JPS63116198A - Sentence-voice conversion - Google Patents

Abstract

(57) [Abstract] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed Description of the Invention] Industrial Application Field The present invention relates to a text/speech conversion method for automatically converting text input into a word processor into speech, and for comparing the speech with a manuscript or for outputting speech in a message communication. be.

BACKGROUND OF THE INVENTION FIG. 3 shows the configuration of a conventional sentence/speech conversion device. In Fig. 3, 1 is a keyboard for inputting codes of kanji-kana mixed sentences into this device, 2 is a kanji-kana conversion dictionary, 3 is a kanji-kana conversion unit, 4 is a syllable file, 5 is a speech synthesis unit, 6 is the audio output speaker.

A kanji-kana mixed code inputted from a keyboard 1 is searched in a kanji-kana conversion dictionary 2 by a kanji-kana conversion section 3, and based on the information, a reading and an accent type are given, and the code is converted into a kana code string. The speech synthesis section 5 searches the syllable synthesis parameter corresponding to the kana code string from the syllable file 4, and synthesizes speech based on this. The synthesized audio signal is amplified by the amplifier 6 and output from the speaker 7.

The Kanji-Kana conversion dictionary 2 stores information such as Kanji codes, pronunciations, accent types, and parts of speech that constitute words. However, this dictionary did not include pronouns written in hiragana (where, who, etc.). As a result, the accent patterns of clauses containing pronouns written in hiragana were unnatural.

For example, if a sentence like (1st sentence example) is input, (2nd sentence example)
(Example sentence) is converted as follows.

(Example 1) The question is who and where they came from.

(Second Sentence Example) Darega Tokokara/Kitakaga/Mondaides However, when uttered by humans, accent types are given, such as ``furega'' and ``tokokara.''

In conventional sentence/speech conversion methods that do not include information on pronouns written in hiragana in dictionaries, a different accent than when spoken by a human is given, making the sentences less natural and less understandable. It had drawbacks.

Problems to be Solved by the Invention The present invention aims to eliminate the drawbacks of the conventional example described above, and aims to eliminate the unnaturalness of converted speech and improve the intelligibility of sentences.

Means for Solving the Problems In order to achieve the above object, the present invention adds information on pronouns written in hiragana to a dictionary, and searches this dictionary for hiragana that precedes particles. , if identified, performs accent combination on the pronoun and particle.

Effect: Therefore, according to the present invention, sentences including pronouns can be converted into natural and highly intelligible speech.

Example An embodiment of the present invention will be described below.

Accent information for pronouns composed of hiragana such as ``dore'' and ``dare'' is added to the conventional dictionary. For hiragana that precedes a particle, search the hiragana pronoun dictionary. If it is identified as a pronoun, the following particle and accent are combined. For example, in the first sentence, "who" is "that" and "from where" is "
Accent combinations are performed, such as ``7 Kokara''. This result (first sentence example) is accented as in (third sentence example),
available.

(Third Sentence Example>"'i; Ruga 7 Kokara/T Fukaga/Mondaides An example of accent combination rules is shown below.

When a particle follows a pronoun with type 1 accent, such as ``徲roko'' or ``Fre,'' the whole sentence is combined to form an l-type accent.

(Example) When a particle follows an adjective with type 0 accent, such as from who→darekara ``τE'' or ``here'', if the accent type of the following particle is type O, it becomes the entire type 0, When it is a type other than 0, it connects high up to the accent position of the following particle.

(Example) This is → This is this → This is Leela Cho Furi Figure 2 is an explanatory diagram of the accent type, (Sanseido,
(Quoted from Shinmeikai Japanese Dictionary, 3rd Edition) FIG. 1 is a flowchart showing the outline of the accent combination processing of this embodiment.

Effects of the Invention As described above, the present invention adds a dictionary of pronouns written in hiragana and uses that information to combine accents between pronouns and the particles that follow them, so it is possible to obtain a natural accent that is close to human speech. It has this effect.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing an accent combination processing flow of a sentence/speech conversion method according to an embodiment of the present invention, FIG. 2 is an explanatory diagram of an accent type, and FIG. 3 is a block diagram of a sentence/speech conversion apparatus. 1...Keyboard, 2...Kanji/Kana conversion dictionary, 3
... Kanji-kana conversion section, 4... Syllable file, 5...
-Speech synthesis section, 6...amplification section, 7...speaker. Name of agent: Patent attorney Toshio Nakao (1st person)
Figure τNl) Figure 3

Claims (1)

[Claims] A sentence/speech conversion method characterized by combining accents for clauses written in hiragana consisting of pronouns and particles according to rules determined by the type of pronoun.