JP2011090483A - Information processing apparatus and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an information processing apparatus capable of properly controlling the range retroacting from the point of time corresponding to the input point of time of voice on output time series of voice data. <P>SOLUTION: The information processing apparatus for reproducing the voice data includes: an input unit which accepts voice given by a user; a calculation unit which calculates the speed of sound uttered by the user; and a control unit which determines the range retroacting from the point of time corresponding to the input point of time of the voice on the output time series of the voice data according to the speed of sound uttered by the user. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to an information processing apparatus that reproduces audio data.

  Even if the voice is generated, it disappears immediately, so that the person can only remember the voice information within the storage capacity of the person. Therefore, when there is information that the user wants to hear again due to missed listening or the like during the reproduction of the audio data, the person usually searches for a part including the information to be reheard by rewinding the audio data.

  A technique in which an acquisition range in which audio data is acquired retrospectively by a predetermined method is determined based on an instruction from an instruction unit that indicates a reconfirmation point of audio data, and words in the acquisition range are extracted and displayed (For example, Patent Document 1).

  In addition, there is a technique of detecting a break in the sound retroactively when an instruction to listen back is given and reproducing the sound again from the detection point (for example, Patent Documents 2 and 3).

JP 2007-257341 A Japanese Patent Laid-Open No. 62-40577 JP 2000-267687 A

  However, in the technique described in Patent Document 1, the acquisition range of audio data is not changed from the initial setting. For this reason, Patent Document 1 has a problem that information to be reconfirmed is not necessarily included in the designated acquisition range. Further, when the initial acquisition range is too large, there is a problem that voice data is acquired far back from the position of the information to be reconfirmed, and the efficiency of the word extraction process is lowered.

  Also, in the techniques described in Patent Documents 2 and 3, playback is performed retroactively to the break of a phrase (sound break), but it is not possible to control whether or not a specific word is included. There was a problem that it was not always possible to hear back.

  An object of one embodiment of the present invention is to provide an information processing apparatus capable of appropriately controlling a range that goes back from a time point corresponding to a time point of input of sound on an output time series of sound data.

One aspect of the present invention is an information processing apparatus. This information processing device
An information processing apparatus for reproducing audio data,
An input unit for receiving a voice uttered by the user;
A calculation unit for calculating an utterance speed of the voice;
A control unit that determines a range that goes back from the time corresponding to the input time of the sound on the output time series of the sound data, according to the speech rate;
Is provided.

Another aspect of the present invention is a method for determining a range that goes back from the voice input time point described above. In addition, another aspect of the present invention can include a program that causes the information processing device to function as a determination device for a range that goes back from the voice input time point, and a computer-readable recording medium that records the program.

  According to the information processing apparatus of the disclosure, it is possible to appropriately control a range that goes back from the time corresponding to the input time of the sound on the output time series of the sound data.

It is a figure which shows the hardware structural example of information processing apparatus. It is explanatory drawing of the function implement | achieved by running the program for determining the range which the processor of information processing apparatus goes back. It is a figure which shows the structural example of a calculation part. It is a figure which shows an example of the relationship between an utterance speed and the range to go back. It is a figure which shows an example of the relationship between an utterance speed and the range to go back. It is a figure which shows an example of the relationship between an utterance speed and the range to go back. It is a figure which shows an example of the relationship between an utterance speed and the range to go back. It is a figure which shows an example of the relationship between an utterance speed and the range to go back. It is a figure which shows an example of the relationship between an utterance speed and the range to go back. It is a figure which shows an example of the relationship between an utterance speed and the range to go back. It is a figure which shows an example of the relationship between an utterance speed and the range to go back. It is a figure which shows the example of the processing flow of information processing apparatus. It is explanatory drawing of the function implement | achieved when the processor of information processing apparatus runs an audio | voice data reproduction | regeneration program. An example in which the search unit searches for a keyword using the word spotting technique is shown. It is a figure which shows an example of the reset process of a search range. It is a figure which shows an example of the reset process of a search range. It is a figure which shows an example of the reset process of a search range. It is a figure showing an example of processing in case a plurality of keywords are detected in partial voice data. It is a figure showing an example of processing in case a plurality of keywords are detected in partial voice data. It is a figure showing an example of processing in case a plurality of keywords are detected in partial voice data. It is a figure showing an example of processing in case a plurality of keywords are detected in partial voice data. It is a figure which shows an example of the process which determines the head position when reproducing audio | voice data from a memory | storage part. It is a figure which shows an example of the process which determines the head position when reproducing audio | voice data from a memory | storage part. It is a figure which shows the example of the processing flow of information processing apparatus.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. The configuration of the following embodiment is an exemplification, and the present invention is not limited to the configuration of the embodiment.

<Example of hardware configuration of information processing apparatus>
FIG. 1 is a diagram illustrating a hardware configuration example of the information processing apparatus. The information processing apparatus 1 includes a processor 101, a main storage device 102, a microphone 103, an output device 104, an auxiliary storage device 105, a network interface 107, and a tuner 108. They are connected to each other by a bus 109.

  The microphone 103 collects sound emitted by the user. The microphone 103 outputs an electrical signal corresponding to the collected sound to the processor 101. Hereinafter, the electrical signal corresponding to the voice is referred to as “voice signal”.

The network interface 107 is an interface for inputting / outputting information to / from the network. The network interface 107 is connected to a wired network and a wireless network. The network interface 107 is, for example, a NIC (Network Interface Card) or a wireless LAN (Local Area).
Network) card. The network interface 107 receives audio signals such as Internet radio and Internet TV from a connected network. The audio signal received by the network interface 107 is output to the processor 101.

  The tuner 108 selects a reception frequency by selecting a reception frequency, and receives broadcast radio waves such as radio and television. The tuner 108 outputs the received broadcast radio sound signal to the processor 101.

  The main storage device 102 provides the processor 101 with a storage area and a work area for loading a program stored in the auxiliary storage device 105, and is used as a buffer. The main storage device 102 is, for example, a semiconductor memory such as a RAM (Random Access Memory).

  The auxiliary storage device 105 stores various programs and data used by the processor 101 when executing each program. The auxiliary storage device 105 is, for example, an EPROM (Erasable Programmable ROM) or a hard disk drive (Haard Disc Drive). The auxiliary storage device 105 can include a removable medium, that is, a portable recording medium. The removable medium is a recording medium such as USB (Universal Serial Bus), flash memory, CD (Compact Disc), or DVD, for example. The auxiliary storage device 105 holds, for example, an operating system (OS), a program for determining a range that goes back from the user's voice input point, a voice data reproduction program, and various other application programs.

  The processor 101 is, for example, a CPU (Central Processing Unit) or a DSP (Digital Signal Processor). The processor 101 loads the OS and various application programs held in the auxiliary storage device 105 into the main storage device 102 and executes them, thereby executing various processes related to voice.

  For example, the processor 101 performs digital conversion processing on an audio signal input from the microphone 103 by executing a program to obtain audio data. The audio data is stored in the main storage device 102 and / or the auxiliary storage device 105.

  Further, the processor 101 generates audio data from audio signals received by the network interface 107 and the tuner 108 by executing the program, and records the audio data in the main storage device 102 and / or the auxiliary storage device 105.

  The audio data generated from the audio signal that is received by the network interface 107 and the tuner 108 and reproduced by real-time processing is accumulated in a buffer in the main storage device 102 for a predetermined time. The processor 101 sequentially reads out audio data stored in the buffer of the main storage device 102, performs a decoding process, decodes the audio data, and outputs the audio signal to the output device 104.

  In addition, the processor 101 executes the program for determining the range that goes back from the user's voice input time point, so that the voice of the voice data input from the network interface 107, the tuner 108, etc. Processing to determine the range that goes back from the point of input. Details of the processing will be described later.

  In addition, when the audio data is input during the reproduction of the audio data sequentially input from the network interface 107, the tuner 108, and the like by the execution of the audio data reproduction program, the processor 101 points back a predetermined range from the audio input time point. Process to play from. Details of the processing will be described later.

  The output device 104 outputs the processing result of the processor 101. The output device 104 includes a display, a speaker, a speaker interface circuit, and the like.

  The information processing apparatus 1 is a general-purpose computer such as a personal computer, for example. Further, the information processing apparatus 1 reproduces a moving image including audio data and audio data such as a mobile phone, a car navigation system, a receiver of a one-segment partial reception service (generally called “one-segment broadcasting”), a radio, It is a device that provides audio information. Also, an IC chip or the like stored in them.

<First Embodiment>
In the information processing apparatus according to the first embodiment, when a voice is input from the user during playback of the voice data, the input of the voice of the user triggers a predetermined range from the user's voice input time. This is an information processing apparatus that reproduces sound from the point in time. For example, when the user misses the sound being reproduced, the user utters some sound, so that the information processing apparatus reproduces the sound data from a time point that is a predetermined range backward from the time point when the user's sound is input.

  FIG. 2 is an explanatory diagram of functions realized by the processor 101 of the information processing apparatus 1 executing a program for determining a range that goes back from the user's voice input time point. The information processing apparatus 1 implements the input unit 11, the calculation unit 12, the control unit 13, and the extraction unit 14 by executing a program for the processor 101 to determine a range that goes back from the user's voice input time point. be able to. That is, the information processing apparatus 1 is an apparatus including the input unit 11, the calculation unit 12, the control unit 13, and the extraction unit 14 by executing a determination program for determining a range that goes back from the user's voice input time point. Function.

  The input unit 11 includes a microphone 103, and obtains an audio signal of a voice uttered by a user input to the microphone 103 as an input. The input unit 11 performs digital conversion processing on the audio signal and converts the audio signal into audio data. The input unit 11 gives the time when the sound data is input to the sound data as a time stamp. The time stamp is assigned with any one of an elapsed time since the activation of the information processing apparatus 1 based on a clock (not shown) provided in the information processing apparatus 1, a time managed by the information processing apparatus 1, and the like. The input unit 11 outputs voice data of a voice uttered by the user to the calculation unit 12 and the extraction unit 14. Hereinafter, the voice data of the voice uttered by the user is referred to as input voice data.

Input voice data is input from the input unit 11 to the calculation unit 12. The calculation unit 12 calculates the utterance speed of the user from the input voice data.

  FIG. 3 is a diagram illustrating a configuration example of the calculation unit 12. The calculation unit 12 includes a section detection unit 121, a voice recognition unit 122, a mora number calculation unit 123, and an utterance speed calculation unit 124.

  The section detection unit 121 obtains input voice data from the input unit 11 as an input. The section detection unit 121 measures the time from the beginning to the end of the input voice data from the time stamp given to the input voice data. The section detection unit 121 outputs the measurement time to the utterance speed calculation unit 124 as the voice section length of the input voice data. The section detection unit 121 outputs the input voice data to the voice recognition unit 122.

  The voice recognition unit 122 obtains input voice data from the section detection unit 121 as an input. The voice recognition unit 122 performs voice recognition processing based on the input voice data. For example, the voice recognition unit 122 obtains as a recognition result that the content of the input voice data is “cat (cat)” by voice recognition processing. As the voice recognition processing performed by the voice recognition unit 122, any existing voice recognition technique can be applied. The voice recognition unit 122 outputs the result of the voice recognition process to the mora number calculation unit 123.

  The mora number calculation unit 123 obtains the speech recognition result of the input speech data from the speech recognition unit 122 as an input. The mora number calculation unit 123 calculates the mora number from the result of speech recognition.

  Mora is a segmental unit of sound having a certain length of time in phonological theory. In Japanese language studies, Mora is generally called “beat”. For example, the number of mora of the word “cat (cat)” is 2 mora for “ne” and “ko”. For example, the number of mora for the word “kappa” is 3 mora for “ka”, “tsu”, and “pa”. For example, the number of mora of the word “chocolate” is 5 mora for “cho”, “co”, “le”, “-” and “to”. The mora number calculation unit 123 outputs the mora number of the input voice to the utterance speed calculation unit 124.

  The utterance speed calculation unit 124 obtains as input the voice section length and the number of mora of the input voice data. The utterance speed calculation unit 124 calculates the utterance speed of the user at the time of voice input from the voice section length of the input voice data and the number of mora. The utterance speed calculation unit 124 calculates the utterance speed as, for example, utterance speed = number of mora / speech interval length. For example, when the voice section length of the input voice data is 0.5 seconds and the number of mora of the input voice data is 4, the speech speed is: speech speed = 4 mora / 0.5 seconds = 8 mora / second. Calculated. The utterance speed calculation unit 124 outputs the calculated utterance speed to the control unit 13.

  The control unit 13 obtains the speech rate as an input. From the time corresponding to the input time of the user's voice (input voice) on the output time series of the voice data (hereinafter also referred to as playback voice data or playback voice data) based on the speech rate. Determine the range to go back.

  4A, 4B, 4C, 4D, 4E, 4F, 4G, and 4H are diagrams showing examples of the relationship between the speech rate and the range that goes back from the input time point of the user's voice. A correspondence table between one or more of the speech speeds shown in FIGS. 4A to 4H and the retroactive range is stored in the auxiliary storage device 105. When the speech rate of the input voice data is input, the control unit 13 determines a range that goes back, for example, based on at least one of the correspondence tables shown in FIGS. 4A to 4H. The retroactive range may be defined by any of the time length, the number of syllables, the number of mora, and the like. In 1st Embodiment, the control part 13 demonstrates the case where the retroactive range is defined with time.

Humans have the property of being quick when speaking, that is, speaking speed increases. If the user hears the voice being played back and utters the voice related to the information that the user missed in a quick manner, the user may be in a state of being in a hurry or impatient. Therefore, when the utterance speed of the user's voice data (input voice data) input to the information processing apparatus 1 is high, the information that the user wants to listen to is within a range retroactive from the user's voice input time point. There is a high possibility that it exists in. On the other hand, if a person is lost or thinking, he / she utters slowly, that is, the speaking rate decreases. When the user utters slowly, there is a high possibility that there is information that the user wants to listen to again at a location that goes back from the time when the user's voice was input.

  The example shown in FIG. 4A is an example of a correspondence table between the utterance speed and the retroactive range associated with each other so that the retroactive range becomes smaller as the utterance speed increases in view of the human nature. FIG. 4A shows an example in which the relationship between the speech rate and the retroactive range is linear.

  For example, the speech rate of a human is 15 mora / sec for a very fast mouth and 8 mora / sec for a normal case. That is, since there is a limit to the number of mora that a person emits per second, a minimum value and a maximum value can be set for the speech rate. For example, in FIG. 4A, the minimum value of the speech rate is set to 0 mora / second, and the maximum value is set to 15 mora / second. In FIG. 4A, the range that goes back when the speech rate is the minimum value is set to the maximum value, and the range that goes back when the speech rate is the maximum value is set to the minimum value. In FIG. 4A, the maximum value of the retroactive range corresponding to the minimum value 0 mora / second of the speech rate is set to 15 seconds. In FIG. 4A, the retroactive range corresponding to the maximum value of the speech rate of 15 mora / second is set to the minimum value of 5 seconds.

  Similar to FIG. 4A, the correspondence tables shown in FIG. 4B, FIG. 4C, and FIG. 4D also associate the utterance speed with the retroactive range so that the retroactive range decreases as the utterance speed increases. Similarly to FIG. 4A, the correspondence table examples shown in FIG. 4B, FIG. 4C, and FIG. 4D also have the minimum and maximum utterance speeds set, and the utterance speed becomes the minimum value. When the range that goes back becomes the maximum value and the speech rate becomes the maximum value, the range that goes back becomes the minimum value.

  FIG. 4B shows an example in which the relationship between the speech rate and the retroactive range is stepped. FIG. 4C shows an example where the relationship between the speech rate and the retroactive range is non-linear. The example shown in FIG. 4C is an utterance that best matches the human nature, in that the region where the utterance speed is low is easier to control, that is, the human is easier to control speaking slowly than speaking fast. It is an example which implement | achieves the relationship between speed and the retroactive range. FIG. 4D shows an example in which the relationship between the speech rate and the retroactive range is non-linear. Compared to the example shown in FIG. 4C, the example shown in FIG. 4D has the non-linear irregularities reversed. The example shown in FIG. 4D is an example that realizes control capable of setting a wide range that goes back even in a fast-speech utterance that is difficult for humans to control, that is, in a region where the speech rate is high.

  FIG. 4E to FIG. 4H show examples of correspondence tables between the speech speeds associated with the retroactive ranges so that the retroactive ranges become wider as the speech speed increases. Similarly to the examples shown in FIGS. 4A to 4D, the minimum value and the maximum value can be set for the speech rate in the examples shown in FIGS. 4E to 4H. In the example of the correspondence table shown in FIGS. 4E to 4H, the minimum value of the range that goes back corresponding to the minimum value of the speech rate and the maximum value of the range that goes back corresponding to the maximum value of the speech rate are set. In the example shown in FIG. 4E to FIG. 4H, when the speaking rate becomes the minimum value, the retroactive range becomes the minimum value. In the example shown in FIG. 4E to FIG. 4H, when the speaking rate becomes the maximum value, the retroactive range becomes the maximum value.

FIG. 4E shows an example where the relationship between the speech rate and the retroactive range is linear. FIG. 4F shows an example in which the relationship between the speech rate and the retroactive range is stepped. FIG. 4G shows an example in which the relationship between the speech rate and the retroactive range is non-linear. The example shown in FIG. 4G shows an utterance speed that best matches the human nature, in which the region where the utterance speed is low is easier to control, that is, it is easier to control speaking slowly than speaking fast. This is an example of realizing the relationship of the range that goes back to. FIG. 4H shows an example where the relationship between the speech rate and the retroactive range is non-linear. Compared to the example shown in FIG. 4G, the example shown in FIG. 4H has the non-linear irregularities reversed. The example shown in FIG. 4H is an example of realizing control capable of setting a wide range that can be traced back even in a fast-speech utterance that is difficult for humans to control, that is, in a region where the speech rate is high.

  For example, the control unit 13 determines the range to be traced using the correspondence table between the speech rate and the range to be traced in FIGS. 4A to 4H and outputs the range to the extraction unit 14.

  The extraction unit 14 obtains input voice data from the input unit 11 and a range that goes back from the control unit 13 as inputs. The extraction unit 14 extracts the reproduction audio data buffered in the main storage device 102 from a time point that corresponds to a range that goes back from the time point when the user's voice is input to the input unit 11.

  The reproduced audio data is obtained by converting the audio signal input from the network interface 107 (FIG. 1) or the tuner 108 (FIG. 1) as it is in the case of a digital signal or digitally converting it in the case of an analog signal. The audio data is buffered in the storage device 102 (FIG. 1). The buffer in the main storage device 102 has the same content as the audio data output from the output device 104 from the present time to a time that is a predetermined time long enough to go back from the time when the user's sound is input. Audio data is accumulated. The reproduced audio data is given a time stamp when the same audio data as the reproduced audio data is output from the information processing apparatus 1. For example, the time when the reproduced audio data is stored in the buffer of the main storage device 102 may be regarded as the time when the audio data having the same content as the reproduced audio data is output from the information processing apparatus 1. In addition, the time stamp given to the reproduced audio data starts from the elapsed time, time, and the beginning of the reproduced audio data from the start of the information processing device 1 based on a clock (not shown) provided in the information processing device 1 (0 : Zero), any output time may be used. In the first embodiment, the time series indicated by the time stamp given to the reproduced audio data is called an output time series of the reproduced audio data.

  Since the time stamp at the time of input to the information processing apparatus 1 is also given to the input sound data, the extraction unit 14 is the time point on the output time series of the reproduced sound data corresponding to the input time point of the input sound data. Can be requested. For example, when both the playback audio data and the input audio data are given as time stamps, the time indicated by the time stamp of the input audio data is the time of the input audio data. This is the time point on the output time series of the reproduced audio data corresponding to the input time point. When the time stamp of the reproduced audio data and the time stamp of the input audio data are different time series, the extraction unit 14 uses the time indicated by the time stamp of the input audio data as the output time series of the reproduced audio data. By converting to the above time, the time point on the output time series of the reproduced sound data corresponding to the input time point of the input sound data is obtained.

  The extraction unit 14 starts from a point in time that corresponds to the retroactive range input to the control unit 13 from the point in time on the output time series of the reproduced audio data corresponding to the input point in time of the input audio data obtained. Sequentially reproduced audio data is extracted. The extraction unit 14 outputs the extracted reproduced audio data. In the first embodiment, input voice data input from the input unit 11 is not used, but input voice data may be used. An embodiment of an information processing apparatus that performs processing using input voice data will be described later.

  The reproduced audio data output from the extraction unit 14 is decoded into an audio signal by the decoding process by the processor 101, and is reproduced and output from the output device 104.

FIG. 5 is a diagram illustrating an example of a processing flow of the information processing apparatus 1. The example shown in FIG. 5 shows a case where the user's voice is input from the microphone 103 while the information processing apparatus 1 is playing back playback voice data such as Internet radio by real-time processing.

  For example, when a user who listens to the sound output from the information processing apparatus 1 wants to listen again to the output sound immediately because the user has missed the sound, he / she can write a word related to the information he / she wants to hear again. To emit. For example, when the information processing apparatus 1 outputs the voice information “Today's electricity-related stock price closing price is A company X yen, B company Y yen...”, It desires to listen to the stock price of company A again. The user says “Company A”.

  The user's voice signal “Company A” is input to the input unit 11 through the microphone 103. When the user's voice signal “Company A” is input, the input unit 11 detects the input voice (OP1). The input unit 11 converts the user's voice signal “Company A” into input voice data “Company A”, and outputs it to the extraction unit 14 and the section detection unit 121.

  When the input voice data “Company A” is input, the section detection unit 121 measures the voice section length of the input voice “Company A” (OP2). For example, it is assumed that the voice section length of the input voice data “Company A” is 0.5 seconds. The section detection unit 121 outputs 0.5 seconds which is the voice section length of the input voice data “Company A” to the utterance speed calculation unit 124. The section detection unit 121 outputs the input voice data “Company A” to the voice recognition unit 122.

  When the input voice data “Company A” is input, the voice recognition unit 122 executes voice recognition processing (OP3). The speech recognition process of the speech recognition unit 122 reveals that the wording of the input speech data is “Company A”. The voice recognition unit 122 outputs the word “Company A” of the input voice data, which is the result of the voice recognition process, to the mora number calculation unit 123.

  When the word “Company A” of the input voice data, which is the result of the voice recognition process, is input, the mora number calculation unit 123 calculates the number of mora of the word “Company A” of the input voice data (OP4). When the result of the speech recognition process is “Company A”, the mora number calculation unit 123 calculates the mora number of “Company A” as 3 mora with “D”, “−”, and “Sha”. The mora number calculation unit 123 outputs to the utterance speed calculation unit 124 that “Company A” is 3 mora.

  The utterance speed calculation unit 124 calculates the utterance speed of the input voice data “Company A” when the voice section length of 0.5 seconds and the mora number 3 mora of the input voice data “Company A” are input (OP5). . The utterance speed calculation unit 124 sets the utterance speed of the input voice data “Company A” to the utterance speed = the number of mora of the input voice data “Company A” ÷ the voice section length of the input voice data “Company A” = 3 mora ÷ 0.5. Calculated as second = 6 mora / second. The utterance speed calculation unit 124 outputs the utterance speed 6 mora / second of the input voice data “Company A” to the control unit 13.

  When the speech rate (6 mora / second) of the input speech data “Company A” is input, the control unit 13 determines a range that goes back from the input speech data input time based on the speech rate (OP6). The control unit 13 refers to a correspondence table (see FIGS. 4A to 4H) between the speech rate stored in the auxiliary storage device 105 and the range that goes back, and determines the range that goes back from the input time point of the input voice data. For example, when the speech rate of the input voice data “Company A” is 6 mora / second, the control unit 13 determines the range that goes back from the input time point of the input voice data as 9 seconds. The control unit 13 outputs a range “9 seconds” that goes back from the input time point of the input voice data to the extraction unit 14.

When the input sound data “Company A” and the range “9 seconds” that goes back from the input time point of the input sound data are input, the extraction unit 14 stores the reproduced sound data stored in the buffer in the main storage device 102. A time point corresponding to the input time point of the input voice data on the output time series is obtained. The extraction unit 14 has a range (9) that extends from the reproduced audio data stored in the buffer of the main storage device 102 from the time on the output time series of the reproduced audio data corresponding to the time when the input audio “Company A” is input. The playback audio data is sequentially extracted starting from the time point that goes back (second) (OP7). The extraction unit 14 outputs the extracted reproduced audio data. The reproduced audio data output from the extraction unit 14 is reproduced and output from the output device 104.

  The information processing apparatus 1 sets a range that goes back from the time point on the output time series of the reproduced sound data corresponding to the input time point of the user's sound, according to the speaking rate of the sound that the user utters. For example, the retroactive range is set to be smaller as the utterance speed increases. For example, the range that goes back as the utterance speed increases is set to increase. Thus, according to the information processing apparatus 1, it is possible to control the setting of the range that goes back from the time point on the output time series of the reproduced voice data corresponding to the voice input time point of the user according to the speech speed. .

  In addition, when the information processing apparatus 1 sets a long range when the utterance speed is low, that is, when the user speaks slowly, information to be reconfirmed such as missed information is again displayed. The possibility that the user listens from the information processing apparatus 1 is increased.

  In addition, when the speaking rate is high, that is, when the user speaks quickly, there is a high possibility that the information that the user wants to reconfirm is present at a point near the user's voice input point. By setting the range that goes back when the information processing apparatus 1 has a high utterance speed to be small, the processing amount of the information processing apparatus 1 can be reduced, and improvement in the efficiency of reproduction processing of audio data can be expected.

Second Embodiment
In the information processing apparatus according to the second embodiment, when voice related to information to be reconfirmed, such as information missed from the user, is input during reproduction of the audio data, a voice word (keyword) related to information to be reconfirmed. And retrieves the wording from the buffered audio data. As a result of the search, the information processing apparatus reproduces the voice data including the keyword pinpoint. In addition, the information processing apparatus according to the second embodiment controls the search range of the audio data based on the voice rate of the voice uttered by the user. The configuration of the information processing apparatus of the second embodiment is partially in common with the configuration of the information processing apparatus of the first embodiment. In the second embodiment, descriptions of parts common to the first embodiment are omitted.

<< Configuration example of information processing apparatus >>
FIG. 6 is an explanatory diagram of functions realized when the processor of the information processing apparatus executes the audio data reproduction program. The hardware configuration of the information processing apparatus 2 shown in FIG. 6 is the same as that of the information processing apparatus 1 shown in FIG. In the information processing apparatus 2, when the processor 101 executes the audio data reproduction program, the data input unit 21, the recording unit 22, the audio input unit 24, the calculation unit 25, the control unit 26, the search unit 27, and the output range determination unit 28. , And the output unit 29 can be realized. That is, the information processing apparatus 2 performs the data input unit 21, the recording unit 22, the storage unit 23, the audio input unit 24, the calculation unit 25, the control unit 26, the search unit 27, and the output range determination unit by executing the audio data reproduction program. 28, and an output unit 29.

The data input unit 21 is connected to the network interface 107 or the tuner 108 shown in FIG. 1 and obtains an audio signal as an input from another device by wireless communication or wired communication. The data input unit 21 receives, for example, an audio signal such as a radio broadcast radio wave or a one-segment broadcast as an input. The data input unit 21 converts the audio signal into audio data that can be handled by the information processing apparatus 2, and outputs the audio data to the recording unit 22 and the output unit 29. The data input unit 21 converts, for example, an analog signal into a digital signal and encodes the digital signal to obtain voice data. Hereinafter, the audio data input to the information processing apparatus 2 via the data input unit 21 is referred to as reproduction audio data.

  Further, the data input unit 21 assigns the time when the reproduction audio data is output to the reproduction audio data as a time stamp. The time stamp includes, for example, an elapsed time since the start of the information processing device 2 based on a clock (not shown) provided in the information processing device 2, a time managed by the information processing device 2, and the beginning of the reproduced audio data (0 : Zero) may be any elapsed time. In the second embodiment, the time required for the reproduction audio data to be input to the information processing apparatus 2 and output at the time when the reproduction audio data is input to the data input unit 21 is predicted. Is added as the output time point of the audio data for reproduction. In the second embodiment, the time series indicated by the time stamp given to the reproduction audio data is called the output time series of the reproduction audio data.

  The recording unit 22 obtains reproduction audio data from the data input unit 21 as an input. The recording unit 22 stores the reproduction audio data in the storage unit 23.

  The storage unit 23 is a part of the storage area of the main storage device 102 shown in FIG. The storage unit 23 holds the playback audio data recorded by the recording unit 22 for a predetermined time.

  The voice input unit 24 is connected to the microphone 103 shown in FIG. 1 and receives a voice signal emitted by the user through the microphone 103 as an input. The voice input unit 24 converts the input voice signal into voice data that can be handled by the information processing apparatus 2. For example, the voice input unit 24 converts an input voice signal from an analog voice signal to a digital voice signal, and encodes the digital voice signal to obtain voice data. The voice input unit 24 outputs user voice data to the calculation unit 25 and the search unit 27. The utterance content of the voice uttered by the user is a phrase indicating information that the user desires to reconfirm. The user's voice and voice data input to the voice input unit 24 are hereinafter referred to as keyword voice and keyword voice data.

  Further, the voice input unit 24 assigns the time when the keyword voice data is input as a time stamp to the keyword voice data. The time stamp may be, for example, an elapsed time from the activation of the information processing apparatus 2 based on a clock (not shown) provided in the information processing apparatus 2 or a time managed by the information processing apparatus 2.

  The calculation unit 25 receives keyword voice data from the voice input unit 24 as an input. The calculation unit 25 calculates the utterance speed of the keyword voice data. The calculation unit 25 includes a section detection unit 251, a speech recognition unit 252, a mora number calculation unit 253, and an utterance speed calculation unit 254. The calculation unit 25, the section detection unit 251, the speech recognition unit 252, the mora number calculation unit 253, and the speech rate calculation unit 254 included in the calculation unit 25 are included in the calculation unit 15 and the calculation unit 15 in the first embodiment. This is the same as each of the section detection unit 121, the speech recognition unit 122, the mora number calculation unit 123, and the speech rate calculation unit 124. The calculation unit 25 outputs the calculated utterance speed of the keyword voice to the control unit 26.

The control unit 26 obtains the utterance speed of the keyword voice data as an input. For example, the control unit 26 determines a search range for searching for a phrase (keyword) of the keyword voice in the reproduction voice data held in the storage unit 23 based on the utterance speed. The control unit 26 includes a correspondence table storage unit 261. The correspondence table storage unit 261 is a part of the data storage area of the auxiliary storage device 105 shown in FIG. 1, and holds a correspondence table (see FIGS. 4A to 4H) of the speech rate and the search range. The control unit 26 determines the search range based on the correspondence table between the utterance speed and the search range held in the correspondence table storage unit 261 in the same manner as the retrospective range determination process of the control unit 13 of the first embodiment. . The control unit 26 outputs the determined search range to the search unit 27. The search range may be defined by any of time length, syllable number, mora number, and the like. In 2nd Embodiment, the control part 26 demonstrates the case where a search range is defined by time.

  The search unit 27 receives the keyword voice data and the search range as inputs. The search unit 27 obtains a time point on the output time series of the reproduction sound data held in the storage unit 23 corresponding to the input time point of the keyword sound data. The search unit 27 uses the time stamp given to the keyword voice data and the time stamp given to the playback voice data, the time point on the output time series of the playback voice data corresponding to the input time point of the keyword voice data. Can be requested. For example, when both the playback audio data and the keyword audio data are given as time stamps from the time when the information processing apparatus 2 is activated by the clock, the time indicated by the time stamp of the keyword audio data is the keyword audio data. Is the time point on the output time series of the reproduction audio data corresponding to the input time point. In addition, when the time stamp of the playback audio data and the time stamp of the keyword audio data are different, the search unit 27 uses the time indicated by the time stamp of the keyword audio data as the output time of the playback audio data. The time on the output time series of the reproduced audio data corresponding to the input time of the input audio data is obtained by converting the time into the time on the sequence.

  The search unit 27 traces the time corresponding to the search range from the playback audio data held in the storage unit 23 from the time corresponding to the input time of the keyword voice of the user on the output time series of the playback audio data. Audio data for reproduction up to the point in time is read out as partial audio data. The search unit 27 performs a search for whether or not a keyword is included in the read partial audio data. Further, the search unit 27 stores, as partial audio data, reproduction audio data from the time point of the silent part immediately before or immediately after the input time of the keyword sound from the user to the time point that goes back the time corresponding to the search range. 23 may be read out. The silent part refers to a part of exhalation at the time of breathing and a part of silence between a sentence and a certain time or more. Hereinafter, when the keyword voice input time is referred to, the time on the output time series of the reproduction voice data corresponding to the keyword voice input time is indicated.

  The search unit 27 uses, for example, a speech recognition technique such as word spotting to search for a keyword issued by a user in the partial voice data read from the storage unit 23.

  FIG. 7 shows an example in which the search unit 27 searches for a keyword using the word spotting technique. Word spotting is a method for extracting a specific word to be detected using a parameter having a characteristic such as a frequency component of audio data.

  In the example shown in FIG. 7, when the user utters the keyword “Company A” while the playback audio is being output from the information processing apparatus 2, the partial voice data “today's stock price” extracted from the playback audio data Is an example of searching whether the keyword “Company A” is included in “Company A 500 yen, Company B. FIG. 7 shows a speech waveform of the partial speech data with the content “Today's stock price is 500 yen for Company A, Company B...” And a speech waveform of the keyword speech data of “Company A”. The search unit 27 reads, from the storage unit 23, the playback audio data included from the time point when the keyword voice is input to the point when the time corresponding to the search range is traced back. The search unit 27 detects the keyword “Company A” from the partial speech data by comparing the partial speech data with the keyword speech data “Company A”. The search unit 27 divides the partial voice data into, for example, syllables or words. A parameter is calculated for each partial voice data segmented by syllables or words, and this parameter is compared with the parameter of the keyword voice data “Company A”. The search unit 27 detects the keyword “Company A” from the partial speech data when one parameter of the partial speech data divided by syllables or words matches the parameter of the keyword speech data “Company A”, Determine successful detection. When a keyword is detected in the read partial audio data, the search unit 27 outputs the detection result to the output range determination unit 28.

  If the parameter of the partial voice data divided by syllables or words does not match the parameter of the keyword voice data “Company A”, the search unit 27 cannot detect the keyword “Company A” from the partial voice data and detects it. Determine failure. If no keyword is detected from the read partial audio data, the search unit 27 outputs a search range reset request to the control unit 26.

  When receiving a search range reset request from the search unit 27 as an input, the control unit 26 resets the search range.

  FIG. 8A, FIG. 8B, and FIG. 8C are diagrams showing processing 1 to processing 3, which are examples of search range resetting processing. The control unit 26 may perform any one of the processing 1, processing 2, and processing 3 shown in FIGS. 8A, 8B, and 8C, respectively, as the search range resetting processing. 8A, 8B, and 8C show cases where the search range at the time of the first search is set to 5 seconds by the control unit 26. FIG. 8A, 8B, and 8C show a case where the keyword is “Company A”.

  FIG. 8A is a diagram illustrating a process 1 which is an example of a search range resetting process. When a search range reset request is input from the search unit 27, the control unit 26 determines the second search range with the same size as the first time. The control unit 26 outputs the determined second search range to the search unit 27.

  When the second search range having the same size as the first search range is input, the search unit 27 further searches for the second search range from the time when the first search range is traced back from the input time of the keyword voice. The partial voice data up to the point in time is read and the keyword is searched. For example, in FIG. 8A, the search unit 27 goes back the second search range (5 seconds) from the time when the first search range (5 seconds) is traced from the input time point of the keyword voice “Company A”. Audio data for reproduction up to the time is read as partial audio data. That is, the reproduction audio data included in the range from 5 seconds to 10 seconds before the input time point of the keyword audio “Company A” is read as partial audio data. FIG. 8A shows a case where the keyword “Company A” is detected in the second search for searching within the partial voice data included in the range from 5 seconds to 10 seconds before the input time of the keyword voice. When the keyword “Company A” is detected in the second search, the search unit 27 outputs the detection result to the output range determination unit 28.

  If the keyword “Company A” is not detected in the second search, the search unit 27 again outputs a search range reset request to the control unit 26. When the search range reset request is input again, the control unit 26 sets the third search range to the same search range as the first time and the second time. The control unit 26 outputs the third search range to the search unit 27. When the third search range is input, the search unit 27 starts from the time when the first search range and the second search range are traced back to the time when the third search range is traced. The partial voice data is read out and a keyword is searched. In FIG. 8A, the third search range (5 seconds) is further traced back from the time when the first search range (5 seconds) and the second search range (5 seconds) are traced from the input time point of the keyword voice data “Company A”. Reproduction audio data included up to the point of time) is read out as partial audio data. That is, the search unit 27 reads out reproduction audio data included in a range from 10 seconds to 15 seconds before the input time point of the keyword audio data “Company A” as partial audio data. The search unit 27 executes a third search for the keyword “Company A” in the read partial voice data.

The search unit 27 and the control unit 26 perform the above-described search processing until the keyword “Company A” is detected from the partial voice data read from the storage unit 23, and n + 1 (n is a natural number not including 0). ) Repeat times. If the keyword “Company A” is not detected after repeating n + 1 times, the search unit 27 outputs “detection failure” to the output range determination unit 28. When “detection failure” is input, the output range determination unit 28 notifies the user that the keyword detection has failed through the output unit 29 and the output device 104 connected to the output unit 29. As a cause of keyword detection failure, for example, a case where the user's utterance is unclear and speech recognition cannot be performed correctly is considered.

  FIG. 8B is a diagram illustrating a process 2 that is an example of a search range resetting process. When a search range reset request is input from the search unit 27, the control unit 26 determines α times the first search range (α> 1 as the second search range. The determined second search 8B is output to the search unit 27. For example, in the example illustrated in FIG. 8B, α = 2, and the control unit 26 sets α times (2 times the first search range (5 seconds) as the second search range. Is 10 seconds).

  When the second search range, which is α times the first search range, is input, the search unit 27 further searches for the second search range from the time when the first search range is traced back from the input time of the keyword voice. Partial audio data included in the retroactive time point is read from the storage unit 23. In FIG. 8B, the search unit 27 further goes back to the second search range (10 seconds) from the time when the first search range (5 seconds) is traced from the input time point of the keyword voice data “Company A”. Is read out as partial audio data. That is, the search unit 27 reads out the reproduction audio data included in the range from 5 seconds to 15 seconds before the input time point of the keyword audio data “Company A” as partial audio data. FIG. 8B shows a case where the keyword “Company A” is detected in the second search for searching within the partial speech data included between 5 seconds and 15 seconds before the input time point of the keyword speech data “Company A”. ing.

  If the keyword “Company A” is not detected even in the second search, the search unit 27 and the control unit 26 until the keyword “Company A” is detected as in the case of the process 1 shown in FIG. 8A. The search process is repeated n + 1 times. If the (n + 1) th search process fails, the search unit 27 outputs “search failure” to the output range determination unit 28.

  FIG. 8C is a diagram illustrating a process 3 which is an example of a search range resetting process. When the search range reset request is input from the search unit 27, the control unit 26 stores the reproduction audio data stored in the storage unit 23 from the time when the first search range is traced back from the input time of the keyword voice. The first search range is determined up to the beginning, and the second search range is output to the search unit 27.

  When the search range for the second time is input, the search unit 27 starts from the time when the first search range is traced from the time point when the keyword voice is input to the beginning of the playback audio data stored in the storage unit 23. Data is read from the storage unit 23. The search unit 27 executes a second search for the keyword “Company A” in the read partial audio data. In the process 3 of FIG. 8C, the search unit 27 searches from the beginning of the data, so the possibility that the keyword “Company A” is detected in the second search is higher than in the process 1 and the process 2.

The search range resetting process described with reference to FIGS. 8A, 8B, and 8C is summarized as follows.
(Processing 1) As the search range for the second and subsequent times, the control unit 26 sets the same range as the first search range.
(Process 2) As a search range for the second and subsequent times, the control unit 26 sets a search range that is α times (α> 1) the previous search range.
(Process 3) As the second search range, the control unit 26 sets from the beginning of the reproduction audio data to the start point of the partial audio data of the first search.

  FIGS. 9A to 9D are diagrams showing process A to process D, which are examples of processes when a plurality of keywords are detected in the partial voice data. The search unit 27 may execute any of the process A to the process D when a plurality of keywords are detected in the partial voice data. The examples shown in FIGS. 9A to 9D show an example in which the search unit 27 executes a search process of “Company A” as a keyword.

  FIG. 9A is a diagram illustrating a process A that is an example of a process when a plurality of keywords are detected in partial audio data. In the process A, the search unit 27 outputs, to the output range determination unit 28, the keyword closest in time from the keyword input point in time as the search result.

  FIG. 9B is a diagram illustrating a process B which is an example of a process when a plurality of keywords are detected in the partial voice data. In the process B, the search unit 27 outputs the keyword farthest in time from the keyword input time point to the output range determination unit 28 as a search result.

  FIG. 9C is a diagram illustrating a process C which is an example of a process when a plurality of keywords are detected in the partial voice data. In the process C, the search unit 27 outputs an arbitrary keyword among the detected keywords to the output range determination unit 28 as a search result.

  FIG. 9D is a diagram illustrating a process D which is an example of a process when a plurality of keywords are detected in the partial voice data. In the process D, the search unit 27 outputs all detected keywords to the output range determination unit 28 as search results.

  When the keyword search result is input from the search unit 27, the output range determination unit 28 determines the head position when reading the reproduction audio data from the storage unit 23. The head position when the reproduction audio data is read from the storage unit 23 is the reproduction start position of the reproduction audio data.

  FIG. 10A and FIG. 10B are diagrams illustrating an example of processing for determining the head position when reading audio data for reproduction from the storage unit 23. In FIG. 10A and FIG. 10B, for example, a user who wants to rehearse the stock price of Company A while outputting the voice information “The closing price of today's electrical equipment-related stock price is A Company X Yen, B Company Y Yen”. An example of issuing a keyword “A company” is shown. In the example shown in FIGS. 10A and 10B, when the keyword “Company A” is detected by the keyword search processing by the search unit 27, the output range determination unit 28 reads the playback audio data from the storage unit 23. An example of deciding the head of is shown.

  In the example shown in FIG. 10A, the output range determination unit 28 determines the detected keyword “Company A” as the head when reading playback audio data from the storage unit 23. The output range determination unit 28 sequentially reads out the reproduction audio data from the storage unit 23 with the keyword “Company A” at the head, and outputs it to the output unit 29. From the output device 104 through the output unit 29, playback audio data with the detected keyword “Company A” as the start position, such as “Company A X Yen, Company B Yen Yen. Is output.

In the example shown in FIG. 10B, the output range determination unit 28 determines the silent part immediately before the detected keyword “Company A” as the head when reading the reproduction audio data from the storage unit 23. The output range determination unit 28 reads the reproduction audio data from the storage unit 23 with the silent part immediately preceding the detected keyword “Company A” in time as the head, and outputs the reproduction audio data to the output unit 29. The silent part is a silent part of a certain time or more between sentences and a breathing part of a person who reads out information. From the output device 104 through the output unit 29, “Today's electric machinery related stock price closing price is A company X yen, B company Y yen. The sound data for reproduction is output starting from the silent part.

  The information processing device 2 executes either the process shown in FIG. 10A or the process shown in FIG. 10B.

  The output unit 29 receives the playback audio data from the data input unit 21 as an input. For example, the output unit 29 decodes the reproduction audio data into a digital signal. The output unit 29 is connected to an output device 104 (FIG. 1) such as a speaker, and outputs the decoded digital signal to the output device 104. The digital signal output from the output unit 29 is converted into an analog signal and output from the output device 104 as an audio signal.

  Further, the output unit 29 switches the input of the reproduction audio between the data input unit 21 and the output range determination unit 28. When the reproduction audio data is input from the output range determination unit 28, the output unit 29 switches the input of the reproduction audio data from the data input unit 21 to the output range determination unit 28. After that, for example, when the audio data for reproduction for a predetermined time is not input from the output range determination unit 28, the output unit 29 switches the input of the reproduction audio data from the output range determination unit 28 to the data input unit 21.

  The output unit 29 may reproduce the reproduction audio data sequentially input from the output range determination unit 28 at a single speed. Further, the output unit 29 may reproduce the reproduction audio data sequentially input from the output range determination unit 28 at a double speed, for example. When the reproduction audio data input from the output range determination unit 28 is reproduced at a double speed, the reproduction audio data input from the output range determination unit 28 at a certain point in time is processed in real time. Keep up with audio data for playback. Thereafter, since the playback audio data is not input from the output range determination unit 28, the output unit 29 switches the input of the playback audio data from the output range determination unit 28 to the data input unit 21, and is input from the data input unit 21. Playback audio data is played back at 1x speed.

<< Processing flow of information processing apparatus >>
FIG. 11 is a diagram illustrating an example of a processing flow of the information processing apparatus 2. The example shown in FIG. 11 is an example in the case where a keyword related to information that the user wants to re-listen is issued during reproduction of audio data.

  The information processing apparatus 2 starts the processing flow of FIG. 11 together with the start of reproduction (output start) of the audio data from the output unit 29.

  For example, when the voice information “Today's electric equipment-related stock closing price is A company X yen, B company Y yen... Issue the keyword “Company A”.

  The voice input unit 24 detects the utterance of the keyword “Company A” by the user as the input voice (OP21). The voice input unit 24 outputs the keyword voice data “Company A” to the calculation unit 25 and the search unit 27.

  When the keyword voice data “Company A” is input, the section detection unit 251 of the calculation unit 25 measures the voice section length of the keyword voice data “Company A” (OP22). For example, it is assumed that the voice segment length of the keyword voice data “Company A” is 0.5 seconds. The section detection unit 251 outputs the voice section length of the keyword voice data “Company A” to the utterance speed calculation unit 254. The section detection unit 251 outputs the keyword voice data “Company A” to the voice recognition unit 252.

When the keyword speech data “Company A” is input, the speech recognition unit 252 executes speech recognition processing (OP23). The voice recognition process of the voice recognition unit 252 reveals that the keyword is “Company A”. The voice recognition unit 252 outputs the result of the voice recognition process “Company A” to the mora number calculation unit 253.

  When the result of speech recognition “Company A” is input, the mora number calculation unit 253 calculates the mora number (OP24). When the result of the speech recognition is “Company A”, the mora number calculation unit 253 calculates the mora number of “Company A” as 3 mora with “D”, “−”, and “Sha”. The mora number calculation unit 253 outputs to the utterance speed calculation unit 254 that “Company A” is 3 mora.

  The utterance speed calculation unit 254 calculates the utterance speed of the keyword voice data “Company A” when the voice interval length 0.5 seconds and the mora number 3 mora of the keyword voice data “Company A” are input (OP25). . The speech rate calculation unit 254 calculates the speech rate of the keyword speech data “Company A” as speech rate = number of mora of the keyword speech data “Company A” ÷ speech interval length of the keyword speech data “Company A” = 3 mora ÷ 0.5. Calculated as second = 6 mora / second. The utterance speed calculation unit 254 outputs the utterance speed 6 mora / second of the keyword voice data “Company A” to the control unit 26.

  When the utterance speed 6 mora / second of the keyword voice data “Company A” is input, the control unit 26 determines a search range based on the utterance speed (OP26). The control unit 26 determines the search range with reference to a correspondence table (see FIGS. 4A to 4H) between the speech rate and the search range held in the correspondence table storage unit 261. For example, when the speech rate of the input voice data “Company A” is 6 mora / second, the control unit 26 determines the search range as 3 seconds. The control unit 26 outputs the determined search range “3 seconds” to the search unit 27.

  When the keyword voice data “Company A” and the search range “3 seconds” are input, the search unit 27 retrieves the partial voice data included in the range retroactive to the search range from the input point of the keyword voice data from the storage unit 23. Read and search for keywords in the partial audio data (OP27). The search unit 27 executes the search process using, for example, word spotting shown in FIG.

  When the search process of the search unit 27 fails, that is, when the keyword “Company A” is not detected in the partial voice data (OP28: No), the search unit 27 sends a search range reset request to the control unit 26. Output. When the search range reset request is input, the control unit 26 resets the search range (OP29). The control unit 26 outputs the reset search range to the search unit 27. The search unit 27 reads the partial voice data included in the range retroactive to the search range reset from the search range of the previous search from the storage unit 23 and searches for the keyword “Company A” again (OP27).

  When the search process of the search unit 27 is successful, that is, when the keyword “Company A” is detected in the partial voice data (OP28: Yes), the search unit 27 sends the detection result to the output range determination unit 28. Output. The output range determination unit 28 determines a playback start point of the playback audio data, sequentially reads the playback audio data from the storage unit 23, and outputs the playback audio data to the output unit 29. When the reproduction audio data starts to be input from the output range determination unit 28, the output unit 29 interrupts the output processing of the reproduction audio data input from the data input unit 21, and the reproduction input from the output range determination unit 28 Audio data is output (OP30).

  The processing in OP27, OP28, and OP29 is one of processing A, processing B, and processing C, which are examples of search range resetting processing, shown in FIGS. 8A, 8B, and 8C.

<< Effects of Second Embodiment >>
The information processing apparatus 2 according to the second embodiment recognizes a keyword related to information that the user wants to re-listen, and searches for the keyword in the partial voice data included in the range that goes back to the search range from the keyword input time point. The information processing device 2 resets the search range when no keyword is detected in the partial voice data. In this way, the information processing apparatus 2 can accurately output information that the user wants to listen to again by searching for a keyword and changing the search range according to the search result and searching again.

  Further, since the user can reproduce the information to be reconfirmed only by speaking a keyword related to the information to be reconfirmed such as missed listening, the operation is easy. Further, by searching for a keyword related to information to be reconfirmed, the user can pinpoint the information to be reconfirmed.

  The information processing device 2 sets the search range according to the utterance speed of the user's utterance, as in the information processing device 1 of the first embodiment. For example, the search range is set to become smaller as the utterance speed increases. For example, the search range is set to increase as the utterance speed increases. Thus, according to the information processing apparatus 1, it is possible to control the setting of the keyword search range in accordance with the speech rate.

  Further, when the information processing device 2 sets the search range to be large when the utterance speed is low, that is, when the user speaks slowly, the possibility that the keyword is included in the search range increases. The accuracy with which keywords are detected in one search is improved.

  In addition, when the speaking rate is high, that is, when the user speaks quickly, there is a high possibility that the information that the user wants to reconfirm is present at a point near the user's voice input point. By setting the search range to be small when the information processing device 2 has a high utterance speed, it is possible to reduce the amount of keyword search processing performed by the information processing device and to expect improvement in efficiency.

  The information processing apparatus 2 can control the reading start position (playback start position) of the playback audio data held in the storage unit 23 according to the keyword search result.

<Modification>
In the first embodiment and the second embodiment, the utterance speed calculation unit 124 and the utterance speed calculation unit 254 calculate the utterance speed from the number of mora and the time length of the input speech. Instead of the number of mora of the input speech, the speech rate calculation unit 124 and the speech rate calculation unit 254 may calculate the speech rate using the spectral characteristics of the input speech. The utterance speed calculation unit 124 and the utterance speed calculation unit 254 can use an utterance speed calculation method that is generally widely used.

  In 1st Embodiment and 2nd Embodiment, the control part 13 and the control part 26 each determined the range or search range which goes back in units of time. Instead of determining the range or search range that goes back in time units, the control unit 13 and the control unit 26 determine the range or search range that goes back using the number of syllables, the number of words, the expiratory paragraph, the silent period, and the like. Also good.

  In the first embodiment and the second embodiment, the information processing apparatus 1 and the information processing apparatus 2 are triggered by the user's voice input while reproducing the audio signals sequentially input from the network interface 107 and the tuner 108 in real time. The process of reproducing audio data from a time point that is a predetermined range backward from the user's voice input time point has been described. The information processing apparatus 1 and the information processing apparatus 2 can execute the processes described in the first embodiment and the second embodiment even during reproduction of audio data held in advance in the auxiliary storage device 105 or the like.

In the second embodiment, the search unit 27 uses the word spotting technique as a method for searching for keywords in the search range. Instead of the word spotting technique, the search unit 27 may use another voice recognition technique. For example, the playback audio data is converted into text and stored in the storage unit 23, and the search unit 27 converts the keyword audio data into text, and searches for keywords from the text of the playback audio data included in the search range. You may go.

  In the second embodiment, after the voice recognition unit 252 finishes the voice recognition process for keyword voice data, the voice recognition result of the keyword voice data may be confirmed to the user. For example, the voice recognition unit 252 outputs the voice recognition result of the keyword voice data to the output unit 29. When the voice recognition result of the keyword voice data is input, the output unit 29 outputs a voice for inquiring the user whether the voice input result is correct. The voice for inquiring the user about the voice recognition result is stored in the auxiliary storage device 105 (FIG. 1). For example, when the speech recognition result of the keyword speech data is “Company A”, the output unit 29 outputs a speech “Are you sure that“ Company A ”?”. As described above, by confirming the speech recognition result of the keyword speech data with the user, it is possible to prevent a keyword detection failure in the search unit 27 due to the keyword speech recognition failure.

  Further, in the second embodiment, an example has been described in which a user issues a keyword (word) regarding information that the user wants to explicitly hear again. When the user designates information that he / she wants to hear again during the output of the playback audio data, the user may, for example, issue a sentence such as “How much is the stock price of Company A?”. When the keyword voice input from the user is a sentence or a plurality of words, for example, the information processing apparatus 2 performs the following process.

  For example, a case where a keyword voice input “What is the stock price of Company A?” Is taken as an example. The auxiliary storage device 105 (FIG. 1) holds a list of keyword candidates. When the keyword speech data “What is the stock price of Company A?” Is input, the speech recognition unit 252 selects keyword candidates in the list held in the auxiliary storage device 105 from the keyword speech data, for example, Detect using word spotting techniques. The voice recognition unit 252 can detect the qi word “Company A” and “stock price” from the keyword voice data using the keyword candidates in the list. The voice recognition unit 252 may output both the detected keyword “Company A” and “stock price” to the search unit 27. Alternatively, in the keyword candidate list, priority is given according to the type of word (for example, proper noun, general noun, etc.), and either the keyword “Company A” or “stock price” detected according to the priority is selected. May be output to the search unit 27. When a keyword is input, the search unit 27 executes the process described in the second embodiment. As described above, the information processing apparatus 2 includes a list of keyword candidates, and by extracting a keyword from the keyword voice data, even if the voice uttered by the user is a sentence or the like, information that the user wants to hear properly Can be output.

<Others>
The following is further disclosed regarding the above embodiment.
(Appendix 1)
An information processing apparatus for reproducing audio data,
An input unit for receiving a voice uttered by the user;
A calculation unit for calculating an utterance speed of the voice;
A control unit that determines a range that goes back from the time corresponding to the input time of the sound on the output time series of the sound data, according to the speech rate;
An information processing apparatus comprising:
(Appendix 2)
The information processing apparatus according to appendix 1, wherein the control unit determines the range so that the retroactive range decreases as the utterance speed increases.
(Appendix 3)
The control unit sets the maximum value of the retroactive range corresponding to the lower limit value of the speech rate, sets the minimum value of the retroactive range corresponding to the upper limit value of the speech rate, and the speech rate is the lower limit value The information processing apparatus according to appendix 1, wherein the retroactive range is set to be narrower between the maximum value and the minimum value as the value increases from the maximum value to the upper limit value.
(Appendix 4)
The information processing apparatus according to appendix 1, wherein the control unit determines the retroactive range so that the retroactive range increases as the speech rate of the voice increases.
(Appendix 5)
The supplementary note according to any one of appendices 1 to 4, further including an extraction unit that extracts partial voice data from a time point corresponding to the voice input time point on the output time series of the voice data to a time point retroactive to the retroactive range. Information processing device.
(Appendix 6)
A retrieval unit for retrieving whether or not the speech utterance content is included in the partial speech data from the time corresponding to the speech input time on the output time series of the speech data to the time traced back to the retroactive range. The information processing apparatus according to any one of appendices 1 to 4, further comprising:
(Appendix 7)
The control unit expands the retroactive range when the speech content of the voice is not included in the partial voice data,
The search unit includes, in the partial audio data included from the time corresponding to the input time of the audio on the output time series of the audio data to the time point back to the enlarged retrospective range by the control unit, The information processing apparatus according to appendix 6, which searches whether or not utterance content is included.
(Appendix 8)
When the plurality of speech utterance contents are detected in the partial speech data, the search unit uses at least one of the detected speech utterance contents as a search result. The information processing apparatus according to appendix 6 or 7.
(Appendix 9)
The information processing apparatus according to any one of appendices 6 to 8, further including a determination unit that determines an output start point of the audio data based on a search result of the search unit.
(Appendix 10)
The information processing apparatus according to appendix 9, wherein the determining unit determines the utterance content of the voice detected by the search unit as an output start point of the voice data.
(Appendix 11)
The information processing apparatus according to supplementary note 10, wherein the determination unit determines a silent part existing in a part that goes back from the speech utterance content detected by the search part as an output start point of the voice data.
(Appendix 12)
An information processing device that reproduces audio data
A step of accepting a voice uttered by a user;
Calculating the speech rate of the voice;
Determining a range going back from a time corresponding to an input time of the sound on an output time series of the sound data according to the speech speed;
A program for running

DESCRIPTION OF SYMBOLS 1, 2 Information processing apparatus 11 Input part 12, 25 Calculation part 13, 26 Control part 14 Extraction part 21 Data input part 22 Recording part 23 Storage part 24 Voice input part 27 Search part 28 Output range determination part 29 Output part 101 Processor 102 Main storage device 103 Microphone 104 Output device 105 Auxiliary storage device 107 Network interface 108 Tuner 109 Bus 121, 251 Section detection unit 122, 252 Speech recognition unit 123, 253 Mora number calculation unit 124, 254 Speech rate calculation unit 261 Correspondence table storage unit

Claims (5)

An information processing apparatus for reproducing audio data,
An input unit for receiving a voice uttered by the user;
A calculation unit for calculating an utterance speed of the voice;
A control unit that determines a range that goes back from the time corresponding to the input time of the sound on the output time series of the sound data, according to the speech rate;
An information processing apparatus comprising: The information processing apparatus according to claim 1, wherein the control unit determines the retroactive range so that the retroactive range decreases as the utterance speed increases. A retrieval unit for retrieving whether or not the speech utterance content is included in the partial speech data from the time corresponding to the speech input time on the output time series of the speech data to the time traced back to the retroactive range. The information processing apparatus according to claim 1 or 2. The control unit expands the retroactive range when the speech content of the voice is not included in the partial voice data,
The search unit includes, in the partial audio data included from the time corresponding to the input time of the audio on the output time series of the audio data to the time point back to the enlarged retrospective range by the control unit, The information processing apparatus according to claim 3, wherein a search is made as to whether or not utterance content is included. An information processing device that reproduces audio data
A step of accepting a voice uttered by a user;
Calculating the speech rate of the voice;
Determining a range going back from a time corresponding to an input time of the sound on an output time series of the sound data according to the speech speed;
A program for running

Images