JP2012037726A - Voice synthesizer and computer program - Google Patents

Abstract

Sound data is corrected easily and efficiently.
A control information table storage unit 51 that stores a control information table, a fundamental frequency analysis unit 31 that generates time change information of a fundamental frequency of voice data, a speech rate calculation unit 32 that calculates a speech rate value, The control condition setting unit 52 for setting the control method type and the key information for specifying the control amount of the speech speed or the change amount of the fundamental frequency, and the control information table associated with the control method type are read A control information acquisition unit 53 that acquires the control amount of the speech speed and the control amount of the change width of the fundamental frequency based on the information, the speech speed value, and the time change information of the fundamental frequency; Based on the target basic frequency generation unit 54 for generating the time change information of the target basic frequency from the time change information of the basic frequency based on the control amount of the speech speed and the time change information of the target basic frequency. And a voice data correction unit 80 to which a positive to.
[Selection] Figure 1

Description

  The present invention relates to a speech synthesizer and a computer program that correct prosody of input speech.

  In order to convert voice data such as synthesized voice into voice data that is easy to listen to, it is effective to change the acoustic feature quantity related to prosody. Then, for example, pitch, power, and duration are controlled as acoustic features related to the prosody, and the sound quality is improved by changing the dynamic range of the time variation of the pitch or power or pitch pattern or changing the pause length. A technique for changing is known (see, for example, Patent Document 1).

JP-A-10-97267

In order to convert the sound of the sound data into a sound that is easy to hear, it is effective to correct a plurality of types of acoustic feature values related to the prosody of the sound data. However, in order to adjust multiple types of acoustic features and bring them closer to the desired prosody, you need extensive expertise in sound and rich adjustment experience, such as the relationship between acoustic features and the relationship between acoustic features and ease of listening. is necessary. In addition, it is very complicated and complicated to perform such adjustment work with human hands.
Therefore, the present invention has been made in view of the above circumstances, it is possible to correct the audio data without having specialized knowledge and rich adjustment experience, and it is possible to easily and effectively obtain easy-to-listen audio data. An object is to provide a speech synthesizer and a computer program.

[1] In order to solve the above-described problem, a speech synthesizer according to an aspect of the present invention provides a control information table in which a control amount of speech speed and a control amount of a change amount of a fundamental frequency are associated with each other in a control method. Basic frequency analysis that generates time change information of a fundamental frequency based on the audio data acquired by the audio data acquisition unit, a control information table storage unit that stores the data in association with a type, an audio data acquisition unit , A speech speed calculation unit for calculating a speech speed value based on the voice data, and a key for designating a desired control method type and a control amount of a speech speed or a change amount of a fundamental frequency in a control information table A control condition setting unit for setting information, and a control information table associated with the type of the control method set by the control condition setting unit is read from the control information table storage unit, Based on the key information set by the control condition setting unit, the speech speed value calculated by the speech speed calculation unit, and the time change information of the fundamental frequency generated by the fundamental frequency analysis unit, A control information acquisition unit that acquires a control amount of a change width of the fundamental frequency, and a target basic frequency from the time change information of the fundamental frequency based on the control amount of the change width of the fundamental frequency acquired by the control information acquisition unit. Based on the target fundamental frequency generation unit that generates the time change information of the target, the control amount of the speech speed acquired by the control information acquisition unit, and the time change information of the target basic frequency generated by the target basic frequency generation unit An audio data correction unit that corrects the audio data and outputs the corrected audio data.
Here, the speech speed control amount is correction data or data representing a correction amount for the speech data correction unit to correct the speech speed, which is the prosody acoustic feature amount. Further, the control amount of the change width of the fundamental frequency is data representing the correction data or the correction amount for correcting the change width of the fundamental frequency, which is the acoustic feature quantity of the prosody, in other words, correcting the sound range.
According to this configuration, the speech synthesizer according to one aspect of the present invention can handle either one of the control amounts by designating either the control amount of the speech speed or the control amount of the change width of the fundamental frequency. The other control amount to be acquired is acquired from the control information table storage unit. Then, the speech synthesizer generates time change information of the target basic frequency based on the acquired control amount of the change width of the basic frequency, and the acquired control amount of the speech speed and the generated time change information of the target basic frequency, Based on the above, the prosody related to the speech speed and the change width of the fundamental frequency of the voice data is corrected.
Therefore, according to the speech synthesizer which is an aspect of the present invention, speech data is easily and efficiently corrected into corrected speech data.

[2] In the speech synthesizer according to [1], the absolute value of the speech speed is set as a control amount of the speech speed, the absolute value of the change width of the basic frequency is set as a control amount of the change width of the basic frequency, and the control information table The storage unit stores a control information table in which an absolute value of a change frequency of a fundamental frequency is associated with each absolute value of a plurality of speech speed values in association with a type of control method.
[3] In the speech synthesizer described in [1] above, the control information table storage unit includes a speech speed magnification as a speech speed control amount, a fundamental frequency change width as a fundamental frequency change amount control amount, and Stores a control information table in which a magnification of a change rate of a fundamental frequency is associated with each of a plurality of speech speed magnifications in association with a type of control method, and the control condition setting unit is configured to store a desired magnification for the key information. The control information acquisition unit further includes the key information set by the control condition setting unit, the magnification, the speech speed value calculated by the speech speed calculation unit, and the basic frequency generated by the fundamental frequency analysis unit. Based on the time change information of the frequency, the control amount of the speech speed and the control amount of the change width of the fundamental frequency are acquired.
[4] In the speech synthesizer according to [3] above, the control information table storage unit may multiply the multiples of the plurality of speech speeds by the multiples of the basic frequency corresponding to the plurality of designated speech speed values. The associated control information table is stored in association with the type of control method, and the control information acquisition unit is related to the type of control method set by the control condition setting unit and calculated by the speech speed calculation unit. A control information table corresponding to a designated speech speed value that is the same as the speech speed value is read from the control information table storage unit, and the key information, the magnification, the speech speed value, and the fundamental frequency set by the control condition setting unit Based on the time change information of the fundamental frequency generated by the analysis unit, the control amount of the speech speed and the control amount of the change width of the fundamental frequency are acquired.
[5] In the speech synthesizer described in [1] above, the control information table storage unit may calculate a basic frequency change width that is a control amount of a basic frequency change amount to a speech speed magnification that is a control amount of the speech speed. A control information table in which magnifications are associated is stored in association with a control method type, the control condition setting unit sets a desired control method type, and the control information acquisition unit is stored in the control condition setting unit. A control information table associated with the set type of the control method is read from the control information table storage unit, and a control amount of speech speed and a control amount of a change width of a fundamental frequency are acquired from the control information table. And

  [6] In order to solve the above-described problem, the computer program according to one aspect of the present invention provides a control information table in which a control amount of speech speed and a control amount of a change width of a fundamental frequency are associated with each other by a control method type A computer having a control information table storage unit stored in association with audio data acquisition means for acquiring audio data, and basic for generating time change information of a fundamental frequency based on the audio data acquired by the audio data acquisition means A frequency analysis means, a speech speed calculation means for calculating a speech speed value based on the voice data, a type of a desired control method, and a control amount of a speech speed or a control amount of a change amount of a fundamental frequency in a control information table Control condition setting means for setting key information, and a control information table associated with the type of control method set by the control condition setting means. Based on the key information set by the control condition setting means, the speech speed value calculated by the speech speed calculation means, and time change information of the fundamental frequency generated by the fundamental frequency analysis means, read from the information table storage unit Control information acquisition means for acquiring the control amount of the speech speed and the control amount of the change width of the fundamental frequency, and the basic frequency based on the control amount of the change width of the fundamental frequency acquired by the control information acquisition means Target fundamental frequency generating means for generating time change information of the target fundamental frequency from the time change information of the above, the control amount of the speech speed acquired by the control information acquiring means, and the target basic frequency generated by the target basic frequency generating means Based on the time change information of the frequency, the sound data is corrected and the sound data correcting means for outputting the corrected sound data is made to function.

  According to the present invention, it is possible to correct audio data without having specialized knowledge about sound and rich adjustment experience, and it is possible to easily and efficiently obtain audio data that is easy to listen to.

It is a block diagram which shows the function structure of the speech synthesizer which is 1st Embodiment of this invention. It is a figure which shows the example of the text data showing phoneme division | segmentation time information. It is a figure which shows the example of the control information table which matched the absolute value of the change width of a fundamental frequency with the absolute value of speech speed value. It is a figure which shows the example of the control information table which matched the magnification of the change width of a fundamental frequency with each magnification of several speech speed. It is a figure which shows the example of the control information table which matched the magnification | multiplying_factor of the change width of a fundamental frequency with each of the magnification | multiplying of a several speech speed corresponding to each of three designated speech speed values. It is a figure which shows typically the time change information of the target fundamental frequency which a target fundamental frequency production | generation part produces | generates based on the control amount of the change width of a fundamental frequency. 4 is a flowchart showing a processing procedure of prosody correction processing of the speech synthesizer in the embodiment. It is a block diagram which shows the function structure of the speech synthesizer which is 2nd Embodiment of this invention. It is a figure which shows the example of the control information table in which the control information for hearing-impaired people was stored. It is a figure which shows the example of the control information table in which the control information for quick listening was stored. It is a figure which shows the example of the control information table in which the control information for rapid mouths was stored. It is a figure which shows the example of the control information table for noise environments. 4 is a flowchart showing a processing procedure of prosody correction processing of the speech synthesizer in the embodiment.

Hereinafter, embodiments for carrying out the present invention will be described in detail with reference to the drawings.
[First Embodiment]
FIG. 1 is a block diagram showing a functional configuration of a speech synthesizer according to the first embodiment of the present invention. As shown in the figure, the speech synthesizer 100 includes a speech data acquisition unit 10, a speech attribute information acquisition unit 20, a speech analysis unit 30, a control condition information generation unit 40, a target value generation unit 50, and a control. An information table update unit 60, a generation condition acquisition unit 70, and an audio data correction unit 80 are provided.

  The voice data acquisition unit 10 acquires voice data supplied from the outside, and supplies the acquired voice data to the voice analysis unit 30. The voice data is digital voice data obtained by sampling and quantizing an analog voice signal at a sampling frequency. For example, the voice data is recorded by a recording device (not shown) or voiced by a voice synthesis device (not shown). This is the synthesized voice data that has been synthesized. The audio data is, for example, PCM (Pulse Code Modulation) data, and the format is, for example, WAVE.

The voice attribute information acquisition unit 20 receives and acquires voice attribute information related to the voice data supplied to the voice data acquisition unit 10, and supplies the voice attribute information to the voice analysis unit 30. The voice attribute information is text information in which speech content of voice data is converted into text, or phoneme separation time information of voice data.
When the utterance content of the voice data is “Good morning (blank) how are you? (Blank)”, the text information is, for example, “Good morning. How are you?” Note that “(blank)” indicates that there is a section that is not uttered between the preceding and following sounds. In addition to the hiragana notation, the sentence included in the text information may be any katakana notation, romaji notation, or kanji mixed notation. Further, in the present embodiment, a case where the utterance content of the voice data is Japanese will be described, but the present embodiment can also be applied to a foreign language.
The phoneme separation time information is information indicating the utterance time of each phoneme of the voice data when the time corresponding to the head of the voice of the voice data is set as the reference time. A phoneme is a minimum unit of sound in phonology, and each vowel and consonant corresponds to one phoneme. In addition, each of the repellent sound, the long sound, and the prompt sound corresponds to one phoneme. A specific example of the phoneme separation time information will be described later.

  The voice analysis unit 30 takes in the voice data supplied from the voice data acquisition unit 10 and the voice attribute information supplied from the voice attribute information acquisition unit 20, and analyzes the voice data and the voice attribute information to analyze the voice. Time change information and speech speed value of the fundamental frequency (hereinafter simply referred to as “fundamental frequency”) are obtained and supplied to the target value generation unit 50.

The voice analysis unit 30 includes a fundamental frequency analysis unit 31 and a speech speed calculation unit 32 as functional configurations.
The fundamental frequency analysis unit 31 generates time change information of the fundamental frequency based on the speech data in a voiced section (speeched speech section that is speech accompanied by vocal cord vibration), and for the generated fundamental frequency value, By performing the smoothing process using a spline function or the like using the analysis value of the voiced section, the time change information of the fundamental frequency with a smooth change is generated. At this time, the fundamental frequency analysis unit 31 uses the fundamental frequency value in the voiced section before and after the unvoiced section for an unvoiced section where the fundamental frequency cannot be extracted (a section of unvoiced speech that is a voice not accompanied by vocal cord vibration). Using the inserted value, the value of the fundamental frequency in the unvoiced interval is interpolated.
The fundamental frequency analyzing unit 31 may use the generated fundamental frequency value itself as time change information of the fundamental frequency without performing the smoothing process.

When the speech attribute information is text information, the speech speed calculation unit 32 calculates the speech speed value of the speech data by one of the following two methods.
The method for calculating the first speech speed value is a method for generating phoneme segmentation time information based on speech data and text information and calculating the speech rate based on the generated phoneme segmentation time information. Specifically, the speech speed calculation unit 32 obtains the total time length of the speech section from the speech data, and generates phoneme separation time information by a known speech recognition processing technique based on the total time length and text information. Then, the speech speed calculation unit 32 obtains the speech speed value (unit: “Mora / second”) by obtaining the number of mora per unit time from the phoneme break time information. Mora is a phonological unit of sound length. In Japanese, two kana characters generally correspond to 1 mora for stuttering, and one kana character corresponds to 1 mora for straight sounds. One mora is composed of one or a plurality of phonemes.
The method for calculating the second speech speed value is to obtain the total time length of the speech section from the speech data and the number of mora from the text information, and based on the total time length of the speech section and the number of mora. Is a method of calculating

  When the speech attribute information is phoneme break time information, the speech speed calculation unit 32 obtains the speech speed value by obtaining the number of mora per unit time from the phoneme break time information.

The control condition information generation unit 40 receives supply of a desired control method type, a desired control parameter, and a desired correction purpose type by a user's operation of an input device such as a keyboard or a mouse. Control condition information including the control parameter and the correction purpose type is generated and supplied to the target value generation unit 50.
The type of the control method is data that distinguishes the control method between the control amount of the speech speed and the control amount of the change width of the fundamental frequency. The speech speed control amount is correction data for the speech data correction unit 80 to correct the speech speed that is the first acoustic feature amount of the prosody or data representing the correction amount, and the absolute value of the target speech speed value. , Or expressed as a rate of speech speed. The control amount of the change width of the fundamental frequency is data representing the correction data or the correction amount for correcting the change width of the fundamental frequency, which is the second acoustic feature quantity of the prosody, in other words, correcting the sound range. , The absolute value of the change width of the fundamental frequency, or the magnification of the change width of the fundamental frequency.
The type of control method will be specifically described. The type of control method in the present embodiment is a control method for quantitatively specifying a control amount, and is classified as shown in Table 1 below.

  In Table 1, the control method 1 is a method of designating the control amount of the speech speed and the control amount of the change width of the fundamental frequency by numerical absolute values. The control method 2 is a method for designating the control amount of the speech speed and the control amount of the change width of the fundamental frequency by a magnification. The control method 3 is a method of designating the control amount of the speech speed and the control amount of the change width of the fundamental frequency by a magnification corresponding to the speech speed value (designated speech speed value) supplied from the voice analysis unit 30. . The type of the control method is data associated with each of the control methods 1 to 3.

  The control parameter is key information used by the target value generation unit 50 to acquire the control amount of the speech speed and the control amount of the change width of the fundamental frequency, and is specifically as follows. The control parameter corresponding to the control method 1 is information for designating either the control amount of the speech speed or the control amount of the change width of the fundamental frequency. The control parameters corresponding to the control method 2 and the control method 3 are information specifying either the control amount of the speech speed or the control amount of the change width of the fundamental frequency, and a desired magnification.

  The type of correction purpose is information that specifies the purpose of prosody control, in other words, information that specifies the correction tendency between the control amount of the speech speed and the control amount of the change amount of the fundamental frequency acquired by the target value generation unit 50. . Specifically, the correction purpose type is, for example, information for designating “for hearing-impaired people”, “for quick listening”, and “for quick mouth”.

  The target value generation unit 50 takes in the speech speed value supplied from the voice analysis unit 30 and time change information of the fundamental frequency, and controls the amount of speech speed according to the control condition information supplied from the control condition information generation unit 40. And the control amount of the change width of the fundamental frequency. Further, the target value generation unit 50 generates time change information of the target fundamental frequency based on the control amount of the change width of the fundamental frequency. Then, the target value generation unit 50 supplies the speech data correction unit 80 with the control amount of the speech speed and the time change information of the target fundamental frequency.

  Next, the target value generation unit 50 will be described in detail. The target value generation unit 50 includes a control information table storage unit 51, a control condition setting unit 52, a control information acquisition unit 53, and a target fundamental frequency generation unit 54 as functional configurations.

The control information table storage unit 51 stores a plurality of control information tables in which the control amount of the speech speed and the control amount of the change width of the fundamental frequency are associated with each other in association with the control method type and the correction purpose type. The control information table includes a table in which the absolute value of the change rate of the fundamental frequency is associated with the absolute value of the speech speed value, a table in which the magnification of the change rate of the fundamental frequency is associated with each of the multiple rates of speech speed, and a plurality of Corresponding to each of the designated speech speed values, there are three types, a table in which the magnification of the fundamental frequency is associated with each of a plurality of speech speed magnifications. Specific examples of the various control information tables will be described later.
The control information table storage unit 51 is a readable and writable storage device, such as a magnetic hard disk device or a semiconductor storage device. The control information table storage unit 51 may be a portable recording medium such as a magneto-optical disk.

  The control condition setting unit 52 takes in the control condition information supplied from the control condition information generating unit 40, and sends the control method type, control parameter, and correction purpose type included in the control condition information to the control information acquisition unit 53. Set.

  The control information acquisition unit 53 reads from the control information table storage unit 51 a control information table associated with the control method type and the correction purpose type set by the control condition setting unit 52. Then, the control information acquisition unit 53 converts the read control information table, the control parameters set by the control condition setting unit 52, and the time change information of the speech speed value and the fundamental frequency respectively supplied from the voice analysis unit 30. Based on this, the control amount of the speech speed and the control amount of the change width of the fundamental frequency are obtained. Further, the control information acquisition unit 53 supplies the control amount of the speech speed to the audio data correction unit 80 and supplies the control amount of the change width of the basic frequency to the target basic frequency generation unit 54.

  Specifically, the control condition setting unit 52 controls the type of control method (for example, control method 1), the control parameter (for example, key information is information for specifying the control amount of speech speed), and the type of correction purpose ( For example, when “for the hearing impaired” is set, the control information acquisition unit 53 sets the control information table (the absolute value of the speech speed value to the basic frequency change width) associated with the control method 1 and for the hearing impaired. A table for the hard of hearing who is associated with an absolute value) is read from the control information table storage unit 51. Then, according to the control parameter, the control information acquisition unit 53 uses the absolute value of the speech speed value supplied from the voice analysis unit 30 as a key, and the absolute value of the change width of the basic frequency that is the control amount associated with the key. The value is extracted from the control information table. Then, the control information acquisition unit 53 uses the absolute value of the speech speed value, which is a key, as the control amount of the speech speed, and uses the absolute value of the change frequency of the basic frequency extracted from the control information table as the control amount of the basic frequency change width. To do.

  In addition, the control condition setting unit 52 controls the type of control method (for example, control method 1), the control parameter (for example, key information is information specifying the control amount of the change width of the fundamental frequency), and the type of correction purpose ( For example, when “for the hearing impaired” is set, the control information acquisition unit 53 sets the control information table (the absolute value of the speech speed value to the basic frequency change width) associated with the control method 1 and for the hearing impaired. A table for the hard of hearing who is associated with an absolute value) is read from the control information table storage unit 51. Then, according to the control parameter, the control information acquisition unit 53 uses the absolute value of the change width of the fundamental frequency obtained based on the time change information of the fundamental frequency supplied from the speech analysis unit 30 as a key and is associated with the key. The absolute value of the speech speed value, which is the control amount, is extracted from the control information table. Then, the control information acquisition unit 53 uses the absolute value of the change width of the fundamental frequency as a key as the control amount of the change width of the fundamental frequency, and uses the absolute value of the speech speed value extracted from the control information table as the control amount of the speech speed. To do.

  Further, the control condition setting unit 52 corrects the type of control method (for example, control method 2) and the control parameter (for example, key information is information for specifying the control amount of speech speed and the magnification is also specified) and correction. When the target type (for example, for fast listening) is set, the control information acquisition unit 53 controls the control information table associated with the control method 2 and for fast listening (basic for each of multiple magnifications of speaking speed). A table for quick listening associated with a frequency change width magnification) is read from the control information table storage unit 51. Then, the control information acquisition unit 53 extracts the magnification of the change width of the fundamental frequency, which is a control amount associated with the key, from the control information table using the speech speed magnification based on the control parameter as a key. Then, the control information acquisition unit 53 sets the magnification of the speech speed based on the control parameter as the control amount of the speech speed, and sets the magnification of the change width of the fundamental frequency extracted from the control information table as the control amount of the change width of the fundamental frequency.

  Further, the control condition setting unit 52 specifies the control method type (for example, control method 2) and the control parameter (for example, key information is information for specifying the control amount of the change width of the fundamental frequency, and the magnification is also specified. ) And the type of correction purpose (for example, for fast listening), the control information acquisition unit 53 controls the control information table (multiple of speaking speeds) associated with the control method 2 and for fast listening. A table for quick listening, each of which is associated with a magnification of the change width of the fundamental frequency, is read from the control information table storage unit 51. Then, the control information acquisition unit 53 uses the magnification of the change width of the fundamental frequency based on the control parameter as a key, and extracts the speech speed magnification, which is the control amount associated with the key, from the control information table. Then, the control information acquisition unit 53 uses the magnification of the change width of the fundamental frequency based on the control parameter as the control amount of the change width of the fundamental frequency, and sets the magnification of the speech speed extracted from the control information table as the control amount of the speech speed.

  Further, the control condition setting unit 52 corrects the type of control method (for example, control method 3) and the control parameter (for example, key information is information for specifying the control amount of speech speed, and the magnification is also specified). When the target type (for example, for fast mouth) is set, the control information acquisition unit 53 is associated with the control method 3 and for fast mouth and has the same speech speed value supplied from the speech analysis unit 30. A control information table storage unit 51 stores a control information table of a certain designated speech speed value (a table for a quick mouth in which a magnification of a change width of a fundamental frequency is associated with each of a plurality of speech speed magnifications corresponding to a designated speech speed value). Read from. Then, the control information acquisition unit 53 extracts the magnification of the change width of the fundamental frequency, which is a control amount associated with the key, from the control information table using the speech speed magnification based on the control parameter as a key. Then, the control information acquisition unit 53 sets the magnification of the speech speed based on the control parameter as the control amount of the speech speed, and sets the magnification of the change width of the fundamental frequency extracted from the control information table as the control amount of the change width of the fundamental frequency.

  Further, the control condition setting unit 52 specifies the control method type (for example, control method 3) and the control parameter (for example, key information is information for specifying the control amount of the change width of the fundamental frequency, and the magnification is also specified. ) And the type of correction purpose (for example, for fast mouth), the control information acquisition unit 53 is associated with the control method 3 and for fast mouth and the speech speed value supplied from the speech analysis unit 30 A control information table of a designated speech speed value that is the same as (a table for a quick mouth in which a magnification of a change rate of a fundamental frequency is associated with each of a plurality of speech speed magnifications corresponding to the designated speech speed value) Read from the storage unit 51. Then, the control information acquisition unit 53 uses the magnification of the change width of the fundamental frequency based on the control parameter as a key, and extracts the magnification of the speech speed value, which is a control amount associated with the key, from the control information table. Then, the control information acquisition unit 53 uses the magnification of the change width of the fundamental frequency based on the control parameter as the control amount of the change width of the fundamental frequency, and sets the magnification of the speech speed extracted from the control information table as the control amount of the speech speed.

The target fundamental frequency generation unit 54 determines the time of the target fundamental frequency from the time change information of the fundamental frequency generated by the fundamental frequency analysis unit 31 based on the control amount of the change width of the fundamental frequency supplied from the control information acquisition unit 53. Change information is generated and supplied to the audio data correction unit 80. At this time, the target fundamental frequency generation unit 54 generates time change information of the target fundamental frequency according to the generation condition information stored therein.
The generation condition information is information for specifying a condition when the target fundamental frequency generation unit 54 generates time change information of the target fundamental frequency. The generation condition information includes an average value fixed designation, a minimum value fixed designation, and a designated value fixed designation. Either is specified.

The average value fixed designation is such that the average value is the same as the average value of the frequency values in the predetermined section of the time change information of the fundamental frequency generated by the fundamental frequency analysis unit 31, and the time change information of the target fundamental frequency in the section Is specified.
The minimum value fixed designation is based on the minimum value of the frequency value in the predetermined section of the time change information of the basic frequency generated by the basic frequency analysis unit 31, and the target basic value in the section is set so that the frequency value is equal to or higher than the minimum value. This specifies that the time change information of the frequency is generated.
There are the following two types of fixed value specification. The first specified value fixed designation generates time change information of the target fundamental frequency in the section based on any one frequency value in the predetermined section of the time change information of the fundamental frequency generated by the fundamental frequency analysis unit 31. Is to specify.
In addition, the second specified value fixed designation designates that time change information of the target fundamental frequency in the section is generated with reference to two frequency values corresponding to the start point and the end point of the predetermined section, respectively. is there.

  When receiving the generation condition information from the generation condition acquisition unit 70, the target fundamental frequency generation unit 54 rewrites the generation condition information stored therein with the generation condition information.

The target fundamental frequency generation unit 54 is supplied from the generation condition acquisition unit 70 when the control amount of the change width of the fundamental frequency supplied from the control information acquisition unit 53 is “absolute value of the change frequency of the fundamental frequency”. In accordance with the generation condition information, the time variation information of the target fundamental frequency whose frequency variation width matches the “absolute value of the fundamental frequency variation width” is generated.
Further, the target fundamental frequency generation unit 54 is supplied from the generation condition acquisition unit 70 when the control amount of the change width of the fundamental frequency supplied from the control information acquisition unit 53 is “a magnification of the change width of the fundamental frequency”. According to the generation condition information, the time change information of the target fundamental frequency that matches the change width of the frequency with the “multiplier of the change width of the fundamental frequency” is generated.

The control information table update unit 60 receives a control information table by a user's operation of an input device such as a keyboard or a mouse, and newly writes the control information table in the control information table storage unit 51 or uses the control information table. The control information table that has already been stored is rewritten.
The generation condition acquisition unit 70 designates one of an average value fixed designation, a minimum value fixed designation, a first designated value fixed designation, and a second designated value fixed designation by an operation of an input device such as a keyboard or a mouse by a user. The generation condition information indicating the designation is generated and supplied to the target fundamental frequency generation unit 54.

The audio data correction unit 80 takes in the control amount of the speech speed supplied from the control information acquisition unit 53 and the time change information of the target basic frequency supplied from the target basic frequency generation unit 54, respectively, and the audio data acquisition unit 10 Corrects the prosody of the acquired voice data and outputs the corrected voice data.
Specifically, for example, the audio data correction unit 80 corrects the time change information of the basic frequency by applying the time change information of the target basic frequency to the audio data. Then, the voice data correction unit 80 corrects the speech speed value calculated by the speech speed calculation unit 32 based on the control amount of the speech speed (absolute value of the speech speed value or the multiplication factor of the speech speed), thereby separating the phoneme of the speech data. The time information is corrected based on the corrected speech speed value. Then, the voice data correction unit 80 generates corrected voice data in which the time axis direction is corrected by applying the corrected phoneme segmentation time information to the voice data in which the time change information of the fundamental frequency is corrected.

FIG. 2 is a diagram illustrating an example of text data representing phoneme segmentation time information. As shown in the figure, the phoneme break time information is text data composed of a plurality of lines, and each line corresponds to a phoneme. The data from the first line to the 17th line of the phoneme separation time information is time information indicating the phoneme breakage for the “(blank) good morning (blank)” portion in the voice data.
Each row of phoneme separation time information has three columns separated by a delimiter (for example, a space or a tab). The first column shows the time from the reference time to the end point of each phoneme, with the time from the reference time to the end point of each phoneme. Is expressed in units of 1 / 10,000 second, and the third column indicates phoneme labels of phonemes. For example, in the figure, “0 2740 sil” indicates that the period from the reference time until 0.274 seconds elapses is a silent section. Further, “2740 3168 o (o)” represents that the phoneme from the time point when 0.274 seconds have elapsed from the reference time to the point when 0.3168 seconds have elapsed from the reference time is “o (o)”. . Note that the phoneme label “sil” indicates that there is no phoneme.
Here, the case where the time is in units of 1 / 10,000 seconds has been described as an example. However, the time may be expressed in other units such as 1 / 1000th of a second (millisecond).

  Each of FIGS. 3A to 3C is a diagram illustrating an example of a control information table in which the absolute value of the fundamental frequency is associated with the absolute value of the speech speed value. Each of the control information tables in FIGS. 9A to 9C is associated with a control method type indicating the control method 1. In these control information tables, the absolute value of the speech speed value is the control amount of the speech speed, and the absolute value of the change width of the fundamental frequency is the control amount of the change width of the fundamental frequency. FIG. 6A shows a table for the hearing impaired person, and is associated with the type of correction purpose indicating “for the hearing impaired”. FIG. 5B is a table for quick listening, and is related to the type of correction purpose indicating “for quick listening”. FIG. 6C is a quick mouth table, which is related to the type of correction purpose indicating “for fast mouth”.

  Each of FIGS. 4A to 4C is a diagram illustrating an example of a control information table in which a magnification of a change width of a fundamental frequency is associated with each of a plurality of magnifications of speech speed. Each of the control information tables in FIGS. 9A to 9C is associated with a control method type indicating the control method 2. In these control information tables, the magnification of the speech speed value is a control amount of the speech speed, and the magnification of the change width of the fundamental frequency is a control amount of the change width of the fundamental frequency. FIG. 6A shows a table for the hearing impaired person, and is associated with the type of correction purpose indicating “for the hearing impaired”. FIG. 5B is a table for quick listening, and is related to the type of correction purpose indicating “for quick listening”. FIG. 6C is a quick mouth table, which is related to the type of correction purpose indicating “for fast mouth”.

  Each of FIGS. 5A to 5C is an example of a control information table in which the magnification of the change frequency of the fundamental frequency is associated with each of a plurality of magnifications of the speech speed corresponding to each of the three designated speech speed values. FIG. Each of the control information tables in FIGS. 9A to 9C is associated with a control method type indicating the control method 3. In these control information tables, the magnification of the speech speed value is a control amount of the speech speed, and the magnification of the change width of the fundamental frequency is a control amount of the change width of the fundamental frequency. FIG. 6A shows a table for the hearing impaired person, and is associated with the type of correction purpose indicating “for the hearing impaired”. FIG. 5B is a table for quick listening, and is related to the type of correction purpose indicating “for quick listening”. FIG. 6C is a quick mouth table, which is related to the type of correction purpose indicating “for fast mouth”. FIGS. 9A, 9B, and 9C show examples in which the designated speech speed values are 6, 8, and 10 (mora / second).

  The example of the control information table from FIG. 3 to FIG. 5 is an example in the case where the speech speed value of the voice data supplied from the speech speed calculation unit 32 is about 7 mora / second to 8 mora / second.

FIG. 6 is a diagram schematically illustrating the time change information of the target fundamental frequency generated by the target fundamental frequency generation unit 54 based on the control amount of the change width of the fundamental frequency supplied from the control information acquisition unit 53.
FIG. 6A shows an example of time change information of the target fundamental frequency generated by the target fundamental frequency generation unit 54 when the generation condition information is information indicating that the average value is fixed. FIG. 4A shows the time change information 1 of the fundamental frequency of the audio data in the time from the time t1 to the time t2 (t1 <t2), and the time change information 1 of the fundamental frequency as a control amount of the change width of the fundamental frequency. Based on, the time change information 1a of the expanded fundamental frequency and the time change information 1r of the reduced fundamental frequency are shown. In FIG. _6A , the average value of the frequency values from time t1 to time t2 of the time change information 1 of the fundamental frequency is an average value f _ave . As shown in FIG. 9A, the target fundamental frequency generation unit 54 sets the time change information of the fundamental frequency so that the average value of the time change information of the fundamental frequency from time t1 to time t2 becomes the average value f _ave. 2 is enlarged or reduced.

FIG. 6B is an example of the time change information of the target fundamental frequency generated by the target fundamental frequency generation unit 54 when the generation condition information is information indicating that the minimum value is fixed. FIG. 6B shows the time change information 2 of the fundamental frequency of the audio data in the time from the time t1 to the time t2 (t1 <t2), and the time change information 2 of the fundamental frequency as the control amount of the change width of the fundamental frequency. The basic frequency time change information 2a expanded and the reduced basic frequency time change information 2r are shown. In FIG. 6B, the minimum value of the frequency value from the time t1 to the time t2 of the time change information 2 of the basic frequency is the minimum value _fmin . As shown in FIG. 4B, the target fundamental frequency generation unit 54 expands or reduces the time change information 2 of the fundamental frequency from time t1 to time t2 with reference to the minimum frequency value _fmin. .

FIG. 6C is an example of the time change information of the target fundamental frequency generated by the target fundamental frequency generator 54 when the generation condition information is information indicating the first specified value fixed designation. FIG. 4C shows the time change information 3 of the fundamental frequency of the audio data in the time from time t1 to time t2 (t1 <t2), and the time change information 3 of the fundamental frequency as the control amount of the change width of the fundamental frequency. The basic frequency time change information 3a expanded and the reduced basic frequency time change information 3r are shown. As shown in FIG. (C), the target fundamental frequency generating unit 54, a time change information 3 of the fundamental frequency from the time t1 to the time t2, the enlarged or reduced based on the arbitrary frequency value f _s .

FIG. 6D is an example of the time change information of the target fundamental frequency generated by the target fundamental frequency generator 54 when the generation condition information is information indicating the second specified value fixed designation. FIG. 4D shows the time change information 4 of the fundamental frequency of the audio data in the time from the time t1 to the time t2 (t1 <t2), and the time change information 4 of the fundamental frequency as the control amount of the change width of the fundamental frequency. The basic frequency time change information 4a expanded and the reduced basic frequency time change information 4r are shown. As shown in FIG. 4D, the target fundamental frequency generation unit 54 uses the frequency value f _s1 at time t1 and the frequency value f _{s2 at} time t2 (t1 <t2) as references, from time t1 to time t2. The basic frequency time change information 4 is enlarged or reduced.

Next, the operation of the speech synthesizer 100 according to the first embodiment will be described. FIG. 7 is a flowchart showing a processing procedure of prosody correction processing of the speech synthesizer 100.
First, in step S101, when the control condition information generation unit 40 receives supply of a desired control method type, a desired control parameter, and a correction purpose type by a user's operation of an input device such as a keyboard or a mouse, Control condition information including the type of control method, control parameter, and type of correction purpose is generated and supplied to the target value generation unit 50.
Next, in step S102, the control condition setting unit 52 of the target value generating unit 50 takes in the control condition information supplied from the control condition information generating unit 40, and the type and control parameter of the control method included in the control condition information. And the type of correction purpose are set in the control information acquisition unit 53.
Next, in step S <b> 103, when the audio data acquisition unit 10 receives the supply of the audio data, the audio data acquisition unit 10 takes in the audio data and supplies it to the audio analysis unit 30. Further, when receiving the voice attribute information related to the voice data supplied to the voice data acquiring unit 10, the voice attribute information acquiring unit 20 takes in the voice attribute information and supplies it to the voice analyzing unit 30.

  Next, in step S104, the voice analysis unit 30 takes in the voice data supplied from the voice data acquisition unit 10 and the voice attribute information supplied from the voice attribute information acquisition unit 20, respectively. The information is analyzed, and the time change information and the speech speed value of the fundamental frequency of the voice are obtained and supplied to the target value generation unit 50. Specifically, the fundamental frequency analysis unit 31 of the speech analysis unit 30 generates time change information of the fundamental frequency based on the speech data and supplies it to the target value generation unit 52. When the speech attribute information is text information, the speech speed calculation unit 32 of the speech analysis unit 30 generates phoneme segmentation time information based on the speech data and the text information, and based on the generated phoneme segmentation time information. The speech speed is calculated (method of calculating the first speech speed value), or the total time length of the speech section is obtained from the speech data and the number of mora is obtained from the text information, and the total time length and the number of mora of these speech sections are obtained. Based on the above, the speech speed value is calculated (method of calculating the second speech speed value). When the speech attribute information is phoneme break time information, the speech speed calculation unit 32 obtains the speech speed value by obtaining the number of mora per unit time from the phoneme break time information.

  Next, in step S <b> 105, the control information acquisition unit 53 reads from the control information table storage unit 51 the control information table associated with the control method type and the correction purpose type set by the control condition setting unit 52. . Next, the control information acquisition unit 53 reads the read control information table, the control parameters set by the control condition setting unit 52, the time change information of the speech speed value and the fundamental frequency supplied from the voice analysis unit 30, respectively. Based on the above, the control amount of the speech speed and the control amount of the change width of the fundamental frequency are obtained. Next, the control information acquisition unit 53 supplies the control amount of the speech speed to the audio data correction unit 80 and supplies the control amount of the change width of the basic frequency to the target basic frequency generation unit 54.

Next, in step S <b> 106, the target fundamental frequency generation unit 54 changes time of the fundamental frequency generated by the fundamental frequency analysis unit 31 based on the control amount of the variation width of the fundamental frequency supplied from the control information acquisition unit 53. Time change information of the target fundamental frequency is generated from the information and supplied to the audio data correction unit 80. At this time, the target fundamental frequency generation unit 54 generates time change information of the target fundamental frequency according to the generation condition information stored therein.
Next, in step S107, the audio data correction unit 80 obtains the control amount of the speech speed supplied from the control information acquisition unit 53 and the time change information of the target basic frequency supplied from the target basic frequency generation unit 54, respectively. When captured, the prosody of the audio data acquired by the audio data acquisition unit 10 is corrected and the corrected audio data is output.

As described above, the speech synthesizer 100 according to the first embodiment designates one of the control amount of the speech speed and the control amount of the change width of the fundamental frequency and the correction purpose by the operation of the input device by the user. By doing so, the other control amount corresponding to the designated one control amount can be easily and efficiently acquired in accordance with the correction purpose. Then, the speech synthesizer 100 sets the prosody related to the speech speed and fundamental frequency variation of the speech data in accordance with the user's correction purpose based on the acquired speech speed control amount and fundamental frequency variation control amount. Can be corrected.
Therefore, according to the speech synthesizer 100 according to the present embodiment, it is possible to easily and efficiently correct the sound data to the corrected sound data that is easy to listen without requiring specialized knowledge about sound and rich adjustment experience. .

[Second Embodiment]
FIG. 8 is a block diagram showing a functional configuration of the speech synthesizer according to the second embodiment of the present invention. In the present embodiment, the same components as those of the speech synthesizer 100 according to the first embodiment are denoted by the same reference numerals, and the description thereof is omitted. The speech synthesizer 100a shown in the figure changes the control condition information generation unit 40 and the target value generation unit 50 in the configuration of the speech synthesizer 100 to a control condition information generation unit 40a and a target value generation unit 50a, respectively. Is.

The control condition information generation unit 40a receives supply of a desired control method type by the operation of an input device such as a keyboard or a mouse by the user, generates control condition information including the control method type, and generates a target value generation unit 50a.
The type of control method in this embodiment will be specifically described. The type of control method in the present embodiment is a control method for qualitatively specifying a control amount, and is classified as shown in Table 2 below.

In Table 2, the control method A is a method for designating “slow” for the speech speed of the speech data and for “enlarging” the change width of the fundamental frequency of the speech data. Specifically, for example, the control method A specifies that the speech speed value supplied from the voice analysis unit 30 is between about 0.5 times and about 0.9 times, and from the voice analysis unit 30. This is a method of designating that the change width of the fundamental frequency obtained based on the time change information of the supplied fundamental frequency is between about 1.15 times and about 1.05 times.
This control method A is a control method suitable for providing an easy-to-listen sound with a slow speed and an extended sound range to an elderly person with senile deafness or another deaf person.

In Table 2, the control method B is a method of designating that the speech speed of the voice data is “fast” and that the change width of the fundamental frequency of the voice data is “slightly reduced”. Specifically, for example, the control method B specifies that the speech speed value supplied from the voice analysis unit 30 is between about 1.5 times and about 2.5 times, and from the voice analysis unit 30, This is a method for designating that the change width of the fundamental frequency obtained based on the supplied time change information of the fundamental frequency is between about 0.97 times and about 0.9 times.
This control method B is a control method suitable for providing sound for quick listening with a speed that is high enough to be heard and with a slightly narrower sound range.

In Table 2, the control method C is a method of designating that the speech speed of the voice data is “slightly faster” and that the change width of the fundamental frequency of the voice data is “reduced”. Specifically, for example, the control method C specifies that the speech speed value supplied from the voice analysis unit 30 is between about 1.1 times and about 1.5 times, and from the voice analysis unit 30, This is a method of designating that the change width of the fundamental frequency obtained based on the time change information of the supplied fundamental frequency is between about 0.96 and about 0.8 times.
This control method C is a control method suitable for providing a fast-speech voice with a narrow speed range that does not give a sense of incongruity.

In Table 2, the control method D is a method of designating to “fix” the speech speed of the audio data and to “enlarge” the change width of the fundamental frequency of the audio data. Specifically, for example, the control method D specifies that the speech speed value supplied from the voice analysis unit 30 is not changed, and the basic method obtained based on the time change information of the fundamental frequency supplied from the voice analysis unit 30. This is a method for designating the frequency change width to be between about 1.1 times and about 1.2 times.
This control method D is a control method suitable for expanding the sound range without changing the speech speed and providing a voice that can be easily heard in a noise / noise environment, for example.

  The target value generation unit 50a takes in the speech speed value supplied from the voice analysis unit 30 and the time change information of the fundamental frequency, and controls the amount of speech speed according to the control condition information supplied from the control condition information generation unit 40a. And the control amount of the change width of the fundamental frequency. Further, the target value generation unit 50a generates time change information of the target fundamental frequency based on the control amount of the change width of the fundamental frequency. Then, the target value generation unit 50a supplies the speech data correction unit 80 with the control amount of the speech speed and the time change information of the target fundamental frequency.

  Next, the target value generation unit 50a will be described in detail. The target value generation unit 50a includes a control information table storage unit 51, a control condition setting unit 52, and a control information acquisition unit 53 in the target value generation unit 50 of the speech synthesizer 100, a control information table storage unit 51a, and a control condition setting unit. 52a and the control information acquisition unit 53a.

The control information table storage unit 51a stores a plurality of control information tables in which the control amount of the speech speed is associated with the control amount of the change width of the fundamental frequency in association with the type of control method. The control information table is a table in which the magnification of the fundamental frequency is associated with the magnification of the speech frequency for each correction purpose (for example, for the hearing impaired, fast listening, fast mouth, and noise environment). A specific example of the seed control information table for each correction purpose will be described later.
The control information table storage unit 51a is a readable / writable storage device, such as a magnetic hard disk device or a semiconductor storage device. The control information table storage unit 51a may be a portable recording medium such as a magneto-optical disk.

  The control condition setting unit 52a takes in the control condition information supplied from the control condition information generating unit 40a, and sets the type of the control method included in the control condition information in the control information acquisition unit 53a.

  The control information acquisition unit 53a reads the control information table associated with the control method type set by the control condition setting unit 52a from the control information table storage unit 51a. Then, the control information acquisition unit 53a supplies the control amount of the speech speed stored in the read control information table to the voice data correction unit 80, and the control amount of the change width of the fundamental frequency stored in the control information table. This is supplied to the target fundamental frequency generation unit 54.

  Specifically, for example, when the type of control method (control method A) is set by the control condition setting unit 52a, the control information acquisition unit 53a controls the control information table (for the deaf person) associated with the control method A. Is read from the control information table storage unit 51a. Then, the control information acquisition unit 53a sets the magnification of the speech speed and the magnification of the change frequency of the fundamental frequency stored in the read control information table as the control amount of the speech speed and the change frequency of the fundamental frequency, respectively. .

  For example, when a control method (control method B) is set by the control condition setting unit 52a, the control information acquisition unit 53a stores a control information table (control information for quick listening) associated with the control method B. Table) is read from the control information table storage unit 51a. Then, the control information acquisition unit 53a sets the magnification of the speech speed and the magnification of the change frequency of the fundamental frequency stored in the read control information table as the control amount of the speech speed and the change frequency of the fundamental frequency, respectively. .

  For example, when the control method (control method C) is set by the control condition setting unit 52a, the control information acquisition unit 53a stores the control information table associated with the control method C (control information for fast exit is stored). Table) is read from the control information table storage unit 51a. Then, the control information acquisition unit 53a sets the magnification of the speech speed and the magnification of the change frequency of the fundamental frequency stored in the read control information table as the control amount of the speech speed and the change frequency of the fundamental frequency, respectively. .

  For example, when the control method (control method D) is set by the control condition setting unit 52a, the control information acquisition unit 53a controls the control information table (table for noise environment) associated with the control method D. Read from the information table storage unit 51a. Then, the control information acquisition unit 53a sets the speech speed magnification to 1 to obtain the speech speed control amount, and uses the fundamental frequency change width magnification stored in the read control information table as the fundamental frequency change width control amount. To do.

FIG. 9 is a diagram illustrating an example of a control information table in which control information for the hearing impaired person is stored. The control information table for the hearing-impaired person shown in the figure includes designation for increasing the speech speed of voice data by 0.7 and designation for increasing the change width of the fundamental frequency of voice data by 1.15.
FIG. 10 is a diagram illustrating an example of a control information table in which control information for quick listening is stored. The control information table for quick listening shown in the figure includes a designation for multiplying the speech speed of voice data by 2.0 and a designation for multiplying the change width of the fundamental frequency of voice data by 0.93.
FIG. 11 is a diagram illustrating an example of a control information table in which control information for a quick mouth is stored. The control information table for the quick mouth shown in the figure includes a designation for multiplying the speech speed of voice data by 1.4 and a designation for multiplying the change width of the fundamental frequency of voice data by 0.84.
FIG. 12 is a diagram illustrating an example of a control information table for a noisy environment. The control information table for a noisy environment shown in the figure includes a designation for fixing the speech speed of voice data (1.0 times) and a designation for multiplying the change width of the fundamental frequency of voice data by 1.15. Yes.
Note that the control information tables in FIGS. 9 to 12 are examples in which the speech speed value of the voice data supplied from the speech speed calculation unit 32 is about 7 mora / second to 8 mora / second.

Next, the operation of the speech synthesizer 100a according to the second embodiment will be described. FIG. 13 is a flowchart showing the processing procedure of the prosody correction processing of the speech synthesizer 100a. In the present embodiment, the same processes as those of the speech synthesis apparatus 100 according to the first embodiment are denoted by the same reference numerals and the description thereof is omitted.
First, in step S101a, when the control condition information generation unit 40a receives supply of a desired control method type by an operation of an input device such as a keyboard or a mouse by the user, the control condition information generation unit 40a displays control condition information including the control method type. Generated and supplied to the target value generating unit 50a.
Next, in step S102a, the control condition setting unit 52a of the target value generation unit 50a takes in the control condition information supplied from the control condition information generation unit 40a, and sets the type of the control method included in the control condition information as control information. Set in the acquisition unit 53a.

  In step S105a, the control information acquisition unit 53a reads the control information table associated with the control method type set by the control condition setting unit 52a from the control information table storage unit 51a. Next, the control information acquisition unit 53a supplies the control amount of the speech speed stored in the read control information table to the audio data correction unit 80, and the control amount of the change width of the fundamental frequency stored in the control information table is obtained. This is supplied to the target fundamental frequency generation unit 54.

As described above, the speech synthesizer 100a according to the second embodiment controls the speech speed corresponding to the correction purpose by designating the type of the control method according to the correction purpose by the operation of the input device by the user. The amount and the control amount of the change width of the fundamental frequency can be acquired easily and efficiently. Then, the speech synthesizer 100a determines the prosody related to the speech speed and the change frequency of the fundamental frequency based on the acquired control amount of the speech speed and the control amount of the change frequency of the fundamental frequency according to the user's correction purpose. Can be corrected.
Therefore, according to the speech synthesizer 100a according to the present embodiment, it is possible to easily and efficiently correct the sound data to the corrected sound data that does not require specialized knowledge and rich adjustment experience related to sound. .

  Each of the speech synthesizer 100 according to the first embodiment and the speech synthesizer 100a according to the second embodiment may be configured not to include the speech attribute information acquisition unit 20. In that case, the speech rate calculation unit 32 of the speech analysis unit 30 performs the following processing to calculate the speech rate value of the speech data. That is, the speech speed calculation unit 32 generates text information by applying a known speech recognition process from the speech data, or generates phoneme separation time information from the speech data. Then, as described in detail in the first embodiment, the speech rate calculation unit 32 generates phoneme segmentation time information based on the voice data and the text information, and calculates the speech rate based on the generated phoneme segmentation time information. (A method for calculating the first speech speed value) or obtaining the total time length of the speech section from the speech data and obtaining the number of mora from the text information, and based on the total time length of the speech section and the number of mora The speech speed value is calculated (method of calculating the second speech speed value). Alternatively, the speech speed calculation unit 32 obtains the speech speed value by obtaining the number of mora per unit time from the phoneme break time information.

  In addition, the control information table storage unit 51a of the speech synthesizer 100a according to the second embodiment sets a plurality of control information tables in which the rate of speech speed and the rate of change of the fundamental frequency are associated with each other as the type of control method. The data was stored in association with each other. In addition to this, the control information table storage unit 51a stores a plurality of control information tables in which the absolute value of the speech speed value and the absolute value of the change width of the fundamental frequency are associated with each other in association with the type of the control method. You may do it.

  Further, some functions of the speech synthesizer 100, 100a may be realized by a computer. In this case, a program for realizing the function may be recorded on a computer-readable recording medium, and the program recorded on the recording medium may be read into a computer system and executed. Here, the “computer system” includes an OS (Operating System) and hardware of peripheral devices. The “computer-readable recording medium” refers to a portable recording medium such as a flexible disk, a magneto-optical disk, an optical disk, and a memory card, and a storage device such as a hard disk built in the computer system. Furthermore, the “computer-readable recording medium” dynamically holds a program for a short time like a communication line when transmitting a program via a network such as the Internet or a communication line such as a telephone line. In this case, a volatile memory in a computer system serving as a server or a client in that case may be included and a program that holds a program for a certain period of time may be included. Further, the above program may be for realizing a part of the functions described above, or may be realized by a combination with the program already recorded in the computer system. .

  As mentioned above, although embodiment of this invention was explained in full detail with reference to drawings, the concrete structure is not restricted to this embodiment, The design etc. of the range which does not deviate from the summary of this invention are included.

DESCRIPTION OF SYMBOLS 10 Voice data acquisition part 20 Voice attribute information acquisition part 30 Voice analysis part 31 Fundamental frequency analysis part 32 Speech speed calculation part 40, 40a Control condition information generation part 50, 50a Target value generation part 51, 51a Control information table storage part 52, 52a Control condition setting unit 53, 53a Control information acquisition unit 54 Target fundamental frequency generation unit 60 Control information table update unit 70 Generation condition acquisition unit 80 Audio data correction unit

Claims (6)

A control information table storage unit that stores a control information table in which a control amount of speech speed is associated with a control amount of a change width of a fundamental frequency in association with a type of control method;
An audio data acquisition unit for acquiring audio data;
A fundamental frequency analysis unit that generates time change information of a fundamental frequency based on the speech data acquired by the speech data acquisition unit;
A speech speed calculator that calculates a speech speed value based on the voice data;
A control condition setting unit that sets a type of a desired control method and key information that specifies a control amount of a speech speed or a control amount of a change width of a fundamental frequency in a control information table;
A control information table associated with the type of control method set by the control condition setting unit is read from the control information table storage unit, and the key information set by the control condition setting unit and the speech speed calculation unit are calculated. Based on the speech speed value and the time variation information of the fundamental frequency generated by the fundamental frequency analysis unit, a control information acquisition unit that acquires a control amount of the speech speed and a control amount of the variation range of the fundamental frequency;
A target fundamental frequency generation unit that generates time change information of a target fundamental frequency from time change information of the fundamental frequency based on a control amount of a change width of the fundamental frequency acquired by the control information acquisition unit;
Based on the control amount of the speech speed acquired by the control information acquisition unit and the time change information of the target basic frequency generated by the target basic frequency generation unit, the audio data is corrected and corrected audio data is obtained. An audio data correction unit to output;
A speech synthesizer comprising: The absolute value of the speech speed is set as the control amount of the speech speed, the absolute value of the change width of the basic frequency is set as the control amount of the change width of the basic frequency, and the control information table storage unit stores each of the absolute values of the plurality of speech speed values. The speech synthesizer according to claim 1, wherein a control information table in which absolute values of change widths of fundamental frequencies are associated is stored in association with the type of control method. The rate of speech speed is defined as the amount of control of speech rate, the rate of change in the basic frequency is defined as the control amount of the rate of change in basic frequency, and the control information table storage unit changes the fundamental frequency for each of multiple rates of speech speed. A control information table in which the magnification of the width is associated is stored in association with the type of control method,
The control condition setting unit further sets a desired magnification for the key information,
The control information acquisition unit includes the key information set by the control condition setting unit, the magnification, the speech speed value calculated by the speech speed calculation unit, and time change information of the fundamental frequency generated by the fundamental frequency analysis unit. The speech synthesizer according to claim 1, further comprising: acquiring a control amount of a speech speed and a control amount of a change width of a fundamental frequency based on The control information table storage unit associates a control information table in which a multiple of a plurality of speech speeds is associated with a magnification of a change width of a fundamental frequency in association with each of a plurality of designated speech speed values, and is associated with a type of control method. Remember,
The control information acquisition unit is a control information table corresponding to a designated speech speed value related to the type of the control method set by the control condition setting unit and the same as the speech speed value calculated by the speech speed calculation unit From the control information table storage unit, based on the key information set by the control condition setting unit, the magnification, the speech speed value, and the time change information of the fundamental frequency generated by the fundamental frequency analysis unit, The speech synthesizer according to claim 3, wherein a speech rate control amount and a control amount of a fundamental frequency change width are acquired. The control information table storage unit includes a control information table in which a magnification of the basic frequency that is the control amount of the fundamental frequency is associated with a magnification of the speech speed that is the control amount of the speech speed, and a type of control method Remember and associate with
The control condition setting unit sets a type of a desired control method,
The control information acquisition unit reads a control information table associated with the type of the control method set by the control condition setting unit from the control information table storage unit, and controls a speech rate control amount and a basic frequency from the control information table. The speech synthesizer according to claim 1, further comprising: obtaining a control amount of a change width of A computer including a control information table storage unit that stores a control information table in which a control amount of a speech speed and a control amount of a change width of a fundamental frequency are associated with each other,
Audio data acquisition means for acquiring audio data;
Fundamental frequency analysis means for generating time change information of the fundamental frequency based on the voice data acquired by the voice data acquisition means;
Speaking speed calculation means for calculating a speaking speed value based on the voice data;
A control condition setting means for setting a type of a desired control method and key information indicating a control amount of speech speed or a control amount of a change width of a fundamental frequency in the control information table;
The control information table associated with the type of the control method set by the control condition setting means is read from the control information table storage unit, and the key information set by the control condition setting means and the speech speed calculation means are calculated. Control information acquisition means for acquiring the control amount of the speech speed and the control amount of the change width of the fundamental frequency based on the speech speed value and the time change information of the fundamental frequency generated by the fundamental frequency analysis means;
Target basic frequency generation means for generating time change information of the target fundamental frequency from time change information of the fundamental frequency based on a control amount of the change width of the fundamental frequency acquired by the control information acquisition means;
Based on the control amount of the speech speed acquired by the control information acquisition unit and the time change information of the target basic frequency generated by the target basic frequency generation unit, the audio data is corrected and corrected audio data is obtained. Audio data correction means for outputting;
Computer program to function as.

Images