JPS60205597A - Voice synthesizer - Google Patents

Abstract

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

Description

DETAILED DESCRIPTION OF THE INVENTION (a) Technical Field of the Invention The present invention relates to a speech synthesis device that outputs synthesized speech for an arbitrary input word.

(b) Prior art and problems Speech synthesis devices have been used in various fields in recent years.
For example, it is used in a sentence/chapter reading device that outputs an arbitrary input sentence aloud.

Conventionally, in order to synthesize arbitrary sentences, a method has been used in which parameters are provided for each small unit such as a phoneme or syllable, and arbitrary words are synthesized using those parameters, and then arbitrary sentences are synthesized. However, since synthesis methods using phonemes and syllables have parameters in small units, deterioration in sound quality cannot be avoided due to problems such as interpolation characteristics between sounds.

(c) Purpose of the Invention The object of the present invention is to solve the above-mentioned drawbacks and to enable speech synthesis of arbitrary sentences using speech that is extremely close to natural speech.

(d) Structure of the Invention In order to achieve the above object, the present invention is provided with a phoneme syllable etc. parameter storage means for storing phoneme syllables etc. phonetic parameters constituting the reading information of an arbitrary input word. In a speech synthesis device configured to be capable of extracting the corresponding phoneme-syllable parameters from the phoneme-syllable parameter storage means and synthesizing the speech, the speech synthesis device is configured to be able to extract the corresponding phoneme-syllable parameters from the phoneme-syllable parameter storage means and perform speech synthesis. A word/phrase etc. parameter storage means is provided for storing speech parameters corresponding to words, phrases, etc. having a length, and the contents of the word/phrase etc. parameter storage means are used together with the contents of the phoneme/syllable etc. parameter storage means to store an arbitrary input word. The present invention is characterized in that it is configured to perform speech synthesis processing for.

(E) Examples of the invention Hereinafter, the present invention will be explained with reference to the drawings.

FIG. 1 is a block diagram of a speech synthesis device according to an embodiment of the present invention, and in the figure, ■ indicates an arbitrary sentence holding section.

2 is a sentence analysis section, 3 is a word dictionary section, 4 is a parameter conversion section, 5 is a phoneme/syllable parameter storage section, 6 is a word/phrase etc. parameter storage section, 7 is a prosody setting section, and 8 is a speech synthesis section.

The operation of FIG. 1 is as follows.

First, a series of sentences extracted from the arbitrary sentence holding unit 1 is
The text is input to the text analysis section 2. In the text analysis section 2,
Referring to the word dictionary section 3, the input sentence is divided into words and readings are given to the words.

For example, the sentence ``compose arbitrary sentences'' is divided into words such as ``arbitrary,''``sentence,''``wo,''``synthesize,'' and ``do,'' as well as ``nini,'' `` Hmm","
Create columns with the readings ``wo'', ``goosei'', and ``suru''.

Next, the string of pronunciations of these words is processed by the parameter conversion unit 4.
is input to the prosody setting section 7. Here, the prosody setting section 7
creates a pitch pattern for the word based on accent type information, beat count information, grammar information, etc. stored in the i-word dictionary 3 together with the pronunciation information of the word. This prosody setting section 7 is a section that is not directly related to the present invention.

On the other hand, the parameter conversion section 4 uses the phoneme-syllable etc. parameter storage section 5 and the word/phrase etc. parameter storage section 6, and
Parameter conversion processing is performed according to the operational flow shown in the figure.

First, for one input word, the word/phrase/parameter storage unit 6 is referred to to check whether there is a word/phrase/parameter corresponding to the word. If it exists, a parameter time series is created from the corresponding parameters such as words and phrases.

In addition, when there is no word/phrase/etc. parameter corresponding to the i-word in the word/phrase/etc. parameter storage unit 6, after decomposing the word into minute phonemes, syllables, etc., the corresponding A parameter time series is created from parameters such as phonemes and syllables corresponding to phonemes and syllables. In this way, after obtaining a parameter time series for one word and sending it to the speech synthesis unit 8, the next word in the input word string is extracted and the same processing as described above is performed.

Next, the speech synthesis section 8 performs speech synthesis processing based on the pitch pattern from the prosody setting section 7 and the parameter time series from the parameter conversion section 4, and outputs the desired speech.

In the embodiment, the word/phrase/etc. parameter storage unit 6 stores readings of frequently used words, such as one clause or longer sentence units, and the corresponding parameters in pairs. Further, in the phoneme syllable etc. parameter storage section 5, phoneme syllables etc. and parameters corresponding thereto are stored in pairs.

(f) Effects of the Invention According to the present invention, parameters are stored in units of frequently used words and phrases, and when that word appears in a sentence to be synthesized into speech, it is stored in correspondence with that word. By using parameters that have been stored, and only when a word that is not memorized is found, speech synthesis is performed using parameters in trick units that synthesize arbitrary words, so compared to conventional methods, the difference between sounds and sounds is improved. Interpolation processing is reduced and sound quality can be improved.

[Brief explanation of drawings]

FIG. 1 is a block diagram of a speech synthesizer according to an embodiment of the present invention, and FIG. 2 is a diagram showing an operation flow of a parameter conversion section. In Figure 1, ■ is an arbitrary sentence holding unit, 2 is a sentence analysis unit,
3 is a word dictionary section; 4 is a parameter conversion section 15 is a phoneme-syllable parameter storage section; 6 is a word-phrase parameter storage section; 8
is the speech synthesis section. person P weak power '$r old

Claims (1)

[Scope of Claims] (11) A phoneme/syllable/etc. parameter storage means for storing phonetic parameters of phonemes/syllables/etc. constituting the reading information of an arbitrary input word; In a speech synthesis device configured to be able to extract and synthesize speech by extracting parameters such as phonemes, syllables, etc., which are longer than the above-mentioned phonemes, syllables, etc.,
A word/phrase/etc. parameter storage means is provided for storing audio parameters corresponding to words, phrases, etc. having various lengths longer than beats, and the content of the word/phrase/etc. parameter storage means is used together with the contents of the phoneme/syllable/etc. parameter storage means. A speech synthesis device characterized in that it is configured to perform speech synthesis processing on an arbitrary input word. [1#(For certain words etc. in the 21 human words, first,
When a word, etc. parameter corresponding to the word, etc. exists by referring to the word, phrase, etc. parameter storage means, a speech parameter time series is created from the word, etc. parameter, and on the other hand, a speech parameter time series corresponding to the word, etc. is stored in the word, phrase, etc. parameter storage means. It is characterized by being configured to break down the word, etc. into smaller phonemes, syllables, etc. when the word, etc. parameter does not exist, and extract the phoneme, syllable, etc. parameter from the phoneme, syllable, etc. parameter storage means to create a speech parameter time series. A speech synthesis device according to claim 11.