JP6044490B2 - Information processing apparatus, speech speed data generation method, and program - Google Patents

Description

  The present invention relates to an information processing apparatus, a speech speed data generation method, and a program that generate speech speed data representing the utterance time of audio output in accordance with video.

2. Description of the Related Art Conventionally, in content including video such as movies and television programs, synthesized voice generated by voice synthesis is output in accordance with video output.
As a device for outputting synthesized voice in accordance with this video, the expansion rate of the voice is determined so that the utterance time length of the synthesized audio matches the program broadcast time, and the speech speed in the synthesized speech is determined based on the determined expansion rate. Has been proposed (see Patent Document 1).

JP 2012-078755 A

When the speech speed is converted by the device described in Patent Document 1, since the entire synthesized speech is expanded and contracted, the utterance time of each word included in the sentence uttered by the synthesized speech is also expanded and contracted.
If the utterance time is shortened when the utterance time is expanded or contracted, words included in the sentence may be difficult to hear. For this reason, there has been a problem that it is difficult for a person who has heard a word whose utterance time is shortened to understand the entire content of the utterance.

In other words, the conventional technique has a problem that the speech speed cannot be adjusted in the synthesized speech so that the content of the utterance can be easily understood.
Therefore, an object of the present invention is to make it possible to adjust the speech speed so that the content of the utterance can be easily understood in the synthesized speech.

  The present invention made in order to achieve the above object includes a text acquisition means, an analysis means, a closeness acquisition means, a speech speed determination means, an identification information acquisition means, a history acquisition means, a history analysis means, An information processing apparatus comprising update means.

  In the present invention, the text acquisition means acquires text data representing a character string of information output by sound in accordance with the video, the analysis means analyzes the text data acquired by the text acquisition means, and the text data Each word included in the character string represented by is identified.

  Then, the familiarity acquisition unit acquires the familiarity corresponding to each word specified by the analysis unit from the familiarity database. The familiarity database referred to here is a database in which familiarity information is stored, and the familiarity information is information in which each word is associated with a familiarity representing the recognition degree of each word in advance.

  Furthermore, the speech rate determination means has a longer proportion of the utterance time of the word in the utterance time of the entire information represented by the text data, as the word indicating that the familiarity acquired by the familiarity acquisition means is lower. Thus, speech speed data in which the utterance time of the word is adjusted is generated. The speech speed data referred to here is data representing the utterance time of synthesized speech output by speech synthesis, and data representing the utterance time of each phoneme constituting the character string of information represented by the text data. is there.

  In the present invention, the identification information acquisition unit acquires user identification information for identifying a user. Then, the history acquisition means is acquired by the identification information acquisition means from the use history in which each of the user identification information is associated with viewing information representing a video viewed by the user corresponding to each user identification information in the past. User viewing information corresponding to the user identification information is acquired.

  Further, in the present invention, the history analysis unit acquires and analyzes text data corresponding to each video represented by the viewing information acquired by the history acquisition unit, and each text included in the character string represented by each text data Identify words. The updating unit updates the intimacy associated with the word in the intimacy information stored in the intimacy database so that the degree of recognition of each of the identified words is increased.

  In other words, if the voice output in accordance with the video contains a word with a low recognition level (ie, familiarity), if the time length for uttering the word is short, the person who heard the voice will be There is a possibility that the content of information represented by speech cannot be recognized.

  Therefore, in the information processing apparatus of the present invention, the utterance time to be applied to the word is set so that the word representing lower intimacy has a longer proportion of the utterance time of the word in the total utterance time of the information. Adjusted speech speed data is generated.

  If the output speed of the synthesized speech is determined based on such speech speed data, in the synthesized speech, the ratio of the time length for uttering words with low familiarity in the total utterance time of information can be increased.

As a result, a person who has listened to the synthesized speech can easily hear even a word with low familiarity, and can recognize the entire content of the information represented by the utterance.
In addition, in the information processing apparatus of the present invention, the familiarity information is updated by analyzing text data corresponding to videos viewed by the user in the past.

In this way, each word in the audio in the video viewed by the user in the past is likely to be recognized by the user.
Therefore, according to the information processing apparatus of the present invention, it is possible to use intimacy information that matches the recognition state of words for each user, and it is possible to generate more appropriate speech speed data for the user.

In other words, in the information processing apparatus of the present invention, the speech speed can be adjusted in the synthesized speech so that the content of the utterance can be easily understood.
The utterance time referred to here represents the time required for utterance and includes speed (speech speed).

  By the way, the information processing apparatus of the present invention includes word specifying means for specifying an important word that is a word corresponding to an important part of speech that is defined in advance as a part of speech with high importance from the words specified by the analyzing means. Also good.

In this case, the speech speed determining means in the present invention may generate the speech speed data so that the vowel utterance time included in the important word specified by the word specifying means becomes longer.
According to the information processing apparatus of the present invention, speech speed data can be generated so that the utterance time for an important Japanese word is prolonged.

  Then, the synthesized speech whose speech speed is adjusted based on the speech speed data generated by the information processing apparatus of the present invention can make it easier to hear important words and understand the content of the utterance more easily. .

Furthermore, the word specifying means in the present invention may specify at least one of a noun and a verb as an important part of speech, and specify a word corresponding to each of the important parts of speech as an important word.
In information output by voice, nouns and verbs have large weights.

For this reason, in the present invention, at least one of a noun and a verb may be an important part of speech, and a word corresponding to each important part of speech may be specified as an important word.
According to such an information processing apparatus, speech speed data can be generated so that the utterance time for at least one of a noun and a verb becomes longer.

The synthesized speech whose speech speed is adjusted based on the speech speed data generated by the information processing apparatus of the present invention can make it easier to hear at least one of a noun and a verb.
In addition, the updating unit according to the present invention corresponds to the word in the familiarity information so that the familiarity at the timing when the word appears increases as the number of times the word specified by the history analyzing unit increases. The attached intimacy may be updated.

According to such an information processing device, a word having a greater number of appearances throughout the video can be made more intimate and speech speed data suitable for the video can be generated.
In the present invention, based on the speech speed data generated by the speech speed determining means, the speech synthesizing means becomes the utterance time represented by the speech speed data for each phoneme constituting each word. In this way, the voice may be synthesized and output.

According to such an information processing apparatus, it is possible to output synthesized speech in which the speech speed is adjusted so that the content of the utterance can be easily understood.
Note that each of the text data of the present invention may include a required utterance time defined in advance as a time length that can be applied to the utterance of the character string represented by the text data.

  In this case, the speech speed determination means of the present invention may generate normalized data as speech speed data so that the utterance time of the entire information represented by the text data is maintained at the required utterance time.

According to such an information processing apparatus, since the time length required for uttering the content of information is not changed, it is possible to utter at an appropriate timing along the progress of the video.
By the way, the present invention may be implemented as a speech speed data generation method for generating speech speed data.

  The speech speed data generation method of the present invention includes a text acquisition process for acquiring text data, an analysis process for specifying each word included in a character string represented by the acquired text data, and each of the specified words. The familiarity acquisition process of acquiring the corresponding familiarity, and the word representing that the acquired familiarity is low so that the proportion of the utterance time of the word in the utterance time of the entire information becomes longer A speech speed determination process for generating speech speed data in which the utterance time of the word is adjusted. Furthermore, the speech speed data generation method of the present invention includes an identification information acquisition process for acquiring user identification information, a history acquisition process for acquiring user viewing information corresponding to the acquired user identification information, and the acquisition Text data corresponding to each video represented by the viewing information obtained and analyzed, a history analysis process for identifying each word included in the character string represented by each text data, and each of the identified words An update process for updating the intimacy associated with the word in the intimacy information stored in the intimacy database so that the degree of recognition increases.

With such a speech speed data generation method, the same effect as the information processing apparatus of the present invention can be obtained.
Further, the present invention may be made as a program executed by a computer.

  In the program of the present invention, a text acquisition procedure for acquiring text data, an analysis procedure for specifying each word included in the character string represented by the text data, and a closeness corresponding to each specified word are acquired. The utterance time of the word so that the proportion of the utterance time of the word occupies the utterance time of the entire information becomes longer for the word indicating that the acquired intimacy is lower The computer executes a speech speed determination procedure for generating the spoken speed data.

  Furthermore, in the program of the present invention, an identification information acquisition procedure for acquiring user identification information, a history acquisition procedure for acquiring user viewing information corresponding to the acquired user identification information, and the acquired viewing information The history analysis procedure for identifying each word included in the character string represented by the text data corresponding to each represented image and the degree of recognition of each identified word is stored in the familiarity database. The computer is caused to execute an update procedure for updating the familiarity associated with the word in the familiarity information.

For example, if the present invention is implemented as a program, it can be used by being loaded from a recording medium into a computer as needed and being activated, or by being obtained and activated by a computer via a communication line as necessary. it can. Then, by causing the computer to execute each procedure, the computer can function as the information processing apparatus of the present invention .

  The recording medium referred to here includes, for example, a computer-readable electronic medium such as a DVD-ROM, a CD-ROM, and a hard disk.

1 is a block diagram illustrating an information processing apparatus to which the present invention is applied and a schematic configuration around the information processing apparatus. It is explanatory drawing explaining the structure of text data. It is a flowchart which shows the process sequence of a speech speed data generation process. It is explanatory drawing explaining the information produced | generated in the process of a speech speed data production | generation process. It is a flowchart which shows the process sequence of a closeness update process.

Embodiments of the present invention will be described below with reference to the drawings.
<Content viewing system>
A content viewing system 1 shown in FIG. 1 is a system in which a user views content prepared in advance, and includes an information processing server 10 and at least one information processing device 30.
<Information processing server>
The information processing server 10 is a server that stores various data, and includes a communication unit 12, a control unit 14, and a storage unit 22.

  The various data stored in the information processing server 10 includes at least content data CD including video and audio to be output, sound source data SV including at least audio feature values of audio input in advance, and a content viewing system. User history data HD relating to the history of content viewed by one user, and word familiarity data DD in which familiarity indicating the recognition degree of each word is associated with each word.

In the communication unit 12, the information processing server 10 communicates with the outside through a communication network. The communication network in this embodiment is, for example, a public wireless communication network or a network line.
The control unit 14 is configured around a known computer having at least a ROM 16, a RAM 18, and a CPU 20, and controls the communication unit 12 and the storage unit 22.

  The ROM 16 stores processing programs and data that need to retain stored contents even when the power is turned off. The RAM 18 temporarily stores processing programs and data. The CPU 20 executes various processes according to the processing program stored in the ROM 16 or the RAM 18.

  The storage unit 22 is a non-volatile storage device configured to be able to read and write stored contents. The storage device is, for example, a hard disk device or a flash memory. The storage unit 22 stores content data CD, sound source data SV, user history data HD, and word familiarity data DD.

Among these, the content data CD is data prepared in advance for each content.
The content mentioned here is a product in which at least an image (video) and audio are output along the time axis. As an example of this product, a movie or a TV program can be considered.

  This content data CD includes video data IM, speech audio data SD, and speech text data TD. A code “m” in FIG. 1 is a code for identifying each content data CD.

The video data IM is data composed of a plurality of images constituting a video (moving image) output in the content.
The speech audio data SD is audio data that is output in accordance with the video represented by the video data IM. The speech audio data SD is, for example, speech or narration that is emitted in accordance with the video. The speech audio data SD in the present embodiment may be prepared for each speech or narration in the video, or may be prepared for each unit section defined in advance along the time axis in the video.

  The serif text data TD is text data representing the content of audio output in accordance with the video represented by the video data IM. As shown in FIG. 2, the serif text data TD includes casting information, caption information, and timing information.

  Among these, the caption information is a caption (text) output in accordance with the video. This subtitle is a character string representing contents such as lines and narration. Further, the subtitle language in the present embodiment is Japanese.

  The casting information is information for identifying a person who should read out each caption, and is defined for each caption. This casting information may be information that identifies the person itself, or information that represents the characteristics of the person such as gender and age.

  The timing information is information in which each of the subtitles includes a start timing in which the timing for outputting the subtitle represented by the subtitle information is defined and an end timing in which the output is terminated. These start timing and end timing are associated with the progress of time in the video data IM.

  Further, the timing information includes a required utterance time defined as a time length that can be spent reading out the entire character string represented by the subtitle information included in the serif text data TD.

Note that the serif text data TD in this embodiment is prepared for each subtitle output in accordance with the video.
The sound source data SV is data in which sound parameters and tag data are associated with each sound source. The voice parameter is at least one feature amount representing a waveform of a sound emitted by a person. This feature amount is a feature amount of speech used for so-called formant synthesis, and is prepared for each speaker and for each phoneme. As a feature value in the speech parameter, at least a fundamental frequency F0, a mel frequency cepstrum (MFCC), a phoneme length, a power, and a time difference thereof in each phoneme in the uttered speech are provided.

  The tag data is data representing the nature of the sound represented by the speech parameters, and includes at least speaker feature data representing the features of the speaker. The speaker feature data includes, for example, the sex and age of the speaker.

  Further, the tag data may include facial expression data representing the facial expression of the speaker when the voice is uttered. This facial expression data is data representing a concept as a facial expression including at least emotions, emotions, scenes, and situations, and may include information necessary for estimating the expression of the speaker.

  The sound source data SV in which these voice parameters and tag data are associated with each other is generated, for example, by executing a process defined in advance in the karaoke device when a song is sung using a known karaoke device. And may be registered in the storage unit 22.

In addition, the code “n” in FIG. 1 is a code for identifying each sound source data SV.
The user history data HD is data representing the history of content viewed by the user of the content viewing system 1. The code “L” in FIG. 1 is a code for identifying each user history data HD.

  The user history data HD is data in which a user ID for identifying a user and a content ID for identifying each content viewed by the user are associated with each user.

  In this user history data HD, each time a content is browsed, the content ID of the content may be associated with the user ID, assuming that the content has been viewed. Note that the timing at which the content ID is associated with the user ID is not limited to the timing at which the content is browsed, but may be the timing at which the content is purchased.

  The word familiarity data DD is data in which each word is associated with a familiarity representing the recognition degree of each word in advance. The familiarity here is a larger value as the recognition degree is higher. That is, the word familiarity data DD is an example of the familiarity information described in the claims.

Note that the word familiarity data DD in the present embodiment may store the recognition degree of each word for each user. In the present embodiment, the storage unit 22 in which the word familiarity data DD is stored functions as a familiarity database.
<Information processing device>
The information processing apparatus 30 includes a communication unit 31, an input reception unit 32, a display unit 33, a sound input unit 34, a sound output unit 35, a storage unit 36, and a control unit 40.

  As the information processing apparatus 30 in the present embodiment, for example, a known portable terminal may be assumed, or a known information processing apparatus such as a so-called personal computer may be assumed. Note that portable terminals include well-known electronic book terminals, and portable information terminals such as mobile phones and tablet terminals.

  The communication unit 31 performs information communication with the outside via a communication network. The input receiving unit 32 receives information input via an input device (not shown). The display unit 33 displays an image based on a signal from the control unit 40.

  The sound input unit 34 is a device that converts sound into an electric signal and inputs the electric signal to the control unit 40, and is, for example, a microphone. The sound output unit 35 is a known device that outputs sound, and includes, for example, a PCM sound source and a speaker. The storage unit 36 is a non-volatile storage device configured to be able to read and write stored contents. The storage unit 36 stores various processing programs and various data.

  The control unit 40 is configured around a known computer having at least a ROM 41, a RAM 42, and a CPU 43. The ROM 41 stores processing programs and data that need to retain stored contents even when the power is turned off. The RAM 42 temporarily stores processing programs and data. The CPU 43 executes various processes according to the processing programs stored in the ROM 41 and the RAM 42.

  That is, the information processing apparatus 30 displays the video in the designated content on the display unit 33 based on the content data CD corresponding to the designated content, and outputs the sound from the sound output unit 35 in accordance with the time axis in the video. . The designated content referred to here is content designated by information received by the input receiving unit 32.

  When the information processing apparatus 30 outputs sound in the designated content, the Japanese subtitles (text) represented by the serif text data CD are synthesized using the sound source data SV stored in the information processing server 10. To output synthesized speech. That is, the information processing apparatus 30 according to the present embodiment is configured to be able to execute voice-over.

The ROM 41 of the information processing apparatus 30 stores a processing program for the control unit 40 to execute speech speed data generation processing for generating speech speed data representing the speech time of synthesized speech output by speech synthesis. .
<Speech speed data generation processing>
The speech speed data generation process executed by the control unit 40 of the information processing apparatus 30 is activated when an activation command is input.

  In the speech speed data generation process, as shown in FIG. 3, when activated, the control unit 40 first acquires Japanese text text data CD of designated content (S110). Subsequently, the control unit 40 performs morphological analysis on the text represented by the serif text data CD acquired in S110, and derives morpheme information (S120). A well-known method (for example, “MeCab”) may be used as the method of morphological analysis in S120.

The morpheme information includes a morpheme mo (k), a morpheme phoneme number ph_nu (k), a phoneme ph (k, j), and a part of speech flag pa (k).
Among these, the morpheme mo (k) is each morpheme mo included in the text represented by the serif text data CD. The code “k” is an index number for identifying each morpheme mo included in the text, and is assigned in order along the time axis in the serif text data CD.

  The phoneme ph (k, j) is each phoneme constituting each morpheme mo (k). The code “j” is an index number for identifying each phoneme included in each morpheme mo (k), and is assigned along the time axis in the text. The morpheme phoneme number ph_nu (k) is the number of phonemes ph constituting each morpheme mo (k).

  Further, the part of speech flag pa (k) indicates whether or not the part of speech corresponding to each morpheme mo (k) (word) is a noun or a verb. The part-of-speech flag pa (k) is set to “1” if the part-of-speech is a noun or a verb, and is set to “0” if the part-of-speech is a noun or a verb.

  For example, when the text represented by the serif text data CD is “Tomorrow is sunny”, the morpheme mo (k) shown in FIG. 4 (tomorrow / has / Morphological information including the phoneme ph (k, j) (asu / wa / hall / desne in the figure) is derived.

  Further, in the speech speed data generation process, the control unit 40 determines the familiarity corresponding to each morpheme (word) mo (k) included in each morpheme information derived in S120 from the storage unit 22 of the information processing server 10. Obtain (S130).

  Subsequently, in the speech speed data generation process, the control unit 40 determines whether each phoneme ph (k, j) is a vowel, and sets a vowel flag vw (k, j) (S140). In S140, specifically, as shown in FIG. 4, if the phoneme ph (k, j) in each morpheme mo (k) is a vowel, the vowel flag vw (k, j) is set to “1”. If the phoneme ph (k, j) is a consonant, the vowel flag vw (k, j) is set to “0”.

  Further, in the speech speed data generation process, the control unit 40 sets an initial value of the phoneme length ratio Ph_lr (k, j) (S150). The phoneme length ratio Ph_lr (k, j) referred to here is the reading of each phoneme ph (k, j) in the time length (speech time length) necessary for reading the entire text represented by the serif text data CD. It is the ratio of the time length required for

  In S150 in the present embodiment, specifically, if the phoneme ph (k, j) is a vowel, the initial value of the phoneme length ratio ph_lr (k, j) is set to “1”, and the phoneme ph (k, k, j) is set. If j) is a consonant, the initial value of the phoneme length ratio ph_lr (k, j) is set to the “specified value p”. The specified value p in the present embodiment is a value specified in advance, and is a value that is larger than “0” and smaller than “1”.

  Subsequently, in the speech speed data generation process, the control unit 40 specifies an important word from each morpheme mo (k) (word) derived in S120 based on the part of speech flag included in the morpheme information (S160). . The important word here is a word corresponding to an important part of speech that is defined in advance as a part of speech having a high degree of importance. The important parts of speech in this embodiment include verbs and nouns.

  And the control part 40 of phoneme ph (k, j) corresponding to a vowel in each phoneme ph (k, j) which comprises each morpheme mo (k) identified as an important word in S160. The phoneme length ratio Ph_lr (k, j) is updated (S170). The update in S170 is executed according to the following equation (1), and only the phoneme length ratio Ph_lr (k, j) of the phoneme ph (k, j) corresponding to the vowel included in the important word is lengthened. In the equation (1), α is a constant defined in advance.

That is, in S170 of this embodiment, the time length for uttering the phoneme ph (k, j) whose part-of-speech flag pa (k) is “1” and whose vowel flag vw (k, j) is “1”. Is multiplied by “1 + α / 100”.

  Further, in the speech speed data generation process, the control unit 40 first acquires the familiarity of each morpheme mo (k) from the information processing server 10 and normalizes the familiarity nr_fa (k) based on the acquired familiarity. Is calculated (S180). This normalized familiarity nr_fa (k) has normalized the familiarity of each morpheme mo (k) so that the average of the familiarity for each morpheme mo (k) is “1” and the variance is “1”. Is.

  In S180, the control unit 40 further calculates the magnification β (k) according to the following equation (2), and the phoneme length ratio Ph_lr (k, j) included in each morpheme according to the following equation (3): ) Is corrected.

That is to say, in S180, the phoneme length ratio Ph_lr (k, j) of a word representing a low familiarity increases the ratio of the time required for reading the word to the time required for reading the entire information. It is corrected as follows.

  Subsequently, in the speech speed data generation process, the control unit 40 keeps the phoneme time of each phoneme ph (k, j) so that the utterance time of the entire text represented by the serif text data CD is maintained at the required utterance time. The length Ph_le (k, j) is derived (S190).

  Specifically, the derivation of the phoneme time length Ph_le (k, j) of each phoneme ph (k, j) in S190 of the present embodiment is executed according to the following equation (4).

The denominator in the equation (4) is a value (total) obtained by integrating all phoneme ph (k, j) phoneme length ratios Ph_lr (k, j) included in the serif text data CD. The code “tol” in the equation (4) is a required utterance time. Further, the symbol “N” in the equation (4) is the number of phonemes ph included in the serif text data CD.

  That is, the phoneme time length Ph_le (k, j) is normalized so that the total time length for reading the subtitles represented by the serif text data CD is maintained at the required utterance time in the serif text data CD.

  Subsequently, in the speech speed data generation process, the control unit 40 uses the phoneme time length Ph_le (k, j) derived in S190 as the phoneme ph (k, j) constituting each morpheme mo (k). Spoken speed data defined as data representing the read-out timing is generated (S200).

  Further, in the speech speed data generation process, the control unit 40 acquires sound source data SV that best matches each casting information based on each casting information included in the speech text data CD obtained in S110 (S210). ).

  In the speech speed data generation process, the control unit 40 uses the sound source data SV acquired in S210 to synthesize the content of the subtitle information included in the speech text data CD acquired in S110 (S220). ). In S220 of the present embodiment, the reading timing (speed) of each phoneme constituting the text represented by the caption information is determined based on the speech speed data generated in S200.

In S <b> 220 of this embodiment, the control unit 40 outputs a control signal to the sound output unit 35, and outputs a synthesized speech generated by speech synthesis from the sound output unit 35.
Thereafter, the present speech speed data generation process is terminated. Then, in accordance with the timing at which the next video data IM is output along the time axis, the speech speed data generation process is started, and the next serif text data TD along the time axis of the video data IM is acquired ( S110). Thereafter, S120 to S220 are executed.

  That is, in the speech speed data generation process of the present embodiment, the serif text data TD of the specified content is acquired, and the acquired serif text data TD is morphologically analyzed. Then, based on the word familiarity data stored in the information processing server 10, the familiarity is specified for each morpheme (word) specified in the morphological analysis.

Furthermore, in the speech speed data generation process, the speech speed data is generated so that the word indicating that the familiarity is low, the ratio of the time required for reading the word to the time required for reading the entire information becomes longer. doing.
<Intimacy update processing>
A familiarity update process executed by the control unit 14 of the information processing server 10 will be described.

This closeness update process is activated in synchronization with the activation timing of the speech speed data generation process.
In this familiarity update process, when started, as shown in FIG. 5, first, the control unit 14 acquires a user ID input via the input receiving unit 32 of the information processing apparatus 30 (S310). .

  Subsequently, in the familiarity update process, the control unit 14 acquires all content IDs associated with the user ID acquired in S310 in the user history data HD (S320).

  Further, in the familiarity update process, the control unit 14 acquires all the Japanese text text data TD corresponding to each content ID acquired in S320 (S330).

  Subsequently, in the familiarity update process, the control unit 14 performs morphological analysis on the text represented by each of the serif text data TD acquired in S330, and derives morpheme information (S340). A well-known method (for example, “MeCab”) may be used as the morphological analysis method in S340. The morpheme information here includes at least morpheme mo (k) (word).

  In the familiarity update process, the control unit 14 updates the word familiarity data DD based on the morpheme mo (k) derived in S340 (S350). Specifically, in S350 of the present embodiment, the number of appearances is counted for each morpheme mo having the same content, and the word familiarity data DD is updated so that the morpheme mo (word) with the larger number of appearances has a higher familiarity. To do.

  The update of the familiarity may be realized by adding a value obtained by multiplying the number of appearances by a predetermined coefficient to the familiarity before the update. In addition, the update of the intimacy may be performed on the morpheme mo whose part of speech is an independent word, and the attached word may be excluded.

Thereafter, the closeness update process is terminated.
In other words, in the closeness update process of the present embodiment, the control unit 14 is stored in the storage unit 22 so that the closeness of the morpheme mo (word) that appears more frequently through the content viewed by the user is higher. Update word familiarity data DD.
[Effect of the embodiment]
As described above, in the speech speed data generation process according to the present embodiment, the speech speed data is generated so that the lower the familiarity of the word, the longer the ratio of the reading time of the word in the total reading time is. .

  This is because if the time required to read a word with a low recognition level (ie, intimacy) is short, a person who listens to the sound output in accordance with the video may not be able to recognize the content of the information based on the sound. Because there is.

  That is, if the start timing of each phoneme in the synthesized speech is determined based on the speech rate data generated by the speech rate data generation process of the present embodiment, the familiarity that occupies the total reading time of information in the synthesized speech The ratio of the time length required to read out words with low can be increased.

As a result, even if the word has a low familiarity, it is easy for a person who has listened to the synthesized speech to hear, and the person can recognize the entire content of the information represented by the utterance.
In other words, the information processing apparatus 30 can adjust the reading speed (that is, speaking speed) so that the content of the utterance can be easily understood in the synthesized speech.

  By the way, normally, in information expressed in Japanese speech, nouns and verbs have large weights. For this reason, in the speech speed data generation processing according to the present embodiment, the noun and the verb are important parts of speech, and the speech speed data is generated so that the reading time for the important words corresponding to each of the important parts of speech becomes long.

According to the synthesized speech in which the speech speed is adjusted based on the speech speed data generated in this way, it is possible to make it easy to listen to important parts of speech and to understand the content of the utterance more easily.
Also, in the speech speed data generation process of the present embodiment, the data normalized so that the time length required to read out the entire information represented by one serif text data CD is maintained at the utterance time required. It is generated as speed data.

  For this reason, according to the speech speed data generation process, it is possible to prevent the time length for reading out the subtitle from being changed from a predetermined time length, and to read out the subtitle at an appropriate timing according to the progress of the video.

  In the present embodiment, in the familiarity update process, the familiarity corresponding to the word is increased so that the familiarity is increased according to the number of times the word included in the caption of the content that the user has viewed has appeared. Has been updated.

According to such a familiarity update process, the familiarity can be increased as the number of appearances in the content viewed by the user increases. As a result, according to the content viewing system 1, word familiarity data reflecting the recognition degree for each user can be generated, and speech speed data according to the user's knowledge can be generated.
[Other Embodiments]
As mentioned above, although embodiment of this invention was described, this invention is not limited to the said embodiment, In the range which does not deviate from the summary of this invention, it is possible to implement in various aspects.

For example, in the speech speed data generation process of the above embodiment, both nouns and verbs are important parts of speech, but the important parts of speech may be at least one of nouns and verbs.
Moreover, in the said embodiment, although the control part 40 of the information processing apparatus 30 performed speech speed data generation processing, the apparatus which performs speech speed data generation processing is not restricted to the information processing apparatus 30, and information The processing server 10 may be used.

  In this case, the information processing apparatus 30 may acquire the speech speed data from the information processing server 10 and determine the speech speed when executing speech synthesis that reads out the caption based on the serif text data TD.

  Moreover, in the said embodiment, although the information processing server 10 performed the closeness update process, the apparatus which performs a closeness update process is not restricted to the information processing server 10, and is the information processing apparatus 30. Also good.

In addition, the aspect which abbreviate | omitted a part of structure of the said embodiment as long as the subject could be solved is also embodiment of this invention. Further, an aspect configured by appropriately combining the above embodiment and the modification is also an embodiment of the present invention. Moreover, all the aspects which can be considered in the limit which does not deviate from the essence of the invention specified by the wording described in the claims are the embodiments of the present invention.
[Correspondence between Embodiment and Claims]
Finally, the relationship between the description of the above embodiment and the description of the scope of claims will be described.

  The function obtained by executing S110 of the speech speed data generation process corresponds to the text acquisition means in the claims, and the function obtained by executing S120 of the speech speed data generation process is the analysis means. It corresponds to. Further, the function obtained by executing S130 of the speech speed data generation process corresponds to the familiarity acquisition means in the claims, and can be obtained by executing S140 to S200 of the speech speed data generation process. The function corresponds to speech speed determining means.

  The function obtained by executing S310 of the intimacy update process corresponds to the identification information acquisition means in the claims, and the function obtained by executing S320 of the intimacy update process Corresponds to acquisition means. The function obtained by executing S330 and S340 of the familiarity update process corresponds to the history analysis means in the claims, and the function obtained by executing S350 corresponds to the update means.

  Further, the function obtained by executing S160 of the speech speed data generation process corresponds to the word specifying means in the claims, and the function obtained by executing S210 and S220 of the speech speed data generation process. Corresponds to speech synthesis means.

  DESCRIPTION OF SYMBOLS 1 ... Content viewing system 10 ... Information processing server 12 ... Communication part 14 ... Control part 16 ... ROM 18 ... RAM 20 ... CPU 22 ... Memory | storage part 22 ... Memory | storage device 30 ... Information processing apparatus 31 ... Communication part 32 ... Input reception part 33 ... Display unit 34 ... Sound input unit 35 ... Sound output unit 36 ... Storage unit 40 ... Control unit 41 ... ROM 42 ... RAM 43 ... CPU

Claims (8)

Text acquisition means for acquiring text data representing a character string of information output by sound according to video;
Analyzing the text data acquired by the text acquisition means, and specifying each word included in the character string represented by the text data;
A parent that acquires a familiarity corresponding to each word specified by the analysis means from a familiarity database in which familiarity information in which each word and a familiarity representing the recognition degree of each word are associated in advance is stored. Density acquisition means;
Used for speech synthesis, a data representative of the utterance time of synthesized speech, and data representing the utterance time of each phoneme constituting the string of information represented by the text data and speech speed data, the intimacy degree obtaining The utterance time of the word is set so that the proportion of the utterance time of the word that occupies the utterance time of the entire information represented by the text data becomes longer as the word indicating that the familiarity acquired by the means is lower. Speaking speed determining means for generating the adjusted speaking speed data;
Identification information acquisition means for acquiring user identification information for identifying a user;
From the usage history in which each of the user identification information is associated with viewing information representing videos viewed in the past by the user corresponding to each user identification information, the user identification information acquired by the identification information acquisition unit is used. History acquisition means for acquiring viewing information of corresponding users;
A history analysis unit that acquires and analyzes text data corresponding to each video represented by the viewing information acquired by the history acquisition unit, and identifies each word included in a character string represented by each text data;
Updating means for updating the intimacy associated with the word in the intimacy information stored in the intimacy database so that the degree of recognition of each word specified by the history analysis means is increased. An information processing apparatus characterized by the above. Among the words specified by the analysis means, comprising word specifying means for specifying an important word that is a word corresponding to an important part of speech defined in advance as a part of speech with high importance,
The speech speed determining means is
The information processing apparatus according to claim 1, wherein the speech speed data is generated so that a vowel utterance time included in an important word specified by the word specifying unit becomes longer. The word specifying means is
The information processing apparatus according to claim 2, wherein at least one of a noun and a verb is the important part of speech, and a word corresponding to each of the important parts of speech is specified as the important word. The updating means includes
The familiarity associated with the word in the familiarity information so that the familiarity at the timing of the appearance of the word increases as the number of occurrences of the word specified by the history analysis unit increases. The information processing apparatus according to any one of claims 1 to 3, wherein the information processing apparatus is updated. Based on the speech speed data generated by the speech speed determining means, speech synthesis is performed by synthesizing and outputting the speech time of each phoneme constituting each word to be the speech time represented by the speech speed data. The information processing apparatus according to claim 1, further comprising: means. Each of the text data includes a required utterance time defined in advance as a time length that can be applied to the utterance of the character string represented by the text data,
The speech speed determining means is
The data normalized so that the utterance time of the entire information represented by the text data is maintained at the required utterance time is generated as the speech speed data. The information processing apparatus according to any one of claims. A text acquisition process for acquiring text data representing a character string of information output by sound according to a video,
Analyzing the text data acquired in the text acquisition process, identifying each word included in the character string represented by the text data; and
A parent that acquires a familiarity corresponding to each word specified in the analysis process from a familiarity database that stores familiarity information in which each word and a familiarity representing the recognition degree of each word are associated in advance. Density acquisition process,
Used for speech synthesis, a data representative of the utterance time of synthesized speech, and data representing the utterance time of each phoneme constituting the string of information represented by the text data and speech speed data, the intimacy degree obtaining The utterance time of the word is set so that the word representing the intimacy acquired in the process has a longer proportion of the utterance time of the word in the utterance time of the entire information represented by the text data. A speech speed determination process for generating the adjusted speech speed data;
An identification information acquisition process for acquiring user identification information for identifying a user;
From the usage history in which each of the user identification information is associated with viewing information representing a video viewed by the user corresponding to each user identification information in the past, the user identification information acquired in the identification information acquisition process History acquisition process to acquire viewing information of the corresponding user,
Obtaining and analyzing text data corresponding to each video represented by the viewing information obtained in the history obtaining process, and identifying each word included in the character string represented by each text data;
An update process for updating the intimacy associated with the word in the intimacy information stored in the intimacy database so that the degree of recognition of each word specified in the history analysis process is increased. Speech speed data generation method characterized by the above. A text acquisition procedure for acquiring text data representing a character string of information output by sound according to a video,
Analyzing the text data acquired in the text acquisition procedure, and specifying each word included in the character string represented by the text data; and
A parent that acquires a familiarity corresponding to each word specified in the analysis procedure from a familiarity database in which familiarity information in which each word and a familiarity representing the recognition degree of each word are associated in advance is stored. Density acquisition procedure;
Used for speech synthesis, a data representative of the utterance time of synthesized speech, and data representing the utterance time of each phoneme constituting the string of information represented by the text data and speech speed data, the intimacy degree obtaining The utterance time of the word is set so that the word representing the low intimacy acquired in the procedure becomes longer in the utterance time of the word in the utterance time of the entire information represented by the text data. A speech speed determination procedure for generating the adjusted speech speed data;
An identification information acquisition procedure for acquiring user identification information for identifying a user;
From the usage history in which each of the user identification information is associated with viewing information representing a video viewed by the user corresponding to each user identification information in the past, the user identification information acquired in the identification information acquisition procedure History acquisition procedure to acquire viewing information of the corresponding user,
A history analysis procedure for acquiring and analyzing text data corresponding to each video represented by the viewing information acquired in the history acquisition procedure, and identifying each word included in a character string represented by each text data;
An update procedure for updating the intimacy associated with the word in the intimacy information stored in the intimacy database so that the recognition degree of each word specified in the history analysis procedure is increased. A program characterized by being executed.