JP2007127994A - Voice synthesizing method, voice synthesizer, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To prevent a user from missing a specific word spoken by enabling reading out a desired word in a text in a specified voice output mode referring to a word dictionary in which optional voice output modes are registered, and enabling drawing the user's attention when the specific word is read out. <P>SOLUTION: When the text to be output in voice from a text capture part 2 is input, a language analysis part 3 performs morpheme analysis to divide the text into each word, using inter-word conjunction information stored in a basic dictionary 8 storing a default voice output mode of words. Then, an accent phrase is formed from divided each word using an accent joining rule, and voice synthesis is performed by reading out the voice output mode of each word in the accent phrase from a user dictionary 9 in which the optional voice output mode is stored for each word. Then, the accent phrase is output from a voice synthesis output part 6. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a speech synthesizing method, a speech synthesizing apparatus, and a program for reading input text in a speech output style such as a speaker, emotion, speed, volume, speech style, voice pitch, language, etc. registered in a word dictionary.

  Conventionally, there is a tag specification for controlling how to read speech synthesis, such as Speech Synthesis Markup Language (http://www.w3.org/TR/speech-synthesis/). Thus, for example, it is possible to control such that a certain word “concentration” is read in male voice and a certain word “concentration” is read in female voice. In addition, a standard specification that describes word information including reading information, such as the Pronounciation Lexicon Specification (http://www.w3.org/TR/lexicon-reqs/), has been developed.

Further, patent documents relating to the following speech synthesis processing are disclosed. Patent Document 1 discloses a technique for changing the prosody of an important part in a sentence. Patent Document 2 discloses a technique in which a prosodic dictionary and a waveform dictionary that take emotions into consideration are created, and when an input sentence is input together with a task specification, the emotional speech is read out. Furthermore, Patent Document 3 discloses a technique that does not output a specific word in a word dictionary. Furthermore, Patent Document 4 discloses a technique for extracting an adverb such as “very” and transforming it into “very” or “very” depending on the degree.
Japanese Patent Laid-Open No. 11-231885 JP 2001-034282 A JP 2002-221981 JP 2002-311981 A

  However, with the conventional Speech Synthesis Markup Language and other reading tags, it is necessary to insert a tag for each part of the text that you want to read, which is troublesome. In the above-mentioned patent documents 1 and 2, prosody and emotion are specified in the document, but there is a problem that the prosody and emotion cannot be specified by registering the word dictionary. Further, in Patent Document 3, a specific word in the word dictionary is not output by voice, but there is a problem that the reading method cannot be specified by registering the word dictionary. Furthermore, in Patent Document 4, adverbs are transformed and output according to their degree, but there is a problem that arbitrary words cannot be output with the user's desired volume, speed, prosody, and emotion.

  The present invention has been made in view of such circumstances, and an object thereof is to provide the following speech synthesis method, speech synthesis apparatus, and program. That is, referring to a word dictionary in which an arbitrary voice output format is registered, a desired word in the text can be read out in a specified voice output format, and the user's attention can be drawn when a specific word is read out. Therefore, it is possible to prevent missed listening.

In order to solve the above problems, a speech synthesis method according to the present invention includes:
A receiving process for receiving text information;
A dividing step of dividing the text information received in the receiving step into predetermined units;
A speech output format corresponding to a word equal to or included in each predetermined unit obtained in the dividing step is acquired from a word dictionary in which each word and an arbitrary speech output format are stored in association with each other. Acquisition process;
A speech synthesis step of synthesizing and outputting the text information based on the acquired speech output format.

In addition, in order to solve the above problem, a speech synthesizer according to the present invention provides:
Receiving means for receiving text information;
Dividing means for dividing the text information received by the receiving means into predetermined units;
A speech output format corresponding to a word equal to or included in each predetermined unit obtained by the dividing means is acquired from a word dictionary in which each word and an arbitrary speech output format are stored in association with each other. Acquisition means;
Voice synthesis means for synthesizing and outputting the text information based on the acquired voice output format.

  Furthermore, in order to solve the above-mentioned problem, a program according to the present invention is characterized in that the speech synthesis method is executed using a computer.

  According to the present invention, a word dictionary in which an arbitrary voice output format is registered can be referred to and a desired word in a text can be read out in a specified voice output format, and a user's attention when a specific word is read out. Can be prevented and missed listening can be prevented.

  Hereinafter, with reference to the drawings, details of speech synthesis processing by the speech synthesizer according to the embodiment of the present invention will be described.

  FIG. 1 is a block diagram showing a configuration of a speech synthesizer according to an embodiment of the present invention. As shown in FIG. 1, an input device 1 such as a keyboard and a mouse for inputting text and an output device 7 such as a speaker for outputting synthesized speech can be attached to and detached from the speech synthesizer main body 12 according to the present embodiment. It is.

  Then, the speech synthesizer main body 12 uses the text capturing unit 2 that captures the text input from the input device 1, and uses the word information stored in the basic dictionary 8 and the connection information between the words. A language analysis unit 3 for language analysis is provided. Furthermore, the speech synthesizer 12 includes a prosody control unit 4 that generates a prosody of the synthesized speech by adding a reading, an accent, and a pose from the result of language analysis. Furthermore, the speech synthesizer 12 includes a waveform generation unit 5 that connects speech units recorded in the waveform dictionary 11 according to the generated prosody and generates a speech waveform. Furthermore, the speech synthesizer 12 includes a synthesized speech output unit 6 that synthesizes the generated speech waveform.

  Furthermore, the speech synthesizer 12 includes a basic dictionary 8 in which default notations, readings, accents, parts of speech, and usage types are recorded for basic words. Furthermore, the speech synthesizer 12 is provided with a user dictionary 9 that allows the user to arbitrarily specify notation, reading, part of speech, accent, speaker, emotion, reading speed, volume, and effective number of readings for a desired word. ing. Furthermore, the speech synthesizer 12 includes a user dictionary editing unit 10 for editing information registered in the user dictionary 9 using a GUI or newly registering information.

  FIG. 2 is a diagram illustrating an example of words registered in the basic dictionary 8 included in the speech synthesizer according to the present embodiment. In FIG. 2, a reading “Kyo”, an accent “Type 1”, and a part of speech “Noun” are registered for the notation “Today”. The basic dictionary 8 stores connection information between the words. Using the connection information, the input text is morphologically analyzed and divided into words, and readings and accents are given to the words.

  FIG. 3 is a diagram illustrating an example of word information registered in the user dictionary displayed on the user dictionary editing screen in the user dictionary editing unit 10. As shown in FIG. 3, the word “today” is registered in the user dictionary. At this time, the reading is “Kyo”, the part of speech is “Noun”, the accent is “Type 1”, the speaker is “Taro”, the emotion is “Fun”, the reading speed is “Fast”, the volume is “High”, the number of effective times Sets each property like “once”. Each property can be selected and set from a pull-down menu as shown in FIG. In the language analysis, the information in the user dictionary 9 is used with priority over the information in the basic dictionary 8.

  FIG. 4 is a diagram illustrating an example of a property setting screen for words registered in the user dictionary 9. In this case, other properties are set in the GUI for a word whose notation is “today”, reading is “today”, and part of speech is “noun”.

  Here, for example, if the speaker is set to “Hanako”, the word “today” in the text is synthesized by using the speech segment data of “Hanako” from the waveform dictionary. Also, for example, if the emotion is set to “fun”, a prosody corresponding to “fun” is generated in the prosody control unit for the word “today” in the text, and synthesized. Further, for example, if the reading speed is set to “fast”, the word “today” in the text is read a predetermined rate faster than other words. Furthermore, for example, if the volume is set to “high”, the word “today” in the text is read out with a predetermined amount higher than other words.

  Furthermore, if the effective count is set to “once”, the word “today” in the text is read out in the output format specified in the user dictionary 9 only once and appears in the second and subsequent times. If it is, it will be read in the default output format. If the effective count is designated as “once per sentence”, it is read out in the output format designated only once per sentence (in this case, the effective count of the user dictionary is set in step S9 in FIG. 6 described later). There is no need to rewrite).

  FIG. 5 is a diagram for explaining the outline of the process for synthesizing the input text by the speech synthesizer according to the embodiment of the present invention. FIG. 6 is a flowchart for explaining a processing procedure for speech synthesis of input text by the speech synthesizer according to the embodiment of the present invention.

  First, in step S <b> 1, text is input using the input device 1, and the text is input from the text capturing unit 2 into the speech synthesizer main body 12. Here, as shown in FIG. 5A, it is assumed that the text “Today is a good weather” is input. Next, in step S <b> 2, the language analysis unit 3 performs morphological analysis on the input text using the basic dictionary 8 and the user dictionary 9. Here, as shown in FIG. 5B, the words are divided into words using the utilization information and connection information of each word such as “today”, “ha”, “good”, “weather”, and “is”.

  In step S3, the user dictionary 9 is referred to and the number of effective readings of each word is confirmed. Here, it is assumed that the word “today” is specified to be read only once in the user dictionary.

In steps S4 to S5, reading, accent, and speaker / emotion / speed / volume information are assigned to each word. Each word is also given grammatical information such as part of speech and accent combining rules, but is omitted in the following description. Here, “Kyo '(Speaker: Hanako) (Emotion: Fun) (Reading speed: Fast) (Volume: High) (Effective number: 1)”,
“Wa (Speaker: Taro) (Emotion: Normal) (Reading speed: Normal) (Volume: Normal) (Effective frequency: Infinite)”,
“Yo (Speaker: Taro) (Emotion: Normal) (Reading speed: Normal) (Volume: Normal) (Effective frequency: Infinite)”,
“Tenki (Speaker: Taro) (Emotion: Normal) (Reading speed: Normal) (Volume: Normal) (Effective frequency: Infinite)”,
“De'su (Speaker: Taro) (Emotion: Normal) (Reading speed: Normal) (Volume: Normal) (Effective frequency: Infinite)”
Thus, each word is given a reading and accent.

Then, using the accent combination rule, each accent phrase is expressed as “Cho '(Speaker: Hanako) (Emotion: Fun) (Reading speed: Fast) (Volume: Large) (Effective) Number of times: 1) (Speaker: Taro) (Emotion: Normal) (Reading speed: Normal) (Volume: Normal) (Effective frequency: Infinite)
“Yo (Speaker: Taro) (Emotion: Normal) (Reading speed: Normal) (Volume: Normal) (Effective frequency: Infinite)”,
"I'm Tekin (Speaker: Taro) (Emotion: Normal) (Reading speed: Normal) (Volume: Normal) (Effective frequency: Infinite)"
And generate.

  Further, in step S6, prosody according to the accent type and emotion is given to the words in each accent phrase as shown in FIG. Next, in step S <b> 7, speech segment data of a speaker designated for each word (that is, “Hanako” or “Taro”) is acquired from the waveform dictionary 11. Then, the speech unit data for each pitch is connected in the waveform generation unit 5 according to the generated prosody. In step S8, each word is output as a voice at a designated speed and volume. Thereafter, in step S9, the number of effective readings of the word “today” in the user dictionary 9 is reduced by one.

  As described above, in registering the user dictionary, the speaker, emotion, reading speed, volume, and effective number of readings when reading each word can be easily specified, and can be suitably reflected in speech synthesis.

<Example 2>
Using the same device as the above-described speech synthesizer, the entire accent phrase including a word designating a desired reading format may be read out by a desired reading method. That is, using the same text, for example,
"Kyo'wa (speaker: Hanako) (emotion: fun) (reading speed: fast) (volume: loud) (effective number: 1)",
“Yo (Speaker: Taro) (Emotion: Normal) (Reading speed: Normal) (Volume: Normal) (Effective frequency: Infinite)”,
"I'm Tekin (Speaker: Taro) (Emotion: Normal) (Reading speed: Normal) (Volume: Normal) (Effective frequency: Infinite)"
And generate.

<Example 3>
Using a device similar to the above-described speech synthesizer, a word designating a desired reading format may be read out as an independent accent phrase without being accent-coupled. That is, using the same text, for example,
"Kyo '(Speaker: Hanako) (Emotion: Fun) (Reading speed: Fast) (Volume: Large) (Effective number: 1)",
“Wa (Speaker: Taro) (Emotion: Normal) (Reading speed: Normal) (Volume: Normal) (Effective frequency: Infinite)”,
“Yo (Speaker: Taro) (Emotion: Normal) (Reading speed: Normal) (Volume: Normal) (Effective frequency: Infinite)”,
"I'm Tekin (Speaker: Taro) (Emotion: Normal) (Reading speed: Normal) (Volume: Normal) (Effective frequency: Infinite)"
An accent phrase is generated as follows.

<Example 4>
In the first to third embodiments, an output format other than the speaker, emotion, speed, and volume may be set in the user dictionary 9 for each word. For example, utterance style (“announcer style”, “reading style”, “conversational style”, etc.), voice pitch (“very high”, “high”, “normal”, “low”, “very low”) Etc.), language (English words are “American-like”, etc.), etc. can be used.

<Other embodiments>
Although the embodiment has been described in detail above, the present invention can take an embodiment as a system, apparatus, method, program, storage medium (recording medium), or the like. Specifically, the present invention may be applied to a system composed of a plurality of devices, or may be applied to an apparatus composed of a single device.

  In the present invention, a software program (in the embodiment, a program corresponding to the flowchart shown in the drawing) that realizes the functions of the above-described embodiments is directly or remotely supplied to a system or apparatus. In addition, this includes a case where the system or the computer of the apparatus is also achieved by reading and executing the supplied program code.

  Accordingly, since the functions of the present invention are implemented by computer, the program code installed in the computer also implements the present invention. In other words, the present invention includes a computer program itself for realizing the functional processing of the present invention.

  In that case, as long as it has the function of a program, it may be in the form of object code, a program executed by an interpreter, script data supplied to the OS, or the like.

  Examples of the recording medium for supplying the program include the following. Floppy (registered trademark) disk, hard disk, optical disk, magneto-optical disk, MO, CD-ROM, CD-R, CD-RW, magnetic tape, nonvolatile memory card, ROM, DVD (DVD-ROM, DVD-R) .

  As another program supply method, the program can be supplied by downloading it from a homepage on the Internet to a recording medium such as a hard disk using a browser of a client computer. That is, it connects to a homepage and downloads the computer program itself of the present invention or a compressed file including an automatic installation function from the homepage. It can also be realized by dividing the program code constituting the program of the present invention into a plurality of files and downloading each file from a different homepage. That is, a WWW server that allows a plurality of users to download a program file for realizing the functional processing of the present invention on a computer is also included in the present invention.

  Further, the program of the present invention is encrypted, stored in a storage medium such as a CD-ROM, and distributed to users. Then, the user who has cleared the predetermined condition is allowed to download key information for decryption from the homepage via the Internet. It is also possible to execute the encrypted program by using the key information and install the program on a computer.

  Further, the functions of the above-described embodiments are realized by the computer executing the read program. In addition, the function of the above-described embodiment can be realized by an OS running on the computer based on an instruction of the program and performing part or all of the actual processing.

  Further, the functions of the above-described embodiments are realized even after the program read from the recording medium is written in a memory provided in a function expansion board inserted into the computer or a function expansion unit connected to the computer. That is, the functions of the above-described embodiments are realized by performing a part or all of the actual processing by the CPU or the like provided in the function expansion board or function expansion unit based on the instructions of the program.

It is a block diagram which shows the structure of the speech synthesizer which concerns on one Embodiment of this invention. It is a figure which shows an example of the word registered into the basic dictionary 8 with which the speech synthesizer which concerns on this embodiment is provided. It is a figure which shows an example of the word information registered into the user dictionary displayed on the edit screen of the user dictionary in the user dictionary editing part. It is a figure which shows the example of a setting screen of each property about the word registered into the user dictionary. FIG. 5 is a diagram for explaining the outline of the process for synthesizing the input text by the speech synthesizer according to the embodiment of the present invention. It is a flowchart for demonstrating the process sequence which carries out the speech synthesis | combination of the input text by the speech synthesizer which concerns on one Embodiment of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Input device 2 Text capture part 3 Language analysis part 4 Prosody control part 5 Waveform generation part 6 Synthetic speech output part 7 Output device 8 Basic dictionary 9 User dictionary 10 User dictionary edit part 11 Waveform dictionary 12 Speech synthesizer body

Claims (16)

A receiving process for receiving text information;
A dividing step of dividing the text information received in the receiving step into predetermined units;
A speech output format corresponding to a word equal to or included in each predetermined unit obtained in the dividing step is acquired from a word dictionary in which each word and an arbitrary speech output format are stored in association with each other. Acquisition process;
A speech synthesis method comprising: synthesizing and outputting the text information based on the acquired speech output format.   The speech synthesis method according to claim 1, wherein the dividing step divides the text information into accent phrase units.   The speech synthesis method according to claim 1, wherein the dividing step divides the text information into predetermined units by performing morphological analysis.   The speech synthesis method according to claim 1, further comprising a registration step of newly registering a speech output format of a word in the word dictionary.   The speech synthesis method according to claim 1, further comprising an editing step of editing a speech output format of words stored in the word dictionary. A setting step for setting the effective number of voice output styles of words stored in the word dictionary;
In the speech synthesis step, when each word in the text information is speech synthesized and output, if the speech synthesis of each word is within the effective number of times, speech synthesis is performed with reference to the speech output format, 2. The speech synthesis method according to claim 1, wherein speech synthesis is performed with reference to a default speech output format when the effective number of times is not exceeded.   The type of voice output style of the word is any one of speaker, emotion, speech style, voice pitch, speed, volume, language, or a combination of two or more thereof. The speech synthesis method according to 1.   The speech synthesis step reads out the speaker waveform data set in the speech output format from the waveform dictionary storing the waveform data of a plurality of speakers for each word in the text information, and synthesizes each of them. The speech synthesis method according to claim 7.   8. The speech synthesis method according to claim 7, wherein the speech synthesis step synthesizes each word in the text information with reference to emotion prosodic data set in the speech output format. .   The speech synthesis method according to claim 7, wherein the speech synthesis step synthesizes speech for each word in the text information at a speed set in the speech output format.   The speech synthesis method according to claim 7, wherein the speech synthesis step synthesizes each word in the text information at a volume set in the speech output format.   The speech synthesis method according to claim 7, wherein the speech synthesis step synthesizes each word in the text information with a speech style set in the speech output format.   The speech synthesis method according to claim 7, wherein the speech synthesis step synthesizes each word in the accent phrase at a voice level set in the speech output format.   The speech synthesis method according to claim 7, wherein the speech synthesis step synthesizes speech for each word in the text information in a language set in the speech output format. Receiving means for receiving text information;
Dividing means for dividing the text information received by the receiving means into predetermined units;
A speech output format corresponding to a word equal to or included in each predetermined unit divided by the dividing unit is acquired from a word dictionary in which each word and an arbitrary speech output format are stored in association with each other. Acquisition means;
A speech synthesizer comprising: speech synthesizer that synthesizes and outputs the text information based on the acquired speech output format.   The program for performing the speech synthesis method of any one of Claim 1-14 using a computer.

Images