JP2006010845A - Synthesized speech uttering device and program thereof, and data set generating device for speech synthesis, and program thereof - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a data receiving device which easily switches speech expressions of the voice quality, intonation, etc., of a synthesized speech according to user's preference for utterance of synthesized speeches with various speech expressions. <P>SOLUTION: A data receiving device 20 is equipped with a speech synthesizing means 25 of acquiring kind information for dividing elementary speech unit groups by voice quality, elementary speech unit information made to correspond to elementary speech units by the elementary speech unit groups, and control information in which a way of deforming the waveform of a synthesized speech generated by connecting elementary speech units made to correspond on the basis of the kind information and elementary speech unit information, prescribing one elementary speech unit group stored in a speech database 251 stored with the elementary speech unit groups by voice quality with the kind information, extracting elementary speech units from the prescribed elementary speech unit group according to the elementary speech unit information, generating a waveform by mutually connecting the extracted elementary speech units, and deforming the generated waveform according to the control information for utterance of the synthesized speech. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a synthesized speech utterance device, a program thereof, and a program for enabling a synthesized speech of various speech expressions to be uttered by easily switching expressions such as voice quality and intonation of synthesized speech according to user's preference, and The present invention relates to a speech synthesis data set generation apparatus and a program thereof.

Recently, with the start of terrestrial digital broadcasting, various types of data broadcasting related to data transmission in which content is transmitted as text on the transmission device side and the received text is synthesized as speech on the reception device side are synthesized. Technology has been developed.
As a conventional data broadcasting system, various rules for speech synthesis of text transmitted from the transmitting device side are registered in a database on the receiving device side, and selection on the transmitting device side or on the receiving device side A device that utters an arbitrary synthesized voice on the receiving device side by selection by a user is known.

For example, a male voice phoneme dictionary storing a symbol string and a male voice phoneme fragment and a female voice phoneme dictionary storing a symbol string and a female voice phoneme fragment are provided on the receiving device side, and the text transmitted from the transmitting device side Is known as a synthesized voice of a male voice or a female voice by referring to a male voice phoneme dictionary or a female voice phoneme dictionary, and a technique for displaying text (data broadcasting display device) is known (for example, see Patent Document 1).
In the data broadcast display device disclosed in Patent Document 1, a database of a male phonemic phoneme dictionary and a female voice phoneme dictionary is provided on the receiving device side, and a man or a woman is selected by appropriately switching a database to be referred to at the time of speech synthesis. The voice quality of the synthesized speech can be changed by selecting it. The selection can be made either on the transmission device side in advance or on the reception device side when the user sets each time.

Further, in the transmission device (data transmission device), additional information for audio used for speech synthesis on the reception device side is incorporated into the program data and transmitted as a broadcast wave, and on the reception device side, it is incorporated into the program data from the received broadcast wave. There is known a technique (data broadcasting system) that discriminates the additional information for voice and that can perform voice synthesis based on the additional information for voice (for example, refer to Patent Document 2).
In the data broadcasting system disclosed in Patent Document 2, on the receiving device side, it is possible to specify a speech synthesis parameter, a reading speed (speech speed) (five levels), and a reading volume (five levels). However, the number of synthesized speech speakers is limited to those with a preset voice quality.

Japanese Patent Laid-Open No. 2001-16553 (paragraphs 0021 to 0029, FIG. 1) Japanese Patent Laying-Open No. 2002-280981 (paragraph 0060, FIG. 19)

  However, in the data broadcasting display device disclosed in Patent Document 1 and the like, since the male voice dictionary and the female phoneme dictionary are limited to predetermined ones, the synthesized voice is uttered as a male voice or a female voice whose voice quality and intonation are determined. I could only do it. Therefore, in this data broadcasting display device, it was possible to select male voice and female voice according to the user's preference, but since both male voice and female voice are determined in voice quality and intonation, either male voice or female voice is selected. Even in this case, it was far from a voice expression that suits the user's preference.

  In the data broadcast display device disclosed in Patent Document 1 and the like, the voice database of the male voice dictionary and the female voice dictionary is often determined in advance and cannot be changed. In addition, when the contents of the voice database can be updated, the user obtains a semiconductor device storing the updated contents by some method such as going to a store and replaces the semiconductor device himself or When a (Compact Disc) -ROM (Read Only Memory) drive is provided, it is necessary to obtain a CD-ROM storing the update contents, read it, and update it. It was inconvenient.

  In the data broadcasting system disclosed in Patent Document 2 and the like, the speech database is limited to that of a single speaker. However, since the speech database has parameter setting means for speech synthesis, It was possible to adjust the volume. Therefore, in this data broadcasting system, the speed and volume of the synthesized voice can be adjusted according to the user's preference, but the synthesized voice with different voice quality and intonation cannot be uttered according to the user's preference, and the user's preference is met. It was not possible to utter the appropriate voice expression.

  Furthermore, even with a technology that simply combines the data broadcast display device disclosed in Patent Document 1 and the like and the data broadcast system disclosed in Patent Document 2 or the like, it is possible to obtain a male voice and a female voice whose voice quality and intonation are determined. Since it is only possible to select and adjust the speed and volume of the synthesized voice of the male voice or female voice, the synthesized voice cannot be uttered with a voice quality or intonation voice expression that suits the user's preference.

  The present invention has been made to solve the above-described problems, and can synthesize synthesized voices of various voice expressions by easily switching voice expressions such as voice quality and intonation of synthesized voices according to user's preference. It is an object of the present invention to provide a synthesized speech utterance device and program thereof, and a speech synthesis data set generation device and program thereof.

  The present invention was created to achieve the above object, and first, the synthesized speech utterance device according to claim 1 is a plurality of speech of the same voice quality constituting a minimum unit waveform to be uttered as synthesized speech. Based on the type information for dividing the speech unit group composed of the units for each voice quality, the unit information associated with the speech unit for each of the speech unit groups, and the type information and the unit information The speech synthesis from the speech synthesis data set generation device that generates at least one type of control information defining a waveform deformation method of the synthesized speech generated by connecting the speech units to be associated with each other as a speech synthesis data set A speech database, a speech synthesis data set acquisition means, a speech synthesis control information setting means, a speech synthesis It was configured and means.

According to this configuration, the synthesized speech utterance apparatus stores each voice quality by the speech database. Here, the speech unit group is a group of speech units constituting the same voice quality and stored in the speech database, and is defined by classification for different types of voice quality. For example, as will be described later, the speech database stores a speech segment group defined by content and speaker differences so that it can be distinguished for each type.
The synthesized speech utterance apparatus acquires the speech synthesis data set from the speech synthesis data set generation apparatus by the speech synthesis data set acquisition means.

  Next, the synthesized speech utterance device receives the speech synthesis data set acquired by the speech synthesis data set acquisition unit by the speech synthesis control information setting unit, and selects and extracts the type information of the speech synthesis data set. Identify the target speech segment group, select the control information of the speech synthesis data set to identify how to transform the waveform, and set the selected type information, control information, and segment information as speech synthesis data .

  The setting is performed by displaying a screen for selecting which type information and control information to select and allowing the user to specify via the interface. For example, when male voice and female voice are given as the segment information, when the user specifies a female voice via the interface, the voice segment specified as the segment information from the voice segment group of the female voice is voiced. A female voice segment group is set as selected from the database. Moreover, as control information, a function, a parameter, and a control information number can be mentioned, for example.

After that, the synthesized speech utterance device receives speech synthesis data from the speech synthesis control information setting unit by the speech synthesis unit based on the synthesized speech data, and generates a segment from the speech unit group of the speech database specified by the type information. Extract speech segments according to the information, connect the extracted speech segments to generate a synthesized speech waveform, transform the waveform according to the waveform modification method specified by the selected control information, Speak as synthesized speech.
For example, the speech synthesizer selects a speech unit from the group of speech units of female voices, generates a waveform by connecting the speech units, and when a fast mouth is set as control information, the generated waveform The waveform is processed so as to shorten the period of the voice, and the synthesized voice is uttered with a quick female voice. The voice synthesizing unit sends the generated waveform to the voice output unit, and causes the voice output unit to utter as synthesized voice. Examples of the voice output means include a speaker that outputs a waveform as an electric signal output by the voice synthesis means as voice.

  By the way, as described above, the type information is for dividing the speech element group for each voice quality, and the speech database is for storing the speech element group for each voice quality. Here, since the voice quality types can be made to be uttered as synthesized speech by the user selecting the type according to the user's preference, there are at least two voice quality types selected by the user of the synthesized speech utterance device. Preferably there is. However, only one type of voice quality may be used.

  According to a second aspect of the present invention, in the synthesized voice utterance device according to the first aspect, the voice synthesis control information setting means includes at least one of a content item type setting means and a speaker setting means. The configuration.

  According to this configuration, in the synthesized speech utterance device, the content item type setting means sets the type of the speech element group stored in the speech database according to the type of the content item included in the type information, and The type of the speech element group stored in the speech database is set by the speaker setting means according to the difference of the speakers included in the type information. As a result, it is possible to utter synthetic speech having different voice qualities for each content. Further, it is possible to synthesize synthesized speech having different voice qualities for each speaker. For example, the speaker setting means can be set as the voice quality of a specific character such as a male voice, female voice, a specific actor or actress, or an animation, and uttered as a synthesized voice thereof.

  Also, the synthesized speech utterance device according to claim 3 is the synthesized speech utterance device according to claim 1 or 2, wherein the speech synthesis control information setting means includes a speech speed setting means and dynamic feature control information setting. And at least one of static feature control information setting means.

According to such a configuration, in the synthesized speech utterance apparatus, the control information for changing the speech speed of the waveform generated by connecting the speech units by the speech speed setting means included in the speech synthesis control information setting means. Set speech speed control information.
Further, in the synthesized speech utterance device, the dynamic feature control information setting means provided in the voice synthesis control information setting means has a dynamic feature that changes the movement of the waveform according to the elapsed time of the waveform generated by connecting the speech segments. Dynamic feature control information, which is control information for changing, is set. Here, as the dynamic feature, for example, a fundamental frequency of a waveform, a spectrum, or “interval” in a story or a sense of tempo can be cited.

  Further, in the synthesized speech utterance device, the static feature control information setting unit included in the speech synthesis control information setting unit includes a static feature that changes the shape of a predetermined time at a predetermined time of a waveform generated by connecting speech units. Static feature control information that is variable control information is set. Here, examples of the static feature include a fundamental frequency average value and a spectrum envelope outline.

  According to a fourth aspect of the present invention, there is provided the synthesized speech utterance device according to any one of the first to third aspects, wherein the update information is used to update the speech element group of the speech database. The acquisition unit and the voice database update unit are provided.

  According to this configuration, the synthesized speech utterance apparatus acquires update information for updating the speech element group stored in the speech database from the speech synthesis data set generation apparatus by the update information acquisition unit. Then, the synthesized speech utterance apparatus updates the speech database according to the update information acquired by the update information acquisition unit by the speech database update unit.

  Further, the synthesized speech program according to claim 5 is a type information for dividing a speech unit group composed of a plurality of speech units of the same voice quality constituting a minimum unit waveform to be uttered as synthesized speech, for each voice quality, The unit information associated with the speech unit for each speech unit group, and the waveform of the synthesized speech generated by connecting the speech units associated based on the type information and the unit information In order to obtain the speech synthesis data set from the speech synthesis data set generation device that generates at least one type of control information defining the method of deformation as a speech synthesis data set, It is configured to function as a speech synthesis data set acquisition unit, a speech synthesis control information setting unit, and a speech synthesis unit.

According to this configuration, the synthesized speech utterance program acquires the speech synthesis data set from the speech synthesis data set generation device by the speech synthesis data set acquisition unit.
The synthesized speech utterance program receives the speech synthesis data set acquired by the speech synthesis data set acquisition unit by the speech synthesis control information setting unit, selects type information of the speech synthesis data set, and selects the voice quality Identify the speech unit group to be extracted from the speech database storing the speech unit group for each, select the control information of the speech synthesis data set to identify the waveform deformation method, the selected type information and Control information and segment information are set as data for speech synthesis.

  Thereafter, the synthesized speech utterance program receives the speech synthesis data from the speech synthesis control information setting unit by the speech synthesis unit, and extracts the speech unit according to the unit information from the speech unit group of the speech database specified by the type information. Then, the extracted speech segments are connected to generate a synthesized speech waveform, the waveform is modified according to the waveform modification method specified by the selected control information, and the deformed waveform is uttered as synthesized speech.

  According to a sixth aspect of the present invention, there is provided a speech synthesis data set generation device for dividing a speech unit group composed of a plurality of speech units of the same voice quality constituting a minimum unit waveform to be uttered as synthesized speech, for each voice quality. Type information, unit information associated with the speech unit for each speech unit group, and synthesis generated by connecting the speech units associated based on the type information and the unit information At least one type of control information defining how to transform the waveform of speech is generated as a speech synthesis data set, and the speech synthesis data set is generated by a synthesized speech utterance device that utters speech as synthesized speech based on the speech synthesis data set. Voice database, material text acquisition means, reading information acquisition means, segment selection means, control information creation means, and speech synthesis And configured to include a datasets providing means.

According to this configuration, the speech synthesis data set generation device stores a speech unit group for each voice quality by the speech database.
Then, the speech synthesis data set generation device acquires material text to be uttered as synthesized speech by the material text acquisition means. This text is character information to be uttered by the synthesized speech utterance device.
Next, the speech synthesis data set generation device acquires the reading information of the text acquired by the material text acquisition unit by the reading information acquisition unit.

Further, the speech synthesis data set generation device selects the speech information of the reading information acquired by the reading information acquisition means for each type of voice quality from the speech database by the segment selection means to obtain the segment information. . In addition, the speech synthesis data set generation device generates type information and control information by the control information generation means. The control information creation means displays a screen for prompting input of type information and control information on the display means of the speech synthesis data set generation device, and allows the operator to input type information and control information via the input means. Get type information and control information.
After that, the speech synthesis data set generation device uses the speech synthesis data set providing means to convert the segment information selected by the segment selection means and the type information and control information created by the control information creation means into the speech synthesis data. Provided as a set to the synthesized speech utterance device.

According to a seventh aspect of the present invention, there is provided the speech synthesis data set generation device according to the sixth aspect, wherein the speech synthesis data set generation device includes a voice database update information creation unit and an update information provision unit. .
According to this configuration, the speech synthesis data set generation device creates update information for updating the speech unit stored in the speech database by the speech database update information creation unit, and updates the speech unit in the speech database. Update with.
Then, the speech synthesis data set generation apparatus provides the update information created by the speech database update information creation means to the synthesized speech utterance apparatus by the update information providing means.

  According to another aspect of the present invention, there is provided a speech synthesis data set generation program for dividing a speech unit group composed of a plurality of speech units of the same voice quality constituting a minimum unit waveform to be uttered as synthesized speech for each voice quality. Type information, unit information associated with the speech unit for each speech unit group, and synthesis generated by connecting the speech units associated based on the type information and the unit information At least one type of control information defining how to transform the waveform of speech is generated as a speech synthesis data set, and the speech synthesis data set is generated by a synthesized speech utterance device that utters speech as synthesized speech based on the speech synthesis data set. To provide a material text acquisition unit, a reading information acquisition unit, a segment selection unit, a control information generation unit, and a speech synthesis device. And configured to function as Tasetto providing means.

According to this configuration, the speech synthesis data set generation program acquires the text of the material to be uttered as synthesized speech by the material text acquisition unit.
Next, the speech synthesis data set generation program acquires the reading information of the text acquired by the material text acquisition unit by the reading information acquisition unit.

  Further, the speech synthesis data set generation program reads the speech unit of the reading information acquired by the reading information acquisition unit from the speech database storing the speech unit group for each voice quality by the unit selection unit. Select for each type as segment information. The speech synthesis data set generation program creates type information and control information by the control information creation means. After that, the speech synthesis data set generation program converts the segment information selected by the segment selection unit by the speech synthesis data set providing unit and the type information and control information created by the control information creation unit into the data for speech synthesis. Provided as a set to the synthesized speech utterance device.

According to the first or fifth aspect of the invention, the speech synthesis control information setting means identifies the speech segment group according to the segment information identified by the type information, and stores the same voice quality stored in the speech database. Since the speech unit is extracted from the speech unit group and the synthesized speech waveform is generated by the speech synthesis means, the synthesized speech of a specific voice quality can be uttered and further generated according to the control information. Since the waveform of the synthesized speech is changed, it is possible to finely set the synthesized speech expression such as a fine voice quality setting or an inflection setting. As a result, the user can easily select the voice quality according to his / her preference, finely set the expression of the synthesized voice, and utter the provided content with the favorite synthesized voice.
As described above, by selecting the voice quality and finely setting the expression of the synthesized speech, it is possible to utter the synthesized speech having various tone. Here, the tone of the voice includes, for example, what is called the magnitude of the voice, the strength of the voice, the level of the voice, the inflection of the voice, the color of the voice, the stagnation, the tongue, the clarity of the ending, and the interval.

  According to the second aspect of the invention, the voice quality can be easily switched for each content by the content item type setting means, and the voice quality can be easily switched for each speaker by the speaker setting means. . As a result, the user can easily switch the voice quality by using items that are easy to distinguish conceptually such as contents and speakers. As a pattern for switching the voice quality according to the content, for example, the case where the news is a male voice and the weather forecast is a female voice can be cited. In addition, examples of the voice quality switching pattern by the speaker include voice quality switching by an announcer, an actor, an actress, and an anime character.

  According to the third aspect of the present invention, the speech speed of the synthesized speech can be switched by the speech speed setting means, and the synthesized speech of the speech expression corresponding to the dynamic feature can be uttered by the dynamic feature control information setting means. Moreover, the synthesized voice of the voice expression according to the static feature can be uttered by the static feature control information setting means. As a result, the synthesized speech utterance device can finely set the waveform of the synthesized speech according to the user's preference, so that speech with rich speech expression can be performed.

  According to the fourth aspect of the present invention, it is possible to change, increase or decrease the type of speech segment group in the speech database, or to change, increase or decrease the speech segment. As a result, the synthesized speech utterance apparatus updates the speech database according to the update information provided from the speech synthesis data set generation apparatus, so that it is possible to easily add a speech unit group having a voice quality desired by the user. become. That is, the user can increase his / her favorite voice quality by adding update information of the provided speech segment group without creating the voice quality himself.

  According to the invention described in claim 6 or claim 8, the data set for speech synthesis can be provided to the synthesized speech utterance device, the control information is provided to the synthesized speech utterance device, and the control information is set. The expression of synthesized speech can be adjusted (fine-tuned). Thus, the user of the synthesized speech utterance apparatus can easily utter the desired synthesized speech by setting control information and adjusting (finely adjusting) the expression of the synthesized speech.

  According to the seventh aspect of the invention, the speech synthesis data set generation device creates the update information of the speech database and provides it to the synthesized speech utterance device, thereby updating the speech database in the synthesized speech utterance device. Therefore, it is possible to improve the usability of the user of the synthesized speech utterance device.

  Therefore, according to the present invention, a synthesized speech utterance apparatus capable of easily switching synthesized speech of various speech expressions by expressing expressions such as voice quality and intonation of synthesized speech according to user's preference, and a program thereof, and A speech synthesis data set generation apparatus and a program thereof can be provided.

  Hereinafter, the best mode for carrying out the present invention (hereinafter referred to as “embodiment”) will be described with reference to the drawings. In the following description, the speech synthesis data set generation device is implemented as a data transmission device that transmits a speech synthesis data set and update information on a broadcast wave, and the synthesized speech utterance device is transmitted from the data transmission device. A data broadcasting system implemented as a data receiving apparatus that receives a broadcast wave, acquires a speech synthesis data set and update information, and utters synthesized speech according to the speech synthesis data set will be described as an example.

[Outline of data broadcasting system]
First, with reference to FIG. 1, the outline | summary of the data broadcasting system 1 comprised from the data transmitter 10 and the data receiver 20 is demonstrated. FIG. 1 is a block diagram showing a configuration of a data broadcasting system according to an embodiment of the present invention.
The data broadcasting system 1 receives a data transmitting apparatus 10 that transmits a voice synthesizing data set created from text of a voice synthesizing material, and receives a voice synthesizing data set transmitted from the data transmitting apparatus 10 to generate a voice. Broadcast transmission / reception is performed with the data receiving apparatus 20 that synthesizes and utters as synthesized speech.
Below, the structure of the data transmitter 10 and the data receiver 20 is each demonstrated.

(Configuration of data transmission device)
This data transmitting apparatus 10 includes type information for dividing a speech unit group composed of a plurality of speech units of the same voice quality constituting a minimum unit waveform to be uttered as synthesized speech, and voice for each speech unit group. Segment information associated with the segments, and at least one type of control information defining how to deform the waveform of the synthesized speech generated by connecting speech segments associated with each other based on the type information and the segment information Are generated as a speech synthesis data set, and the speech synthesis data set is provided to the data receiving apparatus 20 that utters speech as synthesized speech based on the speech synthesis data set.

  For this purpose, the data transmission apparatus 10 includes a material text acquisition unit 11, a speech synthesis data set creation unit 12, a speech synthesis data set addition unit 13, a speech database update information creation unit 14, and a multiplexing unit 15. And transmission means 16 are mainly provided.

  The material text acquisition unit 11 receives a data broadcast material stored in a storage unit such as a data broadcast material processing device (not shown), acquires material text to be uttered as synthesized speech from the data broadcast material, and stores the text. This is output to the voice synthesis data set creation means 12 and the data broadcast material is output to the voice synthesis data set addition means 13. Here, the data broadcast material is assumed to be a stream including video data, audio data, and text, and is stored in storage means such as a data broadcast material processing apparatus (not shown). Further, the text is intended for Japanese such as kanji mixed kana sentences, but is not limited thereto, and may be only hiragana or languages such as English and Chinese.

  The speech synthesis data set creation means 12 generates a speech synthesis data set including unit information, type information, and control information. For this purpose, the speech synthesis data set creation unit 12 mainly includes a reading information acquisition unit 121, a segment selection unit 122, a speech database 123, and a control information creation unit 124.

  The speech synthesis data set creation unit 12 obtains reading information for each text, selects a corresponding speech segment from the speech database 123, and creates segment information. Subsequently, the speech synthesis data set creation unit 12 creates a speech synthesis data set in which speech synthesis data sets including type information, segment information, and control information are collected.

  The reading information acquisition unit 121 acquires the reading information of the text acquired by the material text acquisition unit 11. The reading information acquisition unit 121 recognizes characters by referring to various dictionaries such as a grammar dictionary and a language dictionary, analyzes text morphemes by referring to the various dictionaries, and acquires reading information.

FIG. 2 is a diagram for explaining the concept of acquiring reading information by the reading information acquiring unit of the speech synthesis data set creating unit included in the data transmitting apparatus. FIG. 2A shows an example of text. FIG. 2B is a diagram illustrating an example of morphological analysis of the sentence example in FIG. FIG. 2C is a diagram illustrating an example of reading information of the sentence example in FIG.
When the material text obtaining unit 11 obtains the text “Text is” shown in FIG. 2A, the material text obtaining unit 11 sends the sentence “Text is” to the speech synthesis data set creation unit 12. The reading information acquisition unit 121 of the voice synthesis data set creation unit 12 acquires the sentence “text” sent from the material text acquisition unit 11.

  The reading information acquisition unit 121 first performs morphological analysis of the sentence “text” with reference to various dictionaries, and divides it into morpheme 1 “text” and morpheme 2 “is” shown in FIG. . When the morpheme analysis is completed, the reading information acquisition unit 121 acquires the reading information of each of the morphemes 1 and 2. Here, when the reading information of the morpheme 1 is represented by “(silent) tekisuto” and the reading information of the morpheme 2 is represented by “desu (silent)”, the reading information acquisition unit 121 is illustrated in FIG. Reading information “(silent) tekisutodesu (silent)” is sent to the segment selection means 122.

Returning to FIG.
The segment selection unit 122 selects the speech unit of the reading information acquired by the reading information acquisition unit 121 for each type of voice quality from the speech database 123 and sets it as unit information.
The segment selection unit 122 searches the speech database 123 according to the reading information passed from the reading information acquisition unit 121, and selects the speech unit according to the reading information for each type of voice quality such as a defined speaker. 123.
Note that the unit information includes a speech unit number that specifies which type of speech unit to select from the speech unit group stored in the speech database 123. As will be described later, the speech unit number can be defined, for example, by assigning the speech unit “(silent) tekisuto” as a number “No250” corresponding to this speech unit.

  In addition, the speech unit group conceptually means a classification associated with the type information stored by dividing the speech unit for each type of speaker or the like, but the speech unit number as the unit information. By using, the number of speech segments stored in the speech database 123 can be reduced. For example, when a synthesized voice having a voice quality of Mr. A and Mr. B is uttered, the voice quality of a predetermined phrase may be the same even when the voice quality of Mr. A and Mr. B is different as a whole. In such a case, it is possible to call up speech units of the same voice quality each time, rather than storing speech units as many as the number of speakers classified for each speaker in the speech database 123. The amount of data stored in the voice database 123 can be reduced. In addition, it is possible to reduce the amount of data of the speech synthesis data set transmitted from the data transmitting apparatus 10 to the data receiving apparatus 20 on a broadcast wave.

In addition, the data transmission device 10 includes a user interface (not shown) for the content provider to input various commands and various data to the data transmission device 10 and a storage unit (not shown) for storing various data. It has been.
Then, the content provider performs addition / correction / deletion of the speech unit via the speech database update information creation unit 14. At this time, it is instructed via a user interface (not shown) which voice unit group of voice quality is stored in the voice database 123 and stored in the voice database 123.

  The speech database 123 stores a speech unit group composed of a plurality of speech units of the same voice quality for each voice quality, and stores speech units in association with speech unit numbers, for example. Specifically, the speech database 123 stores the speech unit “No250” in association with the speech unit “(silent) tekisuto”. Further, the voice database 123 defines and stores a voice element group for each speaker in order to specify voice quality. The speech unit group can be defined as a table in which speech units are associated with the speech unit numbers described above.

  Therefore, here, for example, the speech database 123 is a database constructed for each table (speech unit group) in which speech unit numbers and speech units are associated with each other. If a database is constructed for each speech unit group in this way, the speech unit can be increased, deleted, or corrected for each speech unit group. It is possible to respond quickly to and easily update to the user's preference.

  In the case where synthesized speech of different voice qualities is uttered depending on the speaker, the speech segment group is obtained by sampling the speech of a speaker such as a male, female, announcer, actor, or anime character, It consists of speech segments collected by shaping the waveform of the voice quality characteristics. Further, when a synthesized voice having different voice qualities is uttered according to content, the speech segment group is composed of speech segments defined and collected according to content such as news or weather forecast. For example, when defining a speech segment group according to the content, it is defined as a speech segment group that collects speech segments of a male announcer's voice quality that utters news, and the voice quality of a female announcer that utters a weather forecast There is a case where it is defined as a speech unit group in which speech units are collected.

  FIG. 3 is a diagram illustrating the concept of selecting speech units by the unit selection unit of the speech synthesis data set creation unit provided in the data transmission apparatus. Here, the contents of the voice database 123 in the case of the speaker 01 shown in FIG. 3 are shown in Table 1, and the contents of the voice database 123 in the case of the speaker 02 are shown as Table 2. In each table, speech unit numbers are arranged in the left column, and speech units are arranged in the right column. The speech segment is a minimum unit waveform constituting a waveform to be uttered as synthesized speech, and corresponds to the minimum unit of reading information. For this reason, the speech segment itself represents a waveform, but for the sake of convenience, in this specification, the speech segment is described as read information. Note that it is sufficient that the speech segment has a length useful for identifying the difference in voice quality, and the length is preferably determined for each speaker.

  Here, Table 1 shows a case in which a plurality of speech segments are registered in the speech database 123 even if they have the same semantic content when the speaker is 01. For example, the speech unit “(silent) tekisuto” having the speech unit number “No 250” and the speech unit “(silent) tekisuto” having the speech unit number “No 251” represent the same meaning content. Is registered. For example, a speech unit number “No250” represents a speech unit having a bright voice quality of the speaker 01, and a speech unit number “No251” represents a speech unit having a normal tone voice quality of the speaker 01. Is defined as This is the same for the speaker 02 in Table 2.

  As will be described later, it is also possible to define the synthesized speech waveform obtained by connecting speech segments so as to have different voice qualities by modifying them according to the control information. Therefore, as described above, it is not necessary to register both the voice quality of the speaker 01 and the voice quality of the normal tone, but it is necessary to register frequently used speech segments in the speech database 123. preferable. In addition, control information described later is also stored in the voice database 123.

  In addition, since speech segments are waveforms with different voice qualities, even the same semantic content may be stored in the speech database 123 according to the difference in voice qualities. For example, the speech unit number “No251” and the speech unit number “No251” are the same as the speech unit number “No250”, but the speech unit number “No251” is the same as the speech unit number “No251”. The number “No. 250” is a speech unit having a waveform different from each other.

  Next, referring back to FIG. 1 and referring to FIG. 3, a specific example in which the segment selection unit 122 selects a speech unit from the speech database 123 based on the reading information acquired by the reading information acquisition unit 121. Will be explained. FIG. 3 is a diagram for explaining the concept of segment selection by the segment selection means of the speech synthesis data set creation means provided in the data transmission apparatus. When the reading information “(silent) tekisutodesu. (Silent)” of the speaker 01 is passed from the reading information acquisition unit 121, the segment selection unit 122 searches the speech database 123 to search for the speech unit “(( silent) tekisuto "is selected to obtain the speech unit number" No250 "and the speech unit" desu. (silent) "is selected to obtain the speech unit number" No480 ". Similarly, for the speaker 02, the speech unit numbers “No255” and “No497” of the reading information “(silent) tekisutodesu. (Silent)” are acquired.

  However, the segment selection means 122 selects a speech segment based on a selection command input by a content provider through a user interface (not shown). As the selection command, for example, as described above, the unit selection unit 122 may select a speech unit for each speaker. However, the selection command is not limited to this. It may be performed every time, or may be performed according to the voice expression of the speaker. For example, the segment selection means 122 may select a voice quality of a specific speaker with a bright expression or a normal voice quality separately.

  The control information creation means 124 is based on the type information and the unit information for dividing the speech unit group composed of a plurality of speech units of the same voice quality constituting the minimum unit waveform to be uttered as synthesized speech, for each voice quality. Thus, at least one type of control information that defines how to deform the waveform of the synthesized speech generated by connecting the speech units associated with each other is created. Note that the type information includes information indicating the type of classification of speakers 01, 02, and the like.

  The control information is obtained by finely adjusting the waveform of the synthesized speech obtained by connecting the speech units, and changing the synthesized speech waveform obtained by connecting the speech units to different speech expression types. Shape into corrugations. Examples of the control information include speech speed control information, dynamic feature control information, and static feature control information.

  The speech speed control information is used to change the speech speed of a waveform generated by connecting speech segments. As the speech speed control information, for example, there is a voice lead setting function. This voice lead setting function switches, for example, to five stages of speaking speed. This speech speed control information can convert the speech speed by expanding the waveform in the time axis direction.

  The dynamic feature control information is used to vary a dynamic feature that changes the movement of the waveform according to the elapsed time of the waveform generated by connecting speech segments. As this dynamic feature control information, for example, a fundamental frequency or a spectrum of a waveform can be mentioned. In addition, as the dynamic feature control information, “between” in a story and a sense of tempo can be cited.

  The static feature control information is used to vary a static feature that changes the shape of a predetermined time at a predetermined time of a waveform generated by connecting speech segments. Examples of the static feature control information include an average value of fundamental frequencies and an outline of a spectrum envelope.

Here, the concept of the data set for speech synthesis will be described with reference to FIG. FIG. 4 is a diagram for explaining the concept of the speech synthesis data set of the speech synthesis data set creation means provided in the data transmission apparatus.
The data set for speech synthesis can be divided into general items that do not depend on the voice quality of each speaker or the like and special items that differ depending on the voice quality of each speaker or the like.

  General matters include, for example, content information such as news, text such as “Example text”, reading information such as “(silent) te'kisutodesu. (Silent)”, and “type 1 noun 1”. There is an accent type for each part of speech. Note that “′” in the reading information represents an accent. Here, the speech synthesis data set is described as including text such as a sentence example “text”. However, as will be described later, the data receiving device 20 performs speech based on the reading information. In order to perform synthesis, the speech synthesis data set may not include text.

On the other hand, as the special items, for example, type information indicating the type of classification of the speakers 01, 02, etc., unit information for selecting a speech unit in the data receiving device 20, and a type such as pitch are indicated. There is control information.
Here, the speaker 01 is, for example, the segment information “No 250 (tekisuto), No 480 (desu.)”, The control information type 1 “average pitch down”, and the control information type 2 “average pitch rise, accent. "1 emphasis". In addition, in order for speaker 02 to have a voice quality different from that of speaker 01, for example, segment information “No255 (tekisuto), No497 (desu.)”, Control information type, 1 “average pitch down”, control information type 2 “accent 1 emphasis”. Note that the segment information may be only the speech segment number. This is because, as will be described later, the data reception device 20 calls a speech unit from the speech database 251 described later using the speech unit number as a key.

The voice synthesis data set adding means 13 acquires the data broadcast material output from the material text acquisition means 11 and incorporates the voice synthesis data set created by the voice synthesis data set creation means 12 into the data broadcast material. It is.
For example, the voice synthesis data set adding unit 13 adds the voice synthesis data set created by the voice synthesis data set creation unit 12 to the data transmission program. As the data broadcasting program, there is a BML (Broadcast Markup Language) program.

This BML program is written in a script language, and is read as “<dedicated tag> reading information </ dedicated tag>” between “<dedicated tag>” and “</ dedicated tag>” in the script. Various information such as information is inserted and described.
Note that the voice synthesis data set may be transmitted on the broadcast wave as communication data sequentially, even if it is not described so as to be sandwiched between tag information in the data broadcasting program.

  The voice database update information creation unit 14 creates update information to be transmitted to the data receiving device 20 when the voice database 123 is updated by update data input via a user interface (not shown) by the content provider. Is. The update data and the update information have the same contents in principle, but have different names because they may be exceptionally different.

Here, the relationship between the update data and the update information will be described while explaining the outline of the process at the time of updating the voice database with reference to FIG. FIG. 5 is a diagram for explaining the concept of creating update information when updating the voice database.
First, update data is input to the data transmission device 10 via a user interface (not shown). Here, the update data d1 is a case where a new speech unit of an existing speaker is added. The update data d1 in this case includes, for example, a speaker type “speaker 01”, a speech unit number “No500”, reading information “desita”, and a speech unit “desita”.

  The input update data d1 is input to the voice database update information creation unit 14 to create update information D1. Here, the created update information D1 is the same as the update data d1. Here, when the update data d1 and the update information D1 are different, there is a case where the corresponding item of the input update data d1 is omitted when diverting the specific information of the specific speaker already registered. is there. For example, if the speech unit specified by the update data d1 registered as the speaker 01 is already registered as the speech unit of the speaker 02, the speech unit input is omitted and the speech database is updated. The information creating unit 14 creates updated information by calling the omitted speech segment.

  Returning to FIG. 1, the voice database update information creating unit 14 registers the created update information in the voice database 123 and outputs it to the multiplexing unit 15. In addition to the addition of a new speech unit for an existing speaker, the update information includes information obtained by improving a speech unit for a new speaker or a speech unit for an existing speaker.

  The multiplexing means 15 multiplexes the data broadcasting material incorporating the voice synthesis data set output from the voice synthesis data set adding means 13 and the update information output from the voice database update information creating means 14. Data is generated, and the generated multiplexed data is sent to the transmission means 16. Here, since the multiplexing means 15 multiplexes the data broadcast material and the update information, the data broadcast material includes text. However, as will be described later, the data receiving device 20 does not recognize characters from text and synthesizes speech, so that multiplexed data that does not include text may be used.

  The transmission unit 16 transmits the data broadcasting material including the voice synthesis data set and the update information on the broadcast wave as multiplexed multiplexed data to the data reception device 20 by the multiplexing unit 15. . The transmitter 16 is specifically an antenna (not shown), but here, a D / A converter (not shown) for digital / analog conversion of the multiplexed data multiplexed by the multiplexer 15 or not shown. Description will be made assuming that an electronic circuit such as an oscillator is included. In this way, the multiplexed data is transmitted from the data transmitting apparatus 10 to the data receiving apparatus 20, and as will be described later, the data receiving apparatus 20 extracts update information from the multiplexed data, and a later-described voice database 251 of the data receiving apparatus 20. , The contents of the voice database 123 of the data transmitting apparatus 10 and the later-described voice database 251 of the data receiving apparatus 20 can be made the same.

Note that the speech synthesis data set adding means 13, the multiplexing means 15, and the transmission means 16 have the segment information selected by the segment selection means 122 and the type information and control created by the control information creation means 124. It functions as a voice synthesis data set providing means for providing information to the data receiving apparatus 20 as a voice synthesis data set.
In addition, the voice database update information creation unit 14, the multiplexing unit 15, and the transmission unit 16 function as update information provision unit for providing the data reception device 20 with the update information created by the voice database update information creation unit 14. To do.

  The data transmitting apparatus 10 can be realized by causing a general computer to execute a program and operating an arithmetic device or a storage device in the computer. This program (speech synthesis data set generation program) can be distributed via a communication line, or can be distributed on a recording medium such as a CD-ROM.

(Configuration of data receiving device)
The data receiving device 20 as a synthesized speech utterance device includes type information and speech elements for dividing a speech unit group composed of a plurality of speech units of the same voice quality that constitute a minimum unit waveform to be uttered as a synthesized speech for each voice quality. At least defined how to transform the waveform of the synthesized speech generated by connecting the speech units associated with the speech units of each group and the speech units associated with the type information and the segment information In order to acquire a speech synthesis data set from the data transmitting apparatus 10 that generates one type of control information as a speech synthesis data set and to utter as synthesized speech, the reception unit 21, the data identification unit 22, and the speech synthesis control The information setting unit 23, the voice database update unit 24, the voice synthesis unit 25, the user interface 26, and the voice output unit 27 are mainly used. Eteiru.

  The receiving unit 21 receives a broadcast wave transmitted from the data transmitting apparatus 10 and passes received data as an electric signal to the data identifying unit 22. The receiving unit 21 is specifically an antenna (not shown), but here, an amplifier circuit or A / D that performs various conversions until the broadcast wave received by the antenna is passed to the data identifying unit 22 as received data. It demonstrates as what includes electronic circuits, such as a conversion part.

  If the received data passed from the receiving means 21 is multiplexed data transmitted from the data transmitting apparatus 10, the data identifying means 22 identifies data broadcast material and update information from the multiplexed data, and from the data broadcast material A data broadcasting program is extracted, and a speech synthesis data set is identified from the data broadcasting program. Further, the data identification means 22 has a function of sending a voice synthesis data set to the voice synthesis control information setting means 23 and sending update information to the voice database update means 24.

  Note that when the data identification unit 22 identifies the reception data passed from the reception unit 21 as not multiplexed data, there is noise having the same frequency as the frequency transmitted from the data transmission device 10.

Here, the concept of the function of the data identification means 22 will be described with reference to FIG. FIG. 6 is a diagram for explaining the concept of the function of the data identification means provided in the data receiving apparatus.
When receiving the received data from the receiving means 21, the data identifying means 22 determines whether the received data is multiplexed data based on the data amount of the received data. And update information are identified and separated. In the case of the update information D2, the data identification means 22 sends the update information D2 such as segment information, reading information, speaker information (type information), content information (type information) and control information to the voice database update means 24. send. If the received data received by the receiving unit 21 is not multiplexed data, the data identifying unit 22 determines that the data is not transmitted to the data receiving device 20 and ignores the received data.

  On the other hand, when the data broadcasting material is, for example, the data broadcasting program P, the data identification means 22 is sandwiched between the script and the <dedicated tag> and </ dedicated tag> from the data broadcasting program P. The data is identified as a voice synthesis data set, and the voice synthesis data set is sent to the voice synthesis control information setting device 23.

  Referring back to FIG. 1, the speech synthesis control information setting unit 23 synthesizes a speech synthesis group 25 to extract an extraction target from a speech unit group stored in the speech database 251, which will be described later, from the data identification unit 22. Unit information specifying process specified by selecting the type information of the data set for control, and control information setting for specifying the waveform deformation method by selecting the control information of the speech synthesis data set passed from the data identifying means 22 It has a function as a process, and sets type information, control information, and segment information passed from the data identification means 22 as speech synthesis data.

The speech synthesis control information setting unit 23 mainly includes a content item type setting unit 231 and a speaker setting unit 232 in order to function as segment information specifying processing.
The content item type setting means 231 sets the type of the speech element group stored in the audio database 251 according to the type of the content item included in advance in the type information.
The speaker setting means 232 sets the type of the speech element group stored in the speech database 251 according to the difference of speakers included in the type information in advance.

Note that the identification in the segment information identification process of the speech synthesis control information setting unit 23 displays a screen for selecting which speech segment group to select, and allows the user to specify via the user interface 26.
For example, the voice synthesis control information setting unit 23 sets the female voice when the user designates a female voice via the user interface 26 when a male voice and a female voice are given as the piece information by the piece information specifying process. The voice element group specified as the element information from the voice element group is identified as the female voice element group so as to be selected from the voice database 251.

The speech synthesis control information setting unit 23 mainly includes a speech speed setting unit 233, a dynamic feature control information setting unit 234, and a static feature control information setting unit 235 in order to function as a control information setting process. ing.
The speech speed setting means 233 sets speech speed control information, which is control information for changing the speech speed of a waveform generated by connecting speech segments.
The dynamic feature control information setting unit 234 sets dynamic feature control information that is control information for changing dynamic features that change the movement of the waveform according to the elapsed time of the waveform generated by connecting the speech segments. Is. This dynamic feature is a physical property in which a waveform uttered as synthesized speech changes with time.
The static feature control information setting unit 235 sets static feature control information that is control information for changing a static feature that changes a shape of a predetermined time at a predetermined time of a waveform generated by connecting speech segments. Is. This static feature is a physical property that is uniquely determined at a predetermined time at a predetermined time of a waveform to be uttered as synthesized speech.

Then, the speech synthesis control information setting means 23 sets the control information related to the speaker, speech speed, intonation or voice quality designated by the user for each data item to make speech synthesis data. It is sent to the speech synthesis means 25
The voice synthesis control information setting unit 23 sets control information for uniquely designating the waveform deformation method by the control information setting process. Examples of the control information include functions, parameters, and control information numbers. These functions, parameters, and control information numbers are for finely adjusting the waveform deformation method generated by connecting speech segments, and do not particularly mean that the waveform is completely different.

  That is, the setting of the control information does not change the voice quality itself set according to the segment information to a completely different voice quality, but performs fine adjustment of the voice quality. For example, the waveform is deformed so that the sharp corners of the waveform are rounded and smoothed, and the voice quality of the same person is finely adjusted from a normal tone to a gentle tone. Note that the control information number is handled as a number that defines a waveform deformation method as a number corresponding to a function or parameter. The control information does not particularly mean that the control information is transformed into a completely different waveform as described herein, but the voice information may be changed to a completely different waveform according to the user's request.

Furthermore, the speech synthesis control information setting unit 23 has a function of analyzing the speech synthesis data set sent from the data identification unit 22 and dividing it into items. In the speech synthesis control information setting unit 23, the content item type setting unit 231, the speaker setting unit 232, the speech speed setting unit 233, the dynamic feature control information setting unit 234, and the static feature control information setting unit 235 are included in the user interface 26. A selection item for displaying a character for selecting an item or a message on a display unit (not shown) is received from the data identification unit 22 and displayed on the display unit of the user interface 26, and selection from an input unit (not shown) is effective. Let me.
With reference to FIG. 7, the concept of analyzing a speech synthesis data set will be described. FIG. 7 is a diagram for explaining the concept of analyzing a speech synthesis data set.

The voice synthesis control information setting unit 23 performs voice synthesis on the script and the data sandwiched between the <dedicated tag> and the </ dedicated tag> extracted from the data broadcasting program P identified by the data identification unit 22. Data sets are acquired, and general matters and special matters of speech synthesis data sets are distinguished.
As described above, for example, as described above, for example, content information such as news, text such as “example text”, reading information such as “(silent) tekisutodesu. (Silent)”, and “type 1 noun 1” There are accent types for each part of speech. Here, a case where there is no “′” in the reading information is shown.

On the other hand, as described above, as described above, for example, type information indicating the classification type of the speakers 01, 02, etc., unit information for selecting a speech unit in the data receiving device 20, and pitch, etc. There is control information that indicates the type.
Here, the speaker 01, for example, has segment information “No 35 (teki), No 45 (suto), No 15 (desu.)”, Control information type 1 “average pitch descent”, and control information type 2 “ "Average pitch rise". Further, in order to obtain a voice quality different from that of the speaker 01, the speaker 02 has, for example, the element information “No20 (tekisuto), No10 (desu.)”, The type of control information for the same meaning content as the speaker 01. 1 “average pitch down”, control information type 2 “accent 1 emphasis”.

  Here, the case where the content of the data set for speech synthesis shown in the description of the data transmission device 10 is different is shown, but this is for the purpose of illustrating the data set for speech synthesis, and the data from the data transmission device 10 This does not mean that the speech synthesis data sets transmitted to the receiving device 20 are different. In the following description, it is assumed that the content of the speech synthesis data set shown in the data receiving device 20 is transmitted from the data transmitting device 10.

Returning to FIG. 1, the voice database update unit 24 updates the postscript voice database 251 of the voice synthesis unit 25 in accordance with the update information sent from the data identification unit 22.
Here, the concept of the function of the voice database update means 24 will be described with reference to FIG. FIG. 8 is a diagram for explaining the concept of the function of the voice database update means provided in the data receiving apparatus.

  The voice database update unit 24 updates the voice database 251 based on the content of the update information D2 sent from the data identification unit 22 and makes the user perform various settings via the user interface 26 (not shown). Update the display on the display. For example, the voice database update unit 24 selects the voice database update data D3 from the update information D2. As the voice database update data D3, for example, there are voice segments and reading information.

The voice database update unit 24 stores the voice database update data D3 in the voice database 251. The voice database update data D3 may include, for example, a specific information deletion command or a specific information correction command.
The voice database update unit 24 selects user interface setting data D4 from the update information D2. Examples of the user interface setting data D4 include speaker information and control information.

Then, the voice database update unit 24 outputs a command for adding options such as a selection button and a selection number of the user interface setting data D4 on a screen of a display unit (not shown) of the user interface 26 to the user interface 26.
In this case, the voice database update unit 24, when update information for adding the voice segment group of the voice quality of the announcer X is identified by the data identification unit 22, a button for options for causing the user to select the announcer X, etc. A command for displaying the character on the display means (not shown) of the user interface 26 is output to the user interface 26.

  The voice database updating unit 24 displays an unshown display of the user interface 26 when an item already registered in the voice database 251 described later is synthesized by the update information identified by the data identifying unit 22. A command for deleting a character such as an option button displayed on the means is output to the user interface 26.

  Returning to FIG. 1, the speech synthesizer 25 includes a speech database 251. The voice database 251 stores a voice element group for each voice quality. Here, the speech segment group is a collection of speech segments that make up the same voice quality and stored in the speech database 251, and is defined by classifying different types of voice quality. For example, as will be described later, the speech database 251 stores a speech segment group defined by content and speaker differences so that it can be distinguished for each type.

  More specifically, when the voice database 251 stores voice segment groups of four types of voice qualities, a voice of a male announcer, a voice of a female announcer, a voice of an actor, and a voice of an actress. The unit information of the speech synthesis data set is information for identifying which speech unit group of the above four types of speakers is to be selected. As this information, a speech unit number associated with a speech unit can be cited.

  Accordingly, the speech synthesizer 25 receives the speech synthesis data from the speech synthesis control information setting unit 23, and from the speech segment group of the speech database 251 specified by the speech synthesis data type information, Extract speech segments according to fragment information, connect the extracted speech segments to generate a synthesized speech waveform, transform the waveform according to the waveform modification method specified by the speech synthesis data control information, and transform The generated waveform is uttered as synthesized speech.

  For example, the speech synthesizer 25 selects speech units constituting a female speech unit group from the speech database 251 according to the type information, generates a waveform by connecting the speech units, and uses it as speech speed control information. When the fast mouth is set, the waveform is processed so that the cycle of the generated waveform is shortened, and the synthesized voice is uttered by a fast-paced female voice. At this time, when “1.5 times speed” is set as the speech speed control information, the voice synthesizing unit 25 sets the time generated so that the waveform generated by connecting the speech elements to each other becomes 1.5 times speed. The waveform is deformed so as to shrink in the axial direction.

  Therefore, the speech synthesizer 25 multiplexes the speech unit associated with the speech unit number and stored in the speech database 251 into the multiplexed data sent from the data transmitting apparatus 10 as multiplexed data. A speech unit corresponding to the speech unit number, that is, a waveform is selected from the speech database 251 by searching and extracting the speech unit number included in the speech synthesis data set as a key, and the selected speech By connecting the segments, a synthesized speech waveform can be generated, the generated synthesized speech waveform can be transformed according to the control information, and the synthesized speech of the deformed waveform can be output to the speech output means 27.

Here, the concept of speech synthesis by the speech synthesizer 25 will be described with reference to FIG. FIG. 9 is a diagram for explaining the concept of speech synthesis by speech synthesis means provided in the data receiving apparatus.
When the speech synthesis unit 25 acquires the speech synthesis data from the speech synthesis control information setting unit 23, the speech synthesis unit 25 analyzes the speech synthesis data, and includes general matters such as content information, text, reading information, accent type, and part-of-speech type, A speech unit for each speaker set by the user is identified from special items such as an average pitch descent function.

  As speech synthesis data, the content item is “News”, the text is “Text”, the reading information is “(silent) tekisutodesu. (Silent)”, the accent type is “Type 1 noun 1”, the story “Speaker 02” of the speaker number, “speech unit“ tekisuto ”of the speech unit number“ No20 ”and“ desu. ”Of the speech unit number“ No1 ”of the unit information, and a function Pav ( 0.8) is specified.

  In this case, when the speech synthesis unit 25 identifies the speech synthesis data, the speech synthesis unit 25 identifies the speech unit group from which the speech unit is extracted according to the type information, refers to the unit information, and sets the speech unit number. The speech database 251 is searched for the key, and the speech unit waveform corresponding to the speech unit number is extracted. The speech synthesizer 25 connects speech units obtained by searching speech unit waveforms based on speech unit numbers extracted from the speech database 251 and connects the whole text of a text to be uttered as synthesized speech or a predetermined text. A synthesized speech waveform is generated for each length. Thereby, the speech synthesizer 25 can utter the synthesized speech of the voice quality based on the speaker 02.

  Next, the speech synthesizer 25 transforms the waveform of the synthesized speech generated according to control information such as accent type and part of speech type. For example, the accented portion is deformed so that the amplitude of the designated accented portion of the waveform of the synthesized speech is increased so that the accented portion is uttered strongly.

  In this case, the speech synthesizer 25 transforms the waveform generated by the noun part to be uttered as synthesized speech into a waveform in which a part of the waveform specified by the accent type is emphasized based on the accent type, Based on Pav (0.8), the waveform is deformed so as to decrease the average pitch, and the waveform of the synthesized speech of voice quality based on the speaker 02 is deformed. Then, the voice synthesis unit 25 outputs the deformed waveform to the voice output unit 27 and causes the voice output unit 27 to utter as synthesized voice. As a result, the speech synthesizer 25 can deform the synthesized speech waveform generated by connecting speech segments, and utter a synthesized speech that meets the user's preference.

  The user interface 26 is mainly used to input content item, speaker, speech speed, dynamic feature or static feature control information to the data receiving device 20, and the data receiving device 20 issues various commands to the user. In addition to the input means to be performed, there is also provided display means (not shown) for displaying various information from the data receiving device 20.

  For example, as described above, when the speaker setting unit 232 can set four types of voice qualities, the user interface 26 displays four types of buttons on a display unit (not shown), A speaker can be selected by having the user select one of them. As a result, the user can utter a synthesized voice having a favorite voice quality from the data receiving apparatus 20 by performing an easy operation of selection via the user interface 26.

Here, with reference to FIG. 10, a concept of setting control information and the like in the user interface 26 will be described. FIG. 10 is a diagram for explaining the concept of setting speech synthesis control information by a user interface provided in the data receiving apparatus.
Here, the display unit of the user interface 26 displays choices of content item, speaker, speech speed, and speech expression type. Here, the type of phonetic expression is a type defined by the control information, and is a type in which items of the control information are classified for selection by the user. For example, classification is such that the average pitch is lowered for Type 1 and the average pitch is raised for Type 2.

  In addition, the input means (not shown) of the user interface 26 is not limited to this, and may be a so-called touch panel, even if various options are selected by a so-called remote controller capable of data communication with the data receiving device 20 by infrared communication. A selection button as an option may be displayed above, and selection of the selection button may be input by a touch on the screen by the user.

As shown in FIG. 10, the content item includes, for example, news, weather, or program commentary. As the speaker, for example, there are a man, a woman, an actor, an actress or an anime character.
Here, for example, the speaker number is associated with the speaker name, such as speaker 01 as a man, speaker 02 as a woman, speaker 03 as an actor, speaker 04 as an actress, and speaker 05 as an anime character. It is displayed as an option.

  As the speaking speed, for example, the standard speed is set to 1.0.times., And an option that enables selection of the speaking speed from 0.8 to 1.4 times depending on the ratio to the reference speed is displayed. ing. Note that, as described above, the speech speed is not limited to the case indicated by the ratio with respect to the reference speed, and may be set by the number of spoken characters per unit time, for example.

  Also, as the type of speech expression, a defined type number is selected as in type 1 and type 2. These types 1 and 2 are defined for each speaker in the special items of the speech synthesis data set as described above. For example, in the case of the speaker 01 in FIG. 8, type 1 is defined as an average pitch decrease, and type 2 is defined as an average pitch increase. Note that, as shown in FIG. 9, the voice expression type may be set in a stepwise manner as shown in FIG. 9 instead of selecting a type unambiguously as “slightly 1”.

  In addition, the content item name, speaker number, speech speed setting value, and speech expression type are displayed as numbers to distinguish them, and the options can be displayed in a separate window from the screen that displays the options. The user may be provided with choices by distributing a paper medium on which is printed. In this case, the user selects the option by viewing the option and inputting the number displayed on the screen or by touching the screen.

  Here, referring to FIG. 11, the speech synthesis control information setting means 23 having the above-described functions is set according to the setting data sent from the user interface 26 described with reference to FIG. Each of speaker setting means 232, speech speed setting means 233, dynamic feature control information setting means 234, and static feature control information setting means 235 (hereinafter collectively referred to as “content item type setting means 231 etc.”). The case where the option of this is set is demonstrated.

  When the voice synthesis control information setting unit 23 receives the voice synthesis data set from the data identification unit 22, as described above, the voice synthesis control information setting unit 23 identifies the content of the voice synthesis data set, and selects each option of the content item type setting unit 231 or the like. Set, wait for setting data from the user interface 26, select each option of the content item type setting means 231, etc., and set it as voice synthesis data including the selected option, and this voice synthesis data Is output to the speech synthesizer 25.

  When the content item of the setting data indicates news, the content of the content item type setting means 231 is selected as news. When the content item of the setting data indicates the speaker 02, the speaker 02 of the speaker setting means 232 is selected. When the control information of the setting data indicates the speech speed, a double speed value such as 1.2 double speed is selected. When the control information of the setting data is an average pitch lowering function representing dynamic feature control information, the average pitch lowering function Pav of the static feature control information setting means 235 is selected, and a variable value 0.8 specified by the user is set. Substitute and set the function Pav (0.8).

  Here, since the setting data does not include the moving feature control information, the moving feature control information setting unit 234 outputs a message prompting the user to select a moving feature via the user interface 26, and receives a message from the user. Waiting for selection, a value set in advance as a specified value is selected after a predetermined time. As described above, the speech synthesis control information setting unit 23 waits for the input of the setting data from the user interface 26 when the input of the data set for speech synthesis is received from the data identification unit 22, and Those not present are set to predetermined values.

  Returning to FIG. 1, the voice output unit 27 utters the synthesized voice waveform as an electric signal output from the voice synthesis unit 25 as synthesized voice, and specifically, a speaker. Then, the voice output unit 27 utters synthesized speech according to the waveform of the electrical signal sent from the voice synthesis unit 25. Note that, here, the voice output means 27 is used in particular to utter a synthesized voice, and will be described as being separate from the user interface 26. Since it can be output or output as a notification sound, it can be configured to be included in the user interface 26.

Note that the receiving unit 21 and the data identifying unit 22 change the waveform generated by connecting the unit information and the speech unit for specifying the speech unit to be uttered as synthesized speech for each type of voice quality. It functions as a voice synthesis data set acquisition means for acquiring control information for voice synthesis from the data transmission device 10 as a voice synthesis data set.
The receiving means 21 and the data identifying means 22 function as update information acquiring means for acquiring update information for updating the speech segment group stored in the speech database 251 from the data transmitting apparatus 10.

  Further, the speech synthesis control information setting means 23 sets which type of speech unit group to be selected from the speech unit group stored in the speech database 251, and is identical from the set speech unit group. Functions as a unit information specifying unit for specifying unit information for selecting a type of speech unit and a control information setting unit for setting control information acquired by the data set acquisition unit for speech synthesis (21, 22) To do.

  By the way, the data receiving device 20 can be realized by causing a general computer to execute a program and operating an arithmetic device or a storage device in the computer. This program (synthetic voice utterance program) can be distributed via a communication line, or can be written on a recording medium such as a CD-ROM for distribution.

Then, each operation | movement of the data transmitter 10 in the data broadcasting system 1 and the data receiver 20 is demonstrated.
(Operation of data transmitter)
First, the operation of the data transmitting apparatus 10 according to the embodiment of the present invention will be described with reference to FIG. 12 (refer to FIGS. 1 to 11 as appropriate). FIG. 12 is a flowchart showing the operation of the data transmission apparatus according to the embodiment of the present invention.

<Update data acquisition step>
First, a control unit (not shown) of the data transmitting apparatus 10 determines whether or not there is update data for updating the audio database 123 via various interfaces by the content provider, and acquires update data when there is update data. Then, the process moves to step SA2 (Yes in step SA1), and if the update data is not within the predetermined time, the process moves to step SA4 (No in step SA1). Note that the predetermined time is represented by, for example, a relative time when the data transmission device 10 is turned on or when a transmission start button (not shown) is pressed, or when the time is started from when the data transmission device 10 is pressed. It is what is done.

<Update information creation step>
When receiving the update data, the voice database update information creating means 14 creates update information for updating the voice database 123 (step SA2). For example, as shown in FIG. 5, when the content provider inputs the update data including the speaker number speaker 01, the speech unit number, the reading information, and the speech unit, the update information is also stored in the speech information. Person, speech unit number, reading information, and speech unit.

<Audio database update step>
The voice database update information creation unit 14 updates the voice database 123 based on the created update information (step SA3). The speech unit number is input by the content provider together with the speech unit. However, the speech database update information creation unit 14 manages the numbering of the speech unit number, and the speech unit number is input by the user. Each time it is done, a new speech unit number may be assigned.

<Material text acquisition step>
Next, the material text acquisition unit 11 receives a data broadcast material stored in a storage unit such as a data broadcast material processing apparatus (not shown), and acquires a text to be uttered as synthesized speech from the data broadcast material (step SA4) The acquired text is output to the voice synthesis data set creation means 12, and the data broadcast material is output to the voice synthesis data set addition means 13.

<Step of creating a data set for speech synthesis>
Upon receiving the text sent from the material text acquisition unit 11, the speech synthesis data set creation unit 12 starts a speech synthesis data set creation process (step SA5), and performs the processes of steps SA6 to SA8 for all the texts. The voice synthesis data set creation process is repeated until is finished (step SA9).

<Reading information acquisition step>
In the speech synthesis data set creation unit 12, when the reading information acquisition unit 121 acquires the text sent from the material text acquisition unit 11, it recognizes characters by referring to various dictionaries such as a grammar dictionary and a language dictionary, The text morphemes are analyzed with reference to various dictionaries, reading information is acquired (step SA6), and the reading information is passed to the segment selection means 122.

<Unit selection step>
The segment selection unit 122 searches the speech database 123 according to the reading information passed from the reading information acquisition unit 121, and selects a speech unit according to the reading information for each speaker defined from the speech database 123 (step SA7) The selected speech segment is transferred to the control information creation means 124.

<Control information creation step>
The control information creation unit 124 receives the speech unit selected from the speech database 123 by the unit selection unit 122, and generates a synthesized speech waveform obtained by connecting the speech units for each type of voice quality such as a speaker. Control information for changing the type and type information for distinguishing the type are created (step SA8).
Then, the speech synthesis data set creation unit 12 includes the reading information acquired by the reading information acquisition unit 121, the speech unit selected by the unit selection unit 122, the control information generated by the control information generation unit 124, The data set for speech synthesis including the type information is output to the speech synthesis data set adding means 13.

<Adding data set for speech synthesis>
The voice synthesis data set adding means 13 acquires the data broadcast material output from the material text acquisition means 11, and incorporates the voice synthesis data set created by the voice synthesis data set creation means 12 into the data broadcast material. (Step SA10), the data is transferred to the multiplexing means 15.

<Multiplexing step>
The multiplexing means 15 multiplexes the data broadcasting material incorporating the voice synthesis data set output from the voice synthesis data set adding means 13 and the update information output from the voice database update information creating means 14. Data is generated, and the generated multiplexed data is sent to the transmission means 16 (step SA11).

<Transmission step>
The transmission means 16 puts the multiplexed data obtained by multiplexing the data broadcasting material including the voice synthesis data set received from the multiplexing means 15 and the update information on the broadcast wave of a predetermined frequency, toward the data receiving apparatus 20. Transmit (step SA12).

  Through the above processing, the data transmitting apparatus 10 multiplexes the speech synthesis data set including the speech units to be uttered as synthesized speech and the update information indicating the update contents of the speech database 123 in which the speech units are databased. Since the data is transmitted to the data receiving device 20, the data receiving device 20 generates a synthesized speech waveform by connecting speech units of the speech synthesis data set, and control information of the speech synthesis data set. The waveform of the synthesized speech generated according to the above can be shaped and uttered as synthesized speech.

  Finally, referring to FIG. 13 (refer to FIGS. 1 to 11 as appropriate), the operation of the data receiving apparatus according to the embodiment of the present invention will be described. FIG. 13 is a flowchart showing the operation of the data receiving apparatus according to the embodiment of the present invention.

<Reception step>
In the data receiving device 20, when the receiving means 21 receives a broadcast wave of a predetermined frequency (step SB1), analog / digital conversion is performed and multiplexed data as digital data is passed to the data identifying means 22.

<Data identification step>
If the received data delivered from the receiving means 21 is multiplexed data transmitted from the data transmitting apparatus 10, the data identifying means 22 identifies the data set for speech synthesis and the update information from the multiplexed data (step SB 2), the speech synthesis data set is sent to the speech synthesis control information setting means 23, and the update information is sent to the speech database update means 24.

<Audio database update step>
First, when the data identification means 22 identifies update information (Yes in step SB3), the voice database update means 24 is sent from the data identification means 22 because update information is given to the voice database update means 24. In accordance with the incoming update information, the voice database 251 of the voice synthesizer 25 is updated (step SB4). If there is no update information (No in step SB3), the process proceeds to step SB5.

<Speech synthesis control information setting step>
When the speech synthesis data set identified by the data identification means 22 includes type information and control information (step SB5), the speech synthesis control information setting process is started (step SB6), and all speech synthesis The voice synthesis control information setting process is repeated until the process of steps SB6 to SB11 is completed for the data set (step SB12).

<Content item type selection step>
The content item type setting unit 231 acquires a content item (type information) from the speech synthesis data set and displays the content item on a display unit (not shown) of the user interface 26. Then, when the content item type setting unit 231 acquires the setting data of the content item input by the user via the input unit (not shown) of the user interface 26, the content item type setting unit 231 selects the content item based on the setting data. If the content item is not selected within a predetermined time, the process proceeds to the next step.

  As a result, the content item type setting means 231 specifies segment information for selecting a speech segment group from the speech database 251 using the content item selected here as a key (step SB7). For example, in the case of news content, a speech unit can be selected from a group of male speech units.

<Speaker setting step>
The speaker setting unit 232 acquires type information (speaker type and speaker number) related to the speaker from the speech synthesis data set, and displays the speaker information on a display unit (not shown) of the user interface 26. When the speaker setting unit 232 acquires the setting data of the speaker information input by the user via the input unit (not shown) of the user interface 26, the speaker setting unit 232 selects the speaker based on the setting data. If the speaker is not selected within a predetermined time, the process proceeds to the next step.

  Thereby, the speaker setting means 232 specifies unit information for selecting a speech unit group having different voice qualities from the speech database 251 according to the difference between speakers (step SB8).

<Speaking speed setting step>
The speech speed setting means 233 acquires information related to the speech speed from the speech synthesis data set, and displays the speech speed setting item on a display means (not shown) of the user interface 26. Then, when the speech speed setting means 233 obtains the speech speed setting data input by the user via the input means (not shown) of the user interface 26, the speech speed setting means 233 selects the speech speed based on the setting data. If the speech speed is not selected within a predetermined time, the process proceeds to the next step.

  Thereby, the speech speed setting means 233 sets speech speed control information which is control information for changing the speech speed of the waveform generated by connecting the speech elements (step SB9).

<Dynamic feature control setting step>
The dynamic feature control information setting unit 234 acquires dynamic feature control information from the speech synthesis data set, and displays dynamic feature setting items on a display unit (not shown) of the user interface 26. When the dynamic feature control information setting unit 234 acquires the setting data of the dynamic feature control information input by the user via the input unit (not shown) of the user interface 26, the dynamic feature control information setting unit 234 outputs the dynamic feature control information based on the setting data. select. If the control information of the dynamic feature is not selected within a predetermined time, the process proceeds to the next step.

  Accordingly, the dynamic feature control information setting unit 234 sets dynamic feature control information that is control information for changing the movement of the waveform according to the elapsed time of the waveform generated by connecting the speech segments. (Step SB10).

<Static feature control setting step>
The static feature control information setting unit 235 acquires static feature control information from the speech synthesis data set, and displays static feature setting items on a display unit (not shown) of the user interface 26. When the static feature control information setting unit 235 acquires static feature control information setting data input by the user via an input unit (not shown) of the user interface 26, the static feature control information setting unit 235 obtains static feature control information based on the setting data. select. If the static feature control information is not selected within a predetermined time, the process proceeds to the next step.

  Thus, the static feature control information setting means 235 is static feature control information that is control information for changing a static feature that changes the shape of a predetermined time at a predetermined time of a waveform generated by connecting speech segments. Is set (step SB11).

<Speech synthesis step>
The speech synthesizer 25 acquires the speech synthesis data sent from the speech synthesis control information setting unit 23 (step SB13). Then, the speech synthesizer 25 extracts the type information and the speech unit number from the speech synthesis data, searches the speech database 251 according to the type information using the speech unit number as a key, and extracts the speech unit. The speech synthesizer 25 connects speech segments extracted from the speech database 251 to generate a synthesized speech waveform (step SB14). Thereafter, the voice synthesizer 25 extracts various control information from the voice synthesis data, deforms the generated waveform, and outputs the deformed waveform to the voice output unit 27 (step SB15).

<Audio output step>
The voice output means 27 utters a waveform as an electric signal input from the voice synthesis means 25 as synthesized voice (step SB16).

  Through the above processing, the data reception device 20 sets different voice qualities for each type of speaker or the like based on the voice synthesis data set sent from the data transmission device 10, and the voice unit of the set voice quality. Since the user can easily transform the waveform of the synthesized speech obtained by connecting each other by specifying the control information, the synthesized speech can be uttered according to the user's preference.

[Supplement]
Although the receiving means 21 of the above embodiment has been described as a configuration including an electronic circuit such as an amplifier circuit or an A / D converter, the electronic circuit is a circuit constructed in hardware by an electronic component or a semiconductor device. Alternatively, it may be realized by the cooperation of a program that realizes the same type of function and a CPU that processes the program.

  In the above-described embodiment, the unit information of the speech synthesis data set is information including a speech unit number and a speech unit, and the data receiving device 20 includes the speech database 251 that stores the speech unit. However, instead of providing the data reception device 20 with the voice database, a speech unit to be uttered as synthesized speech may be transmitted from the data transmission device 10 to the data reception device 20 each time.

  In the above-described embodiment, when the update data and the update information are different from each other, the voice database update information creation unit 14 of the data transmission device 10 is described as supplementing the voice segment waveform data from which the data is omitted. It is assumed that the speech unit having the same update information as the update data is omitted from the update information transmitted from the transmitting device 10 to the data receiving device 20, and the speech database updating unit 24 of the data receiving device 20 selects the corresponding speech unit as the speech database 251. The voice database 251 may be updated by supplementing the voice database.

  In the above embodiment, the case where the speech database 123 is built in the segment selection unit 122 has been described. However, the speech database 123 may not be built in the segment selection unit 122. Similarly, although the case where the speech database 251 is built in the speech synthesizer 25 has been described, the speech database 251 may be configured not to be built in the speech synthesizer 25.

  In the above-described embodiment, the function and operation of the data transmission apparatus 10 have been described as an apparatus configuration that can be accommodated in one housing. However, normally, the data transmission device 10 is provided as a large facility in a broadcasting station and should be called a data transmission system. This data transmission system is basically the same in function configuration and operation as the data transmission device 10 except that each configuration becomes a large data transmission device 10 and does not fit in one housing. In the specification and the claims, the functional configuration and operation have been described as a device configuration that fits in one housing for convenience. This does not exclude the configuration and function as a large-scale data transmission system from the scope of the right of the present invention.

  In the above-described embodiment, the transmission / reception relationship between one data transmission device 10 and one data reception device 20 has been described. However, the number of mutual devices may be plural. For example, one data transmission device 10 transmits data to a plurality of unspecified data reception devices 20 via radio waves such as terrestrial waves or wires such as optical fiber cables. Further, one data receiving device 20 may receive data from a plurality of data transmitting devices 10 and register the rules determined by each data transmitting device 10.

  Moreover, in the said embodiment, the case where it utilizes for the data broadcasting system 1 comprised from the data receiver 20 as a synthetic | combination speech apparatus and the data transmitter 10 as a data set production | generation apparatus for speech synthesis is made into an example. explained. However, the speech synthesis data set generation device and the synthesized speech utterance device of the present invention are not limited to being used as a data broadcasting system, but are, for example, a data communication system (not shown) via a network such as the Internet. May be.

  In this data communication system (not shown), the data broadcasting system 1 transmits a voice synthesis data set and update information on a broadcast wave, whereas a voice synthesis data set via a network such as the Internet. And update information. Therefore, the process of creating the speech synthesis data set and the process of creating update information by the speech synthesis data set generation apparatus are the same in any system. The process in which the synthesized speech utterance apparatus utters synthesized speech in accordance with the speech synthesis data set and the process in which the speech database is updated in accordance with the update information are the same in any system.

  For this reason, a data communication system (not shown) can be implemented in the same manner as the configuration and operation of the data broadcasting system of the above embodiment, and thus the description thereof is omitted. However, in the data broadcasting system 1 of the embodiment, the transmission unit 16 transmits as a broadcast wave. However, when the transmission unit 16 is implemented as a data communication system, the transmission unit 16 includes an interface for connecting to various networks. Data transmission may be performed.

  In the above embodiment, since the data broadcasting system has been described, the speech synthesis data set acquisition unit is configured by the reception unit 21 and the data identification unit 22. However, even in the case of a data communication system, the receiving means 21 and the data identifying means 22 can be similarly configured. However, in this case, the receiving means 21 needs to be provided with an interface for connecting to various networks.

  In the data transmitting apparatus 10 of the above embodiment, reading information is acquired from text, and a speech unit based on the reading information is selected. However, raw speech is collected by a microphone and recognized. Reading information may be acquired. In this case, the material text acquisition unit 11 according to the embodiment particularly includes a voice input unit (not shown) that collects and inputs a voice as an electric signal, and a waveform of the voice electric signal input by the voice input unit. What is necessary is just to provide the voice recognition means (not shown) which recognizes as text with reference to various dictionaries. And the reading information acquisition means 121 of the said embodiment should just acquire the text recognized by the speech recognition means, and may acquire reading information similarly to the case of the said embodiment. Note that since the voice recognition means generates reading information during the voice recognition processing, the reading information acquisition means 121 may acquire the reading information generated at this time as it is.

  In the embodiment, the case of content as data broadcasting material including video, audio data, and text has been described. However, as long as it includes text (including recorded audio data) to be uttered as synthesized speech, The content is not limited to the content, and may be a content including a still image or a combination thereof.

  In the data broadcasting system 1 of the embodiment, the provision of the speech synthesis data set and the update information from the data transmission device 10 as the speech synthesis data set generation device to the data reception device 20 as the synthesized speech utterance device is as follows. It was carried out by radio broadcast waves. However, the provision of the speech synthesis data set and the update information from the speech synthesis data set generation device to the synthesized speech utterance device is not limited to this, and the use of wired broadcasting (so-called cable television) facilities, wired / wireless The data transmission path may be used or a wired / wireless telephone communication network may be used. In addition, a portable removable medium storing a speech synthesis data set and update information is provided to a user of the synthesized speech utterance device by a mailing method or the like, and the synthesized speech utterance device (data transmission device) is used to remove the removable media from the removable medium. A voice synthesis data set or update information may be read.

  In particular, in the case of the data broadcasting system 1 and the data communication system (not shown) of the above embodiment, the speech synthesis data set is transmitted from the speech synthesis data set generation device (data transmission device) to the synthesized speech utterance device (data reception device). ) In substantially real time. Also, in the case of using removable media, various contents can be provided by speech of synthesized speech to areas where wired or wireless does not reach. Furthermore, when a movie is stored and provided on a DVD-ROM (Digital Versatile Disk-Read Only Memory), the voice database itself is stored together with the segment information by the voice synthesis data set, so that the actor's Speech can be uttered by synthesized speech. Further, the synthesized speech utterance program itself may be stored in the DVD-ROM, and interactively reproduced by a general computer.

It is a block diagram which shows the structure of the data broadcasting system of embodiment which concerns on this invention. It is a figure explaining the acquisition concept of the reading information by the reading information acquisition means of the data set preparation means for speech synthesis with which a data transmission apparatus is provided. It is a figure explaining the selection concept of the speech unit by the segment selection means of the data set preparation means for speech synthesis with which a data transmission apparatus is provided. It is a figure explaining the concept of the data set for speech synthesis of the speech data set creation means with which a data transmission apparatus is provided. It is a figure explaining the creation concept of the update information at the time of audio | voice database update. It is a figure explaining the concept of the function of the data identification means with which a data receiver is provided. It is a figure explaining the concept which analyzes the data set for speech synthesis. It is a figure explaining the concept of the function of the voice database update means with which a data receiver is provided. It is a figure explaining the concept of the speech synthesis | combination by the speech synthesis means with which a data receiver is provided. It is a figure explaining the concept of the setting of the speech synthesis control information by the user interface with which a data receiver is provided. It is a figure explaining the setting concept of the speech synthesis control information by the speech synthesis control information setting means provided in the data receiving device. It is a flowchart which shows operation | movement of the data transmitter which concerns on embodiment of this invention. It is a flowchart which shows operation | movement of the data receiver which concerns on embodiment of this invention.

Explanation of symbols

1 Data Broadcasting System 10 Data Transmission Device (Data Set Generation Device for Speech Synthesis)
DESCRIPTION OF SYMBOLS 11 Material text acquisition means 12 Speech synthesis data set creation means 121 Reading information acquisition means 122 Segment selection means 123 Speech database 124 Control information creation means 13 Speech synthesis data set addition means (speech synthesis data set provision means)
14 Voice database update information creation means (update information provision means)
15 Multiplexing means (speech synthesis data set providing means, update information providing means)
16 Transmitting means (speech synthesis data set providing means, update information providing means)
20 Data receiver (synthesized speech utterance device)
21 Receiving means (speech synthesis data set acquisition means, update information acquisition means)
22 Data identification means (speech synthesis data set acquisition means, update information acquisition means)
23 voice synthesis control information setting means 231 content item type setting means 232 speaker setting means 233 speech speed setting means 234 dynamic feature control information setting means 235 static feature control information setting means 24 voice database update means 25 voice synthesis means 251 voice Database 26 User interface 27 Voice output means

Claims (8)

Type information for dividing a speech unit group composed of a plurality of speech units of the same voice quality constituting a minimum unit waveform to be uttered as synthesized speech, and corresponding to the speech unit for each speech unit group And at least one type of control information that defines how to deform the waveform of the synthesized speech generated by connecting the speech units associated with each other based on the type information and the unit information. A synthesized speech utterance device that obtains the speech synthesis data set from a speech synthesis data set generation device that generates a synthesis data set and utters it as synthesized speech,
A speech database that stores the speech segment group for each voice quality;
Voice synthesis data set acquisition means for acquiring the voice synthesis data set from the voice synthesis data set generation device;
The speech synthesis data set acquired by the speech synthesis data set acquisition means is received, the type information of the speech synthesis data set is selected to identify the speech segment group to be extracted, and the speech synthesis data set A voice synthesis control information setting means for selecting the control information of the data set to identify the waveform deformation method, and setting the selected type information and the control information and the segment information as voice synthesis data;
The speech synthesis data is received from the speech synthesis control information setting means, the speech segment is extracted according to the segment information from the speech segment group of the speech database specified by the type information, and the extracted speech A speech synthesizer that connects the segments to generate a waveform of a synthesized speech, deforms the waveform in accordance with the waveform modification method specified by the selected control information, and utters the deformed waveform as synthesized speech; ,
A synthesized speech uttering device comprising: The speech synthesis control information setting means is
Content item type setting means for setting the type of the speech element group stored in the audio database according to the type of the content item included in the type information;
Speaker setting means for setting the type of the speech segment group stored in the speech database according to the difference of speakers included in the type information;
The synthesized speech utterance device according to claim 1, comprising at least one of the following. The speech synthesis control information setting means is
A speech speed setting means for setting speech speed control information, which is control information for changing the speech speed of a waveform generated by connecting the speech segments;
Dynamic feature control information setting means for setting dynamic feature control information, which is control information for changing dynamic features that change the movement of the waveform with the elapsed time of the waveform generated by connecting the speech units;
Static feature control information setting means for setting static feature control information, which is control information for changing a static feature that changes the shape of a predetermined time at a predetermined time of a waveform generated by connecting the speech elements;
The synthesized speech utterance device according to claim 1, comprising at least one of the following. Update information acquisition means for acquiring update information for updating the speech segment group stored in the speech database from the speech synthesis data set generation device;
Voice database update means for updating the voice database according to the update information acquired by the update information acquisition means;
The synthesized speech utterance device according to any one of claims 1 to 3, further comprising: Type information for dividing a speech unit group composed of a plurality of speech units of the same voice quality constituting a minimum unit waveform to be uttered as synthesized speech, and corresponding to the speech unit for each speech unit group And at least one type of control information that defines how to deform the waveform of the synthesized speech generated by connecting the speech units associated with each other based on the type information and the unit information. In order to acquire the speech synthesis data set from the speech synthesis data set generation device that generates the synthesis data set and utter as synthesized speech,
Speech synthesis data set acquisition means for acquiring the speech synthesis data set from the speech synthesis data set generation device;
From the speech database that receives the speech synthesis data set acquired by the speech synthesis data set acquisition means, selects the type information of the speech synthesis data set, and stores the speech segment group for each voice quality The speech unit group to be extracted is specified, the control information of the speech synthesis data set is selected to specify the waveform deformation method, and the selected type information, the control information, and the unit information are selected. Speech synthesis control information setting means for setting as data for speech synthesis,
The speech synthesis data is received from the speech synthesis control information setting means, the speech segment is extracted according to the segment information from the speech segment group of the speech database specified by the type information, and the extracted speech A speech synthesizer that connects the segments together to generate a waveform of synthesized speech, deforms the waveform according to the method of waveform deformation specified by the selected control information, and utters the deformed waveform as synthesized speech;
A synthesized speech utterance program characterized by functioning as Type information for dividing a speech unit group composed of a plurality of speech units of the same voice quality constituting a minimum unit waveform to be uttered as synthesized speech, and corresponding to the speech unit for each speech unit group And at least one type of control information that defines how to deform the waveform of the synthesized speech generated by connecting the speech units associated with each other based on the type information and the unit information. A speech synthesis data set generation device that provides the speech synthesis data set to a synthesized speech utterance device that is generated as a synthesis data set and uttered as synthesized speech based on the speech synthesis data set,
A speech database that stores the speech segment group for each voice quality;
Material text acquisition means for acquiring material text to be uttered as synthesized speech;
Reading information acquisition means for acquiring reading information of the text acquired by the material text acquisition means;
A segment selection unit that selects a speech unit of the reading information acquired by the reading information acquisition unit for each type of voice quality from the speech database and sets it as segment information;
Control information creating means for creating the type information and the control information;
A speech synthesis data set providing unit that provides the segment information selected by the unit selection unit and the type information and control information created by the control information creation unit to the synthesized speech utterance device as a speech synthesis data set; ,
A data set generating apparatus for speech synthesis, comprising: Creating update information for updating the speech unit stored in the speech database, and updating speech database update information creating means for updating the speech unit of the speech database with the update information;
Update information providing means for providing the update information created by the voice database update information creating means to the synthesized speech utterance device;
The speech synthesis data set generation device according to claim 6, comprising: Type information for dividing a speech unit group composed of a plurality of speech units of the same voice quality constituting a minimum unit waveform to be uttered as synthesized speech, and corresponding to the speech unit for each speech unit group And at least one type of control information that defines how to deform the waveform of the synthesized speech generated by connecting the speech units associated with each other based on the type information and the unit information. In order to provide the speech synthesis data set to a synthesized speech utterance device that generates as a synthesis data set and utters as synthesized speech based on the speech synthesis data set,
Material text acquisition means for acquiring material text to be uttered as synthesized speech,
Reading information acquisition means for acquiring reading information of the text acquired by the material text acquisition means,
A segment selection unit that selects a speech unit of reading information acquired by the reading information acquisition unit for each type of voice quality from a speech database that stores the speech unit group for each voice quality,
Control information creating means for creating the type information and the control information;
Speech synthesis data set providing means for providing the synthesized speech utterance apparatus with the segment information selected by the segment selection means and the type information and control information created by the control information creation means as a speech synthesis data set;
A data set generation program for speech synthesis, characterized in that

Images