JPS58154900A - Sentence voice converter - 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 The present invention relates to a text-to-speech conversion device that converts a text given in the form of text information into speech and utters it.

A text-to-speech conversion device that reads input text aloud has been developed. By the way, such a conventional text-to-speech conversion device is
Both common words and special words such as proper nouns, technical words, and new words are uttered with similar pronunciation characteristics without distinction.

However, while common words and phrases can be easily heard by the general public, people who are not familiar with the above-mentioned special words are likely to miss them. This becomes clear if you imagine a radio broadcast. Radio broadcast announcers read uncommon proper nouns, new words, technical terms, even numbers, dates, etc. at a slower speed, repeat them, or read them more slowly than other common words and phrases. Efforts are being made to help listeners understand the content by changing the language and other methods.

Therefore, an object of the present invention is to provide a text-to-speech conversion device that facilitates listening to proper nouns, new words, technical terms, and the like.

In order to achieve this object, the text-to-speech conversion device according to the present invention has means for identifying a specific type of word in an input text, and according to the identification result by the means (2), the text-to-speech conversion device according to the present invention It is characterized by being constructed so that its pronunciation characteristics are different from those of other types of words.

Hereinafter, the present invention will be described with reference to one embodiment with reference to the drawings.

FIG. 1 is a block diagram of a text-to-speech conversion device according to the present invention.

In the figure, 1 is a text file in which input texts containing kana and kanji are stored in the form of character codes (for example, JIS codes). The input text read from the text file 1 is converted into speech and is uttered by the 5-voice unit 12.

Note that this text file 1 is specifically a storage device such as a magnetic tape device or a magnetic disk device.

3 is a word dictionary file, in which information on various words in kanji and kana are stored in the format shown in Figure 2. Here, W is a word code, C1 is a part-of-speech code indicating the part of speech of the word, and C2 is a reading code indicating the reading of the word. In addition, there are more than two ways to read it.
There are two or more reading codes C2 for a word. Up to this point, the format is the same as the word dictionary file used in conventional text-to-speech conversion devices, but in this embodiment, a classification code C3 has been added. This classification code C
3 indicates whether the corresponding word is a type of word (referred to as a special word) whose pronunciation characteristics should be different from other general words. As the special words, proper nouns such as unusual names of people and places, specialized words, new words, numerals, etc. are selected as necessary.

Returning to FIG. 1, 2 is a search section. This search section 2 is
A known method such as the longest two-clause-dispersion method is used to detect the break between case particles and words in the input sentence, and with reference to this, each word in the input sentence is searched from the word dictionary file 3.

The part-of-speech code -C1 of the searched word is sent to the accent determination section 5, the reading code C2 is sent to the monosyllabic decomposition processing section 6, and the classification code C3 is sent to the pronunciation control section 9.

The configuration of the search section 2 may be similar to the search section of a conventional text-to-speech conversion device. However, the search unit 2 of this embodiment also works as a means for identifying special words. In other words, when searching for a word, the classification code C3 indicating whether or not it is a special word is also read out from the dictionary file 3 at the same time. By making this change, it is possible to identify special words without substantially changing the configuration of the search unit 2.

The monosyllabic decomposition processing unit 6 decomposes the pronunciation of each word into monosyllables according to the phonological rules from the pronunciation code C2 of each word inputted from the search unit 2, and retrieves parameters for each monosyllable from the monosyllabic parameter file 7. and sends it to the combination processing section 8. Furthermore, the monosyllable decomposition processing unit 6 notifies the combination processing unit 8 of the connection or separation between the decomposed individual monosyllables in synchronization with the monosyllable parameters. The combination processing unit 8 performs combination processing (articulation processing) on the single syllable parameters to make the combination between single syllables that are to be pronounced as one continuous voice natural, and sends them to the sound source parameter generation unit 10 .

Note that the monosyllable decomposition processing section 6, monosyllable parameter filter 7, and combination processing section 8 described above may all be the same as those of the conventional device.

4 is an intonation determining section. This intonation determining unit 4 determines the overall intonation of the sentence by determining whether it is a declarative sentence or an interrogative sentence based on the final word of each sentence in the input sentence, for example. (2) Therefore, it is necessary to change the accent and pitch when pronouncing the words in the sentence (especially the final word).
The intonation information is sent from the intonation determining section 4 to the accent determining section 5 and the pronunciation control section 9.

The accent determining section 5 receives the accent from the detecting section 2? The accent of the word to be uttered is determined according to the part of speech code C1 and the intonation information from the intonation determining section 4, and the accent information is transmitted to the pronunciation control section 9.
send to The pronunciation control section 9 determines the duration, pitch, and amplitude, which are the elements that determine the pronunciation characteristics, according to the accent information and intonation information, and outputs the pronunciation characteristics information.

The sound source parameter generation unit 10 generates sound source parameters according to the monosyllabic parameters provided from the combination processing device 8 and pronunciation characteristic information that is modification information thereof. According to the sound source parameters, the voice synthesis section 11 synthesizes a voice signal, and sends it to the voice generation section 12 to generate a voice.

Since the sound source parameters are modified with pronunciation characteristic information,
The pronunciation characteristics, that is, the duration, pitch, and amplitude (volume) of the voice uttered by the voice generating section 12 are controlled according to the pronunciation characteristic information.

As described above, except for the special words, the operations and configurations of the parts 9 to 12 are the same as those of the conventional device. However, when a special word is uttered, that is, when the classification code C3 input to the pronunciation control unit 9 specifies a special word, the pronunciation control unit 9 intentionally changes the pronunciation characteristics determined by the accent information and intonation information, and Control is performed so that words can be clearly distinguished from other common words and phrases. The pronunciation control unit 9 of this embodiment uniformly increases the pitch of the pronunciation characteristics for special words.

Note that the amplitude may also be changed at the same time as the pitch.
This means changing the pronunciation characteristics so that the sound can be heard clearly.

In order to control the pronunciation characteristics of special words in this manner, it is necessary to change the configuration of the pronunciation control section 9 from that of the conventional device. However, such a change in the configuration may be extremely minor and its implementation is easy, so a specific example of the sound generation control section 9 will not be specifically shown.

As explained above, the present invention can be applied to unusual proper nouns, new words, technical terms, and even numerals that are difficult to hear (
Special words) ζSecondly, the pronunciation characteristics such as pitch are intentionally changed to make the words uttered in order to call the listener's attention.

Therefore, according to the present invention, it is possible to provide an excellent text-to-speech conversion device that greatly improves the drawbacks of conventional text-to-speech conversion devices.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention, and FIG. 2 is a diagram showing a code format in a threatening word dictionary file. 1... Sentence file, 2... Search section, 3... Word dictionary file, 4... Intonation determining section, 5.
... Accent determination unit, 6... Monosyllabic decomposition processing unit, 7
. . . Monosyllabic parameter file, 8 .

Claims (1)

[Claims] (1) A text-to-speech conversion device that converts a text input in the form of character information into speech and utters it, which has means for identifying a specific type of word in the input text, and according to the identification result by the means. , a text-to-speech conversion device characterized by making the pronunciation characteristics of the specific type of words different from those of other types of words.