ratio SRs0 of the voice interval increases, the amount of information per unit time increases, so that it becomes difficult for the user to listen. Further, when the value of the speed ratio SRs0 of the voice section is about 2.0, it is difficult to understand the contents unless the user concentrates on listening. Thus, when the speed ratio SRs0 of the voice section is large, considerable difficulty occurs in listening for a long time. Therefore, the speed ratio SRs0 of the voice section is optimally set within a range in which the user can listen without depending on the target reproduction time ratio Rt to some extent. Based on this, a range in which the speed ratio SRs0 of the speech section is normally about 1 to 1.8 is used. Further, if the reproduction at a constant speed ratio without using speech / non-speech determination, practically often with 1.3 to 1.5 the velocity ratio.

In the present embodiment, it is considered that the standard deviation indicates the degree of bias of the existence of the voice section in the entire program as described above while considering the optimum setting range of the speed ratio SRs0 of the voice section . The speed ratio SRs0 of the voice interval is varied using the difference between the voice content rate and the average value of the voice content rate and the standard deviation. That is, the speed ratio SRs0 as a reference value, the audio content is changed to Do value smaller speed ratio of the speech segment than the reference value with respect to higher portion than the average value of the audio content is concentrated is the speech section, reverse audio content is changed to a large value and a speed ratio of the speech segment than the reference value with respect to the lower part than the average value of the audio content to.

Here, N is the number of frames of the entire program, Sn is the standard deviation when the calculation frame length is n minutes, and Ris_n (k) is the audio content rate at the kth frame when the calculation frame length is n minutes. SRs (k) is the speed ratio of the voice section in the k-th frame, An is the average voice content when the calculation frame length is n minutes, and Cn is a weighting factor that differs for each calculation frame length. The speed ratio of the non-voice section is SRns, and the speed ratio of the reference value when the voice content rate is assumed to be constant is SRs0. Here, the speed ratio SRns of the non-speech section is assumed to be a constant value without depending on the speech content rate of the frame. In this case, the speed ratio of the voice section in accordance with the deviation of the presence of speech segment SRs (k), for example is expressed as Equation (3).

数式（４）において、Ｃｎは、算出用フレーム長ごとに異なる重み係数であり、各算出用フレーム長の音声含有率の偏差を音声区間の速度比ＳＲｓ０に反映させる度合いを示すものである。 Furthermore, when the speed ratio SRs (k) of the voice section is calculated as a value reflecting both the global fluctuation and the short-term fluctuation of the voice content rate, a plurality of types of calculation frame lengths having different time lengths are used. It is calculated using the voice information. That is, the speed ratio of the voice section is calculated by using multiple types of calculation frame length voice information in a multiplexed manner. Here, assuming that M types of speech information having a calculation frame length are used, the speed ratio SRs (k) of the speech section of the kth frame is expressed by Equation (4).

In Equation (4), Cn is a different weighting factor for each calculation frame length, and indicates the degree to which the deviation of the voice content rate of each calculation frame length is reflected in the speed ratio SRs0 of the voice section.

ここで、数式（５）により音声情報を多重に用いた速度比の算出結果の一例を図６に示す。図６は、音声情報を多重に用いた速度比の算出結果の一例を示す図である。なお、図６に示す算出例は、数式（５）においてＳＲｓ０＝１．５、Ｃ１＝１、Ｃ５＝１０、Ｃ１０＝２０として算出し、短期的変動よりも長期的な変動に重点を置いた速度比を算出することを意図した例である。また、Ａ１、Ａ５、Ａ１０、Ｓ１、Ｓ５、Ｓ１０、Ｒｉｓ＿１（ｋ）、Ｒｉｓ＿５（ｋ）、およびＲｉｓ＿１０（ｋ）は、それぞれ図２〜図５に示した値である。また、図６では、数式（５）により音声情報を多重に用いた速度比の他に、数式（３）を用いて算出フレーム長（１分、５分、および１０分）に基づく音声情報から算出された各速度比を比較のために示している。 Here, when each voice information when the calculation frame length is 1 minute, 5 minutes, and 10 minutes is used as multiplexed voice information, the speed ratio SRs (k) of the voice section is expressed by Equation (5). .

Here, FIG. 6 shows an example of the calculation result of the speed ratio in which the voice information is multiplexed by the equation (5). FIG. 6 is a diagram illustrating an example of a calculation result of a speed ratio in which audio information is multiplexed. The calculation example shown in FIG. 6 is calculated as SRs0 = 1.5, C1 = 1, C5 = 10, and C10 = 20 in Equation (5), and focuses on long-term fluctuations rather than short-term fluctuations. This is an example intended to calculate the speed ratio. A1, A5, A10, S1, S5, S10, Ris_1 (k), Ris_5 (k), and Ris_10 (k) are the values shown in FIGS. Further, in FIG. 6, in addition to the speed ratio in which voice information is multiplexed by Formula (5), from the voice information based on the calculated frame length (1 minute, 5 minutes, and 10 minutes) using Formula (3). Each calculated speed ratio is shown for comparison.

  In FIG. 6, a graph drawn with rhombus plots indicates a speed ratio of a voice section calculated by using voice information in a multiplexed manner. Further, the graph drawn with a circle plot shows the speed ratio of the voice section calculated using only the voice information when the calculation frame length is 1 minute. The graph drawn by the square plot shows the speed ratio of the voice section calculated using only the voice information when the calculation frame length is 5 minutes. The graph drawn by the triangular plot shows the speed ratio of the voice section calculated using only the voice information when the calculation frame length is 10 minutes.

As shown in FIG. 6, the speed ratio of the voice section calculated using the voice information in a multiplexed manner is higher than the speed ratio calculated using only the voice information of the single calculation frame length. It can be seen that this value reflects both short-term fluctuations and long-term fluctuations. That is, the speed ratio of the audio section calculated by using the multiplexed audio information is a speed ratio according to the bias of the existence of the audio section throughout the program, and is a speed ratio that is considered to be a speed that can be easily heard. .

The speed ratio calculation unit 14 calculates the speed ratio SRns of the non-voice section so as to achieve the target playback time ratio Rt set from the input playback time after calculating the speed ratio SRs of the voice section by the method described above. . Note that the speed ratio SRns of the non-speech section is not variable and is a constant speed ratio as described above. This is based on the fact that most of the useful information is included in the speech section, and therefore, the adjustment of the speed ratio of the speech section is emphasized . Thereby, the audio reproducing apparatus according to the present embodiment can realize efficient reproduction. Hereinafter, a method for calculating the speed ratio SRns of the non-voice section will be described.

The target reproduction time ratio Rt is expressed as Expression (6) using the speed ratio SRs (k) of the voice section for each frame calculated based on Expression (4). Note that Ris (k) is the shortest calculation frame length for obtaining the audio content rate. Considering the above example, the shortest of the three types of calculation frame length is the one-minute calculation frame length.

なお、数式（７）からも分かるように、音声区間の速度比ＳＲｓ（ｋ）がフレーム毎に算出されるのに対して、非音声区間の速度比ＳＲｎｓは、フレームには依存せず（ｋには依存せず）一定速度比として算出される。ここで、非音声区間の速度比ＳＲｎｓの算出例を挙げる。例えば音声区間の速度比が１分、５分、１０分の多重な音声情報を用いて算出されるとする。また、数式（４）において、ＳＲｓ０を１．５と、重み係数をＣ１＝１、Ｃ５＝１０、Ｃ１０＝２０とする。このとき、図６に示したように、音声情報を多重に用いて算出された音声区間の速度比ＳＲｓ（ｋ）は１．２３〜１．６８の範囲の値となる。ここで、目標再生時間比Ｒｔを例えば０．５とする。このとき、非音声区間の速度比ＳＲｎｓは、数式（７）より、３．１７７となる。つまり、非音声区間の速度比ＳＲｎｓは、音声区間の速度比（例えば図６に示す１．２３〜１．６８）より高速の速度比に設定される。このように、速度比算出部１４は、音声情報記録部１３に記録された音声情報を用いて、音声含有率の変動に応じた音声区間の速度比をフレーム毎に算出し、非音声区間の速度比をフレームに関係なく一定の速度比で算出する。そして、算出された音声区間および非音声区間の速度比の情報は、音声速度変換部１５に出力される。 Therefore, the speed ratio SRns of the non-speech section is expressed by Expression (7) by rearranging Expression (6).

As can be seen from Equation (7), the speed ratio SRs (k) of the speech section is calculated for each frame, whereas the speed ratio SRns of the non-speech section does not depend on the frame (k It is calculated as a constant speed ratio. Here, a calculation example of the speed ratio SRns of the non-voice section will be given. For example, it is assumed that the speed ratio of the voice section is calculated using multiplexed voice information of 1 minute, 5 minutes, and 10 minutes. Further, in Equation (4), and 1.5 SRS0, the weighting coefficient is C1 = 1, C 5 = 10 , C 10 = 20. At this time, as shown in FIG. 6, the speed ratio SRs (k) of the voice section calculated by using the voice information in a multiplexed manner is a value in the range of 1.23 to 1.68. Here, the target reproduction time ratio Rt is set to 0.5, for example. At this time, the speed ratio SRns of the non-speech section is 3.177 from Equation (7). That is, the speed ratio SRns of the non-voice section is set to a speed ratio higher than the speed ratio of the voice section (for example, 1.23 to 1.68 shown in FIG. 6). As described above, the speed ratio calculation unit 14 calculates the speed ratio of the voice section corresponding to the fluctuation of the voice content rate for each frame using the voice information recorded in the voice information recording unit 13, and The speed ratio is calculated at a constant speed ratio regardless of the frame. Information on the speed ratio between the calculated speech segment and the non-speech segment is output to the speech speed conversion unit 15.

  The voice speed conversion unit 15 performs speed conversion processing on the voice signal recorded on the input recording medium or the like based on the information on the speed ratio of the voice section and the non-voice section calculated by the speed ratio calculation unit 14. I do. As a method of speed conversion processing, for example, there is a method of performing speed conversion by compressing and expanding an input audio signal on a time axis. However, the speed conversion process may be performed using other known methods without being limited to this method. In this way, the audio signal speed-converted by the audio speed conversion unit 15 of the present embodiment is converted at a speed ratio that varies dynamically according to the discrimination result of the audio / non-audio discrimination unit 11 and the audio content rate. It is an audio signal.

  Next, with reference to FIG. 7, the process flow of the audio reproduction device 1 according to the present embodiment will be described. FIG. 7 is a flowchart showing the flow of processing of the audio reproduction device 1 according to this embodiment. In FIG. 7, first, the user sets a target reproduction time with respect to the recording time of the entire program recorded on the recording medium, for example (step S1). Thereby, the target reproduction time ratio Rt (0 <Rt <1) is set. Next, the entire program recorded on the recording medium or the like is read, and the audio / non-audio discrimination unit 11 discriminates the audio section and non-audio section through the entire program before reproduction (step S2). Then, the voice information calculation unit 12 calculates the voice content ratio for each of a plurality of types of calculation frame lengths based on the information of the voice / non-voice section determined in step S2 (step S3). Next, the audio information calculation unit 12 calculates the average value and the standard deviation of the audio content rates using the audio content rates of the respective calculation frame lengths calculated in step S3 (step S4). Then, the audio information (audio content rate, average value and standard deviation of the audio content rate) calculated in steps S3 and S4 is recorded in the audio information recording unit 13 (step S5). This is the processing performed before reproduction. After the audio information is calculated throughout the entire program, playback for speed conversion is started. At the stage of reproduction, the speed ratio calculation unit 14 calculates the speed ratio of the voice section for each frame according to the bias of the existence of the voice section based on the voice information recorded in the voice information recording unit 13 (step S6). Next, the speed ratio calculation unit 14 calculates the speed ratio of the non-voice section based on the speed ratio of the voice section calculated in step S6 and the target playback time ratio Rt set in step S1 ( Step S7). Then, the speed ratio for each frame is set for the speech / non-speech section discrimination information discriminated by the speech / non-speech discriminator 11 and output to the audio speed converter 15. Subsequent to step S7, speed conversion processing is performed on the audio signal recorded on the input recording medium or the like based on the information on the speed ratio between the voice section and the non-voice section calculated in steps S6 and S7. (Step S8). Above, description about the flow of a process of the audio | voice reproduction apparatus 1 which concerns on this embodiment is complete | finished.

  As described above, according to the audio reproduction device according to the present embodiment, after calculating the audio content rate for the entire audio signal, the average value and the standard deviation of the audio content rate are calculated as statistical values and By calculating in advance the degree of bias of the presence of the voice section and calculating the speed ratio of the voice section using these voice information, the speed ratio of the voice section that varies dynamically according to the fluctuation of the voice content rate is calculated. be able to. That is, the audio reproduction device according to the present embodiment performs a process of reducing the speed ratio in a portion where the sound is concentrated and increasing the speed ratio in a portion where the sound is not concentrated. Thereby, according to the audio reproducing device according to the present embodiment, it is possible to maintain the intelligibility of the audio throughout the television program and the movie. Further, the speed ratio of the non-voice section is calculated as a constant speed ratio based on the speed ratio of the voice section so that a predetermined reproduction time is obtained. As a result, reproduction at an efficient reproduction speed becomes possible. Also, by calculating the statistical value such as the average value by multiplexing the audio information of each calculation frame length, it is more smooth and smoother for long-term fluctuations and short-term fluctuations in the voice content rate. Control of the speed ratio can be realized.

  Note that the speed ratio calculation unit 14 described above calculates the voice section speed ratio SRs (k) by multiplexing the audio information of each calculation frame length, but is not limited thereto. For example, the speed ratio SRs (k) of the voice section may be calculated using only a single calculation frame length. When the calculation is performed using the calculation frame length having a long time length, the calculated speed ratio of the voice section is a global value with respect to the changing voice content rate, and is a more accurate value. . When the calculation is performed using the calculation frame length having a short time length, the calculated speed ratio of the voice section is a value that has better followability with respect to the changing voice content rate.

  In the speed ratio calculation unit 14 described above, the voice content rate Ris_n (k), the average value An of the voice content rate, and the standard deviation Sn are used as the voice information for calculating the speed ratio of the voice section. It is not limited to this. For example, instead of the standard deviation, a statistical value equivalent to the standard deviation such as variance or average deviation may be used. In other words, the speech information for calculating the speed ratio of the speech section includes, in addition to the speech content rate Ris_n (k), a statistical value equivalent to the average value An and standard deviation of the speech content rate.

  Further, in the speed ratio calculation unit 14 described above, the speed ratio of the voice section is calculated for each frame, but for each voice section in the frame, it is further divided into sections such as a sentence head, a sentence, and a sentence end, The speed ratio may be varied in each section. For example, at the beginning of a certain voice section, the speed ratio is slightly reduced with respect to the speed ratio of the voice section calculated by the speed ratio calculation unit 14. And it sets so that a speed ratio may become large as the end of a sentence is reached. This makes it easier for the user to listen to sentence head portions that contain a lot of important information. As described above, the speed ratio calculation unit 14 may change the speed ratio for each section in one voice section.

  Note that the voice / non-voice discrimination unit 11, the voice information calculation unit 12, the speed ratio calculation unit 14, and the voice speed conversion unit 15 described in the first embodiment described above receive, for example, a voice signal and perform voice speed conversion. It can be realized by an information processing apparatus such as a general computer system that outputs the audio signal speed-converted by the unit 15. In this case, the present invention can be realized by storing a program for causing the computer to execute the above-described operation in a predetermined information recording medium, and reading and executing the program stored in the information recording medium. In this case, the playback time desired by the user is input using an input unit such as a keyboard connected to the information processing apparatus. The audio information calculated by the audio information calculation unit 12 is recorded on, for example, a hard disk in the information processing apparatus. The information recording medium for storing the program is, for example, a nonvolatile semiconductor memory such as a ROM or a flash memory, a CD-ROM, a DVD, or an optical disk-like recording medium similar to them. Further, the program may be supplied to the information processing apparatus through another medium or a communication line. Further, although the sound information calculated by the sound information calculation unit 12 is recorded on the hard disk in the information processing apparatus, it may be recorded on a memory in the information processing apparatus or another recording medium outside the information processing apparatus. .

(Second Embodiment)
With reference to FIG. 8, an audio reproducing apparatus according to the second embodiment of the present invention will be described. FIG. 8 is a block diagram showing the configuration of the audio reproduction device 2 according to the second embodiment of the present invention. In FIG. 8, the audio reproduction device 2 includes an input buffer 21, an audio / non-audio discrimination unit 11, an audio information sequential update unit 22, a speed ratio calculation unit 14, and an audio speed conversion unit 15.

Note that the audio reproduction device 2 according to the present embodiment has already recorded the entire audio signal of, for example, a TV program or a movie on a recording medium or the like, and a part of the entire recorded audio signal (for a predetermined time). This is an apparatus that assumes that audio information is sequentially calculated while temporarily storing the audio signal, and that the speed conversion is performed immediately in response to the input of the audio signal. Therefore, the audio reproducing apparatus 2 according to the present embodiment, the audio reproduction apparatus 1 according to the first embodiment described above, newly includes an input buffer 21, the audio information in the audio information sequentially updated section 2 2 It differs greatly in that it is updated sequentially. Hereinafter, different points will be mainly described. Further, since the voice / non-voice discrimination unit 11, the speed ratio calculation unit 14, and the voice speed conversion unit 15 are the same as those in the first embodiment described above, the same reference numerals are given and detailed description is omitted. To do.

  An audio signal recorded on a recording medium or the like is input to the input buffer 21. The input buffer 21 buffers input audio signals as appropriate. That is, in the input buffer 21, audio signal data for a predetermined time necessary for sequentially updating audio information by the audio information sequential updating unit 22 is temporarily recorded. The temporarily stored audio signals for a predetermined time are output to the audio / non-audio discriminating unit 11 and the audio speed converting unit 15, respectively. The voice / non-speech discrimination unit 11 discriminates a speech segment and a non-speech segment from the input audio signal for a predetermined time. Information on the voice / non-voice section determined by the voice / non-voice determination unit 11 is output to the voice information sequential update unit 22 and the speed ratio calculation unit 14, respectively.

  The voice information sequential update unit 22 sequentially updates the voice information based on the discrimination information of the voice / non-voice section. In the first embodiment, in Equations (3) and (4), the speech content rate Ris_n (k) is once calculated for the entire speech signal, and then the average value An and standard deviation of the speech content rate, which are statistical values, are calculated. Sn was calculated. On the other hand, in this embodiment, in order to perform playback with speed conversion immediately according to the input of the audio signal, the average value An of the voice content rate and the initial value of the standard deviation Sn, which are statistical values, are recorded in advance. (Not shown) or the like is recorded and set, and the statistical value is updated while being sequentially recorded in a recording unit or the like. Hereinafter, a method of updating the average value and the standard deviation of the voice content rate that is the voice information will be described.

The average value An of the voice content rate is set to an initial value upon activation . Then, the average value An audio content is Exiled next updated for each frame to which the audio signal is inputted. The initial value varies depending on, for example, the genre of the program to be played back, and is appropriately set according to the genre. For example, in the case of a television news program where the announcer frequently speaks, the average value of the audio content is about 85%. In addition, in the case of a documentary program that frequently uses various video scenes where there are few opportunities for speakers to speak, the average value of the audio content rate is about 50%.

なお、数式（８）において、α１およびβ１は音声含有率の更新平均値Ａｎ（ｋ）の更新速度を規定するパラメータである。すなわち、α１の値が大きいほどｋフレームの１つ前のフレームの更新平均値Ａｎ（ｋ−１）の占める割合が高くなり、更新平均値Ａｎ（ｋ）の更新速度が緩やかになる。また、β１の値が大きいほどｋフレームの音声含有率Ｒｉｓ＿ｎ（ｋ）の占める割合が高くなり、更新平均値Ａｎ（ｋ）の更新速度が速くなる。数値例としては、例えばα１＝０．９８、β１＝０．０２としてもよい。 Here, the predetermined time of the audio signal recorded in the input buffer 21 is set to the above-described calculation frame length (n minutes), for example. Then, it is assumed that the input buffer 21 sequentially records and updates the audio signal for one frame, for example, while securing the audio signal for the calculation frame length (n minutes). In addition, the audio information sequential update unit 22 performs the sequential update of the average value An of the audio content every time the audio / non-audio determination unit 11 determines an audio / non-audio section for one frame, for example. In this case, the average value An of the audio content rate is updated for each frame, and an update value of the average value of the audio content rate that is sequentially updated in the kth frame (hereinafter referred to as an updated average value of the audio content rate) is An ( k). At this time, the update average value An (k) of the voice content rate is expressed by Expression (8).

In Equation (8), α1 and β1 are parameters that define the update rate of the update average value An (k) of the voice content rate. That is, as the value of α1 is larger, the ratio of the update average value An (k−1) of the frame immediately before k frames becomes higher, and the update speed of the update average value An (k) becomes slower. In addition, the larger the value of β1, the higher the proportion of the audio content ratio Ris_n (k) of k frames, and the update speed of the update average value An (k) becomes faster. As numerical examples, for example, α1 = 0.98 and β1 = 0.02 may be used.

なお、数式（９）において、α２およびβ２は標準偏差の更新値Ｓｎ（ｋ）の更新速度を規定するパラメータである。数値例としては、例えばα２＝０．９８、β２＝０．０２としてもよい。 The standard deviation Sn is also set to an initial value at the time of activation , like the average value of the voice content rate. Then, the standard deviation Sn is Exiled following update on a frame-by-frame basis. Similar to the average value An of the audio content rate, the initial value varies depending on, for example, the genre of the program to be played back, and is appropriately set according to the genre. Specifically, the standard deviation Sn is updated using the initial value, the updated average value An (k), and the k frame audio content rate Ris_n (k). Here, if the update value of the standard deviation of the kth frame is Sn (k), the update value Sn (k) of the standard deviation is expressed by Equation (9).

In Equation (9), α2 and β2 are parameters that define the update rate of the standard deviation update value Sn (k). As numerical examples, for example, α2 = 0.98 and β2 = 0.02 may be used.

  Next, the speed ratio calculation unit 14 is based on the voice content rate Ris_n (k), the updated average value An (k) of the voice content rate updated for each frame, and the updated value Sn (k) of the standard deviation. Similarly to the above-described first embodiment, the speed ratio SRs (k) of the speech section is calculated based on the formulas (3) to (5). Further, the speed ratio calculation unit 14 calculates the speed ratio SRns of the non-voice section based on the calculated speed ratio SRs (k) of the voice section and the target reproduction time ratio Rt. Then, the speed ratio calculation unit 14 sets a speed ratio for each frame with respect to the discrimination information of the voice / non-speech section input from the voice / non-speech discrimination unit 11 and outputs it to the voice speed conversion unit 15. The voice speed conversion unit 15 sequentially performs speed conversion processing for each frame on the voice signal input from the input buffer 21 based on the speed ratio information of the voice section and the non-voice section calculated by the speed ratio calculation unit 14. I do.

  As described above, the audio reproduction device 2 according to the present embodiment sequentially updates the average value and standard deviation of the audio content rate, which are statistical values. Thereby, the audio reproduction device 2 according to the present embodiment can perform speed conversion processing immediately according to the input of the audio signal without calculating the audio information in advance for the entire program.

  Note that the audio reproduction device 2 described in the second embodiment described above includes the audio / non-audio discriminating unit 11, the audio information sequential updating unit 22, the speed ratio calculating unit 14, and the audio speed converting unit 15, for example, an audio signal. And an information processing apparatus such as a general computer system that outputs a voice signal whose speed has been converted by the voice speed conversion unit 15. In this case, the present invention can be realized by storing a program for causing the computer to execute the above-described operation in a predetermined information recording medium, and reading and executing the program stored in the information recording medium. In addition, a playback time desired by the user and the initial value described above are input in an input unit such as a keyboard connected to the information processing apparatus. Moreover, the input buffer 21 is comprised in the hard disk in information processing apparatus, for example. The information recording medium for storing the program is, for example, a nonvolatile semiconductor memory such as a ROM or a flash memory, a CD-ROM, a DVD, or an optical disk-like recording medium similar to them. Further, the program may be supplied to the information processing apparatus through another medium or a communication line. Further, although the input buffer 21 is configured in, for example, a hard disk in the information processing apparatus, it may be configured in a memory in the information processing apparatus or another recording medium outside the information processing apparatus.

(Third embodiment)
With reference to FIG. 9, a voice recording / reproducing apparatus according to a third embodiment of the present invention will be described. FIG. 9 is a block diagram showing a configuration of a voice recording / reproducing apparatus 3 according to the third embodiment of the present invention. In FIG. 9, the voice recording / reproducing apparatus 3 includes a voice / non-voice discrimination unit 11, an information recording unit 31, a voice information calculation unit 12, a voice information recording unit 13, a speed ratio calculation unit 14, and a voice speed conversion unit 15. .

  The audio recording / reproducing apparatus 3 according to the present embodiment is an audio recording / reproducing apparatus that records and reproduces audio in the information recording unit 31, and simultaneously records an input audio signal in the information recording unit 31. The information recording unit 31 also records the information of the voice section and the non-voice section discriminated by the non-voice discriminating unit 11. Hereinafter, this feature will be mainly described. Further, the voice / non-voice discrimination unit 11, the voice information calculation unit 12, the voice information recording unit 13, the speed ratio calculation unit 14, and the voice speed conversion unit 15 are the same as those in the first embodiment described above, and thus the same. The detailed description is abbreviate | omitted.

  An audio signal to be recorded is input to the audio / non-audio discriminating unit 11 and the information recording unit 31, respectively. The voice / non-voice discrimination unit 11 discriminates a voice section and a non-voice section from the input voice signal. The discrimination information of the voice / non-voice section discriminated by the voice / non-voice discrimination unit 11 is output to the information recording unit 31. The information recording unit 31 records the input audio signal to be recorded and the discrimination information of the voice / non-voice section.

  The voice information calculation unit 12 reads out the voice / non-speech section information about the entire voice signal recorded in the information recording unit 31 and calculates the voice information. Specifically, the audio information calculation unit 12 calculates the audio content rate through the entire recorded audio signal, and then calculates the average value and the standard deviation of the audio content rate. Then, the voice content rate calculated by the voice information calculation unit 12, the average value of the voice content rate, and the standard deviation are recorded in the voice information recording unit 13.

  Then, at the stage of reproduction, the speed ratio calculation unit 14 uses the audio information recorded in the audio information recording unit 13 to calculate the speed ratio of the audio section corresponding to the change in the audio content rate for each frame. Further, the speed ratio calculation unit 14 calculates the speed ratio of the non-voice section based on the speed ratio of the voice section and the target reproduction time ratio Rt. Then, a speed ratio for each frame is set for the recorded voice / non-voice section discrimination information and output to the voice speed converter 15. The voice speed conversion unit 15 performs speed conversion processing on the voice signal recorded in the information recording unit 31 based on the speed ratio information of the voice section and the non-voice section calculated by the speed ratio calculation unit 14.

  As described above, the audio recording / reproducing apparatus 3 according to the present embodiment records the input audio signal in the information recording unit 31 and also determines the audio interval and non-audio interval determined by the audio / non-audio determination unit 11. Information is also recorded in the information recording unit 31. Thereby, according to the audio recording / reproducing apparatus 3 according to the present embodiment, since the determination of the audio section and the non-audio section for the entire audio signal is completed at the stage of recording the entire audio signal, it is performed before the reproduction. The calculation time of voice information can be shortened.

Incidentally, the information recording section 31 described above, the audio / addition to the non-speech determination unit 11 determine specific information of speech section or a non-voice section is determined by, recorded more audio information calculated by the sound information calculation section 12 Also good. In this case, as shown in FIG. 10, the audio information recording unit 13 is omitted. FIG. 10 is a block diagram showing a configuration of the voice recording / reproducing apparatus 4 that records information of voice sections and non-voice sections and voice information in the information recording unit 31. In FIG. 10, the voice recording / playback apparatus 4 includes a voice / non-voice discrimination unit 11, an information recording unit 31, a voice information calculation unit 12, a speed ratio calculation unit 14, and a voice speed conversion unit 15.

  In FIG. 10, in the information recording unit 31, the input audio signal to be recorded, the information of the voice / non-voice section determined by the voice / non-voice determination unit 11, and the voice information calculation unit 12 are calculated. Audio information is recorded. That is, the voice recording / reproducing apparatus 4 records the voice / non-voice section discrimination information and the voice information in the information recording unit 31 together with the recording. Thereby, according to the audio recording / reproducing apparatus 4, if the reproduction time is input after recording, the speed ratio can be calculated immediately. As a result, the voice recording / playback apparatus 4 can output the playback voice whose speed has been converted in a short time.

  The voice / non-voice discrimination unit 11, the voice information calculation unit 12, the voice information recording unit 13, the speed ratio calculation unit 14, and the voice speed conversion unit 15 described in the third embodiment described above, for example, It can be realized by an information processing apparatus such as a general computer system that inputs and outputs a voice signal speed-converted by the voice speed converter 15. In this case, the present invention can be realized by storing a program for causing the computer to execute the above-described operation in a predetermined information recording medium, and reading and executing the program stored in the information recording medium. In addition, a playback time desired by the user is input in an input unit such as a keyboard connected to the information processing apparatus. The information recording unit 31 and the audio information recording unit 13 are configured in, for example, a hard disk in the information processing apparatus. The information recording medium for storing the program is, for example, a nonvolatile semiconductor memory such as a ROM or a flash memory, a CD-ROM, a DVD, or an optical disk-like recording medium similar to them. Further, the program may be supplied to the information processing apparatus through another medium or a communication line. In addition, although the information recording unit 31 and the audio information recording unit 13 are configured in, for example, a hard disk in the information processing apparatus, they may be configured in a memory in the information processing apparatus or another recording medium outside the information processing apparatus. Good.

  Further, the voice / non-voice discrimination unit 11, the voice information calculation unit 12, the voice information recording unit 13, the speed ratio calculation unit 14, the voice information successive update unit 22, and the voice speed described in the first to third embodiments described above. The conversion unit 15 can also be realized by an integrated circuit that receives, for example, an audio signal, reproduction time information, the above-described initial value, and the like, and outputs the audio signal speed-converted by the audio speed conversion unit 15. In this case, the audio information recording unit 13 in the first embodiment, the input buffer 21 in the second embodiment, the audio information recording unit 13 and the information recording unit 31 in the third embodiment are, for example, memories in an integrated circuit. Composed. The present invention can be realized by integrating an electric circuit that performs the above-described functions in one small package and configuring an audio signal processing circuit DSP (Digital Signal Processor) that performs audio signal processing or the like. Become. The audio information recording unit 13 in the first embodiment, the input buffer 21 in the second embodiment, the audio information recording unit 13 and the information recording unit 31 in the third embodiment are different from the integrated circuit described above. The recording medium may be configured as follows.

  The audio reproducing device, audio recording / reproducing device, and method, recording medium, and integrated circuit according to the present invention perform optimum speed ratio control in accordance with fluctuations in the audio content, thereby realizing reproduction that is easier to listen to. It is also useful for DVD players, HDD players, CD players and the like.

The block diagram which shows the structure of the audio | voice reproduction apparatus 1 which concerns on 1st Embodiment in this invention. The figure which shows the example of calculation of audio | voice content rate Ris_1 (k) when the frame length for calculation is 1 minute The figure which shows the example of calculation of audio | voice content rate Ris_5 (k) when the frame length for calculation is 5 minutes The figure which shows the example of calculation of audio | voice content rate Ris_10 (k) when the frame length for calculation is 10 minutes The figure which shows the calculation result of the average value of the audio | voice content rate of each calculation frame length, and a standard deviation The figure which shows an example of the calculation result of the speed ratio using multiple audio | voice information The flowchart which shows the flow of a process of the audio | voice reproduction apparatus 1 which concerns on this embodiment. The block diagram which shows the structure of the audio | voice reproduction apparatus 2 which concerns on 2nd Embodiment in this invention. The block diagram which shows the structure of the audio | voice recording / reproducing apparatus 3 which concerns on 3rd Embodiment in this invention. The block diagram which shows the structure of the audio | voice recording / reproducing apparatus 4 which records the information and audio | voice information of an audio | voice area and a non-audio | voice area in the information recording part 31. FIG. The block diagram which shows the structure of the conventional audio | voice reproduction apparatus 9.

Explanation of symbols

1, 2 Audio playback device 3, 4 Audio recording / playback device 11 Audio / non-audio discrimination unit 12 Audio information calculation unit 13 Audio information recording unit 14 Speed ratio calculation unit 15 Audio speed conversion unit 21 Input buffer 22 Audio information successive update unit 31 Information recording section

Claims (8)

An audio reproduction device for applying an audio speed conversion process to an input audio signal to shorten the reproduction time than usual and listening to the audio signal,
A discriminator for discriminating a voice section including voice and a non-voice section not containing voice with respect to the voice signal;
Based on the speech segment and the non-speech section, and the audio content indicating a ratio that contains the speech segment against calculated frame length, the mean value and standard deviation of the speech content of the calculation frame length statistics A voice information calculation unit for calculating at least a value ;
A speed ratio with a ratio of the voice speed conversion processing of 1 or more is preset as a reference value, and the speed ratio of the voice section is calculated from the reference value according to the statistical value and the voice content rate of the calculation frame length. A speed ratio calculation unit
The voice information calculation unit sets a plurality of calculation frame lengths each having a different time length, and calculates the voice content rate and the statistical value, respectively.
The speed ratio calculation unit calculates a coefficient calculated using a difference value with respect to the average value of the voice content rate in the calculation frame length and a value obtained for each calculation frame length by the standard deviation of the speed ratio. By multiplying a reference value, the speed ratio of the voice section in the calculation frame length is changed to a value smaller than the reference value when the voice content rate of the calculation frame length is relatively high, and the calculation frame When the speech content rate of the length is relatively low, the speed ratio of the speech section in the calculation frame length is changed to a value larger than the reference value, and adaptively with respect to the temporal variation of the speech content rate An audio reproducing apparatus that performs speed control for adjusting a speed ratio of an audio section.   The speed ratio calculation unit sets the shortened playback time according to a user operation, and the playback time of the audio signal becomes the set playback time based on the changed speed ratio in a voice section. The sound reproducing apparatus according to claim 1, wherein the speed ratio of the non-voice section is calculated as described above.   The sound reproduction apparatus according to claim 2, wherein the speed ratio calculation unit calculates the speed ratio of the non-speech section constant within the set reproduction time. Of the audio signals before Symbol input, a buffer for recording while sequentially updating the audio signals to include the speech signals of at least the calculated frame length fraction,
A speed conversion section for outputting performs voice speed conversion processing on the audio signals recorded before Symbol buffer, further comprising,
The determination unit determines the voice section and the non-voice section with respect to the calculation-frame-length voice signal recorded in the buffer,
The audio information calculation unit sequentially updates the statistics calculated for Until now every unit time,
The speed ratio calculation unit calculates the speed ratio of the voice section according to the statistical value updated every unit time and the voice content rate set in the calculation frame length at the time of the update,
The speed conversion unit sequentially performs a sound speed conversion process on a sound signal sequentially updated in the buffer using a speed ratio of the sound section calculated for each unit time. Item 2. The audio playback device according to Item 1. An audio reproduction method for listening to the audio signal by applying an audio speed conversion process to the input audio signal to shorten the reproduction time than usual,
A determination step of determining a speech section including speech and a non-speech section not including speech for the speech signal;
Based on the speech segment and the non-speech section, and the audio content indicating a ratio that contains the speech segment against calculated frame length, the mean value and standard deviation of the speech content of the calculation frame length statistics A voice information calculation step for calculating at least a value ;
A speed ratio with a ratio of the voice speed conversion processing of 1 or more is preset as a reference value, and the speed ratio of the voice section is calculated from the reference value according to the statistical value and the voice content rate of the calculation frame length. A speed ratio calculating step,
The voice information calculation step sets a plurality of calculation frame lengths each having a different time length, and calculates the voice content rate and the statistical value, respectively.
The speed ratio calculating step calculates a coefficient calculated using a difference value with respect to the average value of the audio content rate in the calculation frame length and a value obtained for each calculation frame length by the standard deviation. By multiplying a reference value, the speed ratio of the voice section in the calculation frame length is changed to a value smaller than the reference value when the voice content rate of the calculation frame length is relatively high, and the calculation frame When the speech content rate of the length is relatively low, the speed ratio of the speech section in the calculation frame length is changed to a value larger than the reference value , and adaptively with respect to the temporal variation of the speech content rate A sound reproduction method for performing speed control for adjusting a speed ratio of a voice section . A computer-readable recording medium that records a sound reproduction program executed by a computer for applying a sound speed conversion process to an input sound signal to shorten the reproduction time and listening to the sound signal. There,
In the computer,
A determination step of determining a speech section including speech and a non-speech section not including speech for the speech signal;
Based on the speech segment and the non-speech section, and the audio content indicating a ratio that contains the speech segment against calculated frame length, the mean value and standard deviation of the speech content of the calculation frame length statistics A voice information calculation step for calculating at least a value ;
A speed ratio with a ratio of the voice speed conversion processing of 1 or more is preset as a reference value, and the speed ratio of the voice section is calculated from the reference value according to the statistical value and the voice content rate of the calculation frame length. A speed ratio calculating step to
The voice information calculation step sets a plurality of calculation frame lengths each having a different time length, and calculates the voice content rate and the statistical value, respectively.
The speed ratio calculating step calculates a coefficient calculated using a difference value with respect to the average value of the audio content rate in the calculation frame length and a value obtained for each calculation frame length by the standard deviation. By multiplying a reference value, the speed ratio of the voice section in the calculation frame length is changed to a value smaller than the reference value when the voice content rate of the calculation frame length is relatively high, and the calculation frame When the speech content rate of the length is relatively low, the speed ratio of the speech section in the calculation frame length is changed to a value larger than the reference value , and adaptively with respect to the temporal variation of the speech content rate A computer-readable recording medium on which a program for speed control for adjusting a speed ratio of a voice section is recorded. An integrated circuit for listening to the audio signal by applying an audio speed conversion process to the input audio signal and reducing the playback time from normal,
A discriminator for discriminating a voice section including voice and a non-voice section not containing voice with respect to the voice signal;
Based on the speech segment and the non-speech section, and the audio content indicating a ratio that contains the speech segment against calculated frame length, the mean value and standard deviation of the speech content of the calculation frame length statistics A voice information calculation unit for calculating at least a value ;
A speed ratio with a ratio of the voice speed conversion processing of 1 or more is preset as a reference value, and the speed ratio of the voice section is calculated from the reference value according to the statistical value and the voice content rate of the calculation frame length. A speed ratio calculation unit
The voice information calculation unit sets a plurality of calculation frame lengths each having a different time length, and calculates the voice content rate and the statistical value, respectively.
The speed ratio calculation unit calculates a coefficient calculated using a difference value with respect to the average value of the voice content rate in the calculation frame length and a value obtained for each calculation frame length by the standard deviation of the speed ratio. By multiplying a reference value, the speed ratio of the voice section in the calculation frame length is changed to a value smaller than the reference value when the voice content rate of the calculation frame length is relatively high, and the calculation frame When the speech content rate of the length is relatively low, the speed ratio of the speech section in the calculation frame length is changed to a value larger than the reference value , and adaptively with respect to the temporal variation of the speech content rate An integrated circuit that performs speed control to adjust the speed ratio of the voice interval . A voice recording / playback apparatus for applying a voice speed conversion process to an input voice signal to shorten the playback time and listening to the voice signal.
An information recording unit for recording the input audio signal;
A discriminating unit for discriminating a voice section including voice and a non-voice section not containing voice with respect to the voice signal before being recorded in the information recording unit;
Based on the speech segment and the non-speech section, and the audio content indicating a ratio that contains the speech segment against calculated frame length, the mean value and standard deviation of the speech content of the calculation frame length statistics A voice information calculation unit for calculating at least a value ;
A speed ratio with a ratio of the voice speed conversion processing of 1 or more is preset as a reference value, and the speed ratio of the voice section is calculated from the reference value according to the statistical value and the voice content rate of the calculation frame length. A speed ratio calculation unit
The voice information calculation unit sets a plurality of calculation frame lengths each having a different time length, and calculates the voice content rate and the statistical value, respectively.
The speed ratio calculation unit calculates a coefficient calculated using a difference value with respect to the average value of the voice content rate in the calculation frame length and a value obtained for each calculation frame length by the standard deviation of the speed ratio. By multiplying a reference value, the speed ratio of the voice section in the calculation frame length is changed to a value smaller than the reference value when the voice content rate of the calculation frame length is relatively high, and the calculation frame When the speech content rate of the length is relatively low, the speed ratio of the speech section in the calculation frame length is changed to a value larger than the reference value , and adaptively with respect to the temporal variation of the speech content rate A voice recording / playback device that performs speed control to adjust the speed ratio of voice sections .

Images