JPH0420996A - Sentence-voice converter - Google Patents

Abstract

PURPOSE:To correctly transmit a document frequently using a written language like a table by providing the mode in which a sentence written in a fixed format is converted to an expression, with which a lister easily understands contents of the sentence, in accordance with this format and it is outputted. CONSTITUTION:A language analyzing part 1 performs language analysis of an input sentence, and a sentence converting part 11 converts the input sentence to an expression easy for the lister to understand in accordance with the format, and a morpheme analyzing part 12 divides the converted sentence into words and give their readings and grammatical information. A rhythm generating part 13 generates rhythm information to output a pronunciation code string, and a voice synthesizing part 3 outputs a synthesized voice based on the pronunciation code string.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a sentence-to-speech conversion device, and more particularly to a language analysis section of the sentence-to-speech conversion device.

Jomai 1: Normally, written language and spoken language are different, so if you use a sentence-to-speech converter to read out the written language as it is, it is often difficult for the listener to understand.Especially, it is difficult to understand a lot of information in a concise manner, such as in a table. In formats intended for expression, there are many abbreviations, symbols, and abbreviations, which makes it difficult to convey information correctly if read aloud. Therefore,
When outputting such a sentence as a voice, it is better to convert the input sentence into an expression that is easy for the listener to understand before outputting it.

However, in terms of converting written words into spoken words and outputting them, conventional sentence-to-speech synthesizers only do things such as adding digits and giving appropriate pronunciations to symbols that represent units when reading out five numbers. If you do not do this, one sentence will become easier to understand in parts, but the format of the entire input sentence is not understood and processed.
It is insufficient to convey information in an easy-to-understand manner to the listener.

The present invention was made in view of the above-mentioned circumstances.
In particular, it was created with the aim of correctly conveying sentences that use a lot of written language, such as tables, to listeners.

In order to achieve the above object, the present invention includes a language analysis unit that performs language analysis on any coded sentence and converts it into a phonetic symbol string containing prosodic information such as pronunciation, accent, and intonation; In a sentence-to-speech conversion device that includes a speech synthesis unit that converts the phonetic symbol string into synthesized speech and outputs the synthesized speech,
It is characterized by having a mode that converts the content of the text into an expression that is easy for the listener to understand according to the format and outputs it.Furthermore, in the mode that converts the content of the text into an expression that is easy to understand and outputs it, Supplementing the content of the input text with appropriate language, or.

It is characterized by replacing words in an input sentence with other words with the same meaning. Hereinafter, the present invention will be explained based on examples.

FIG. 1 is a block diagram for explaining one embodiment of a sentence-to-speech conversion device according to the present invention. In the figure, 1 is a language analysis section, 2 is a phonetic symbol string storage section, 3 is a speech synthesis section, It is assumed that the input to this device is a text that has already been encoded, such as a text sent via teletext, a text created using a word processor, or a text received via personal computer communication. The language analysis unit 1 performs language analysis on an input sentence and outputs a phonetic symbol string including pronunciation information, prosodic information such as accents and pauses, and additional information for controlling speech pronunciation. Either send the phonetic symbol string directly to the speech synthesis section 3, or
It is temporarily stored in the phonetic symbol string storage section 2, and the speech synthesis section 3 reads it as needed. The speech synthesis unit 3 outputs synthesized speech based on the information on this phonetic symbol string.

Figure 2 is a detailed diagram of the language analysis unit 1 shown in Figure 1. In the figure, 11 is a sentence conversion unit, 12 is a morphological analysis unit, 13 is a prosody generation unit, 14 is format information, and 15 is format support. In the dictionary, as shown in the figure, a language analysis unit 1 is connected to a sentence conversion unit 11 that converts an input sentence into an expression that is easy for the listener to understand according to its format, and a sentence conversion unit 11 that divides the converted sentence into words and processes the pronunciation and grammatical information. It consists of a morphological analysis section 12 that provides morphological information, and a prosody generation section 13 that generates prosodic information such as accents and pauses.

In the mode in which the input sentence is converted into an easy-to-understand expression and output, the sentence conversion section 11 converts the input sentence into an expression that is easy to understand for the listener and then sends it to the morphological analysis section 12. In the normal mode, the sentence conversion section 11 is Instead, the input sentence may be sent as is to the morphological analysis unit 12.

A feature of the present invention is that the language analysis section 1 is provided with a text conversion section 11. The text conversion unit 11 will be described in detail below.

The text conversion unit 11 reads the format information 14 corresponding to the input text and supplements the content of the input text with appropriate words.For example, when the input text is as shown in Table 1, Table 1 (weather forecast table) ( 1) Add the item name corresponding to the content before the table content.

(2) Fill in the necessary units for the contents of the table.

(3) Supplement the contents of the table with particles and auxiliary verbs.

By doing this, you can convert it into the following sentence.

(Output after text conversion (underlined words supplemented)
) Tokyobe no J (Nakaharu, ゑ''1u'' student 24-5L, ゑ,, tW middle school 15.5㎜(''Lini hair.

Furthermore, the one-sentence conversion unit 11 replaces the words in the input sentence with other words of the same meaning that are easier to understand by referring to the dictionary 15 that corresponds to the format of the input sentence. For example, if the input sentence is as shown in Table 2, Table 2 (traffic information table) is converted into a sentence as shown below by replacing abbreviations, symbols, etc. with spoken words.

(Output after text conversion (underlined words supplemented)
] 釆五亙speed五藍、traffic jam、6 Saturdays.

Another way to use it is to replace the contents of the table with another language that corresponds to it, so that the input text can be translated and output, or it can be replaced with English and output as shown below.

TOMEI expressway, bumper-
to-bumoer, 6 kilometers.

As is clear from the above description, the present invention involves linguistic analysis of an arbitrary coded sentence and converting it into a string of phonetic symbols containing prosodic information such as pronunciation, accent, and intonation. Department and.

In a sentence-to-speech conversion device that includes a speech synthesizer that converts the phonetic symbol string into synthesized speech and outputs it, the text is written in a standard format and the content of the sentence is conveyed to the listener according to the format. Since it has a mode that converts it into an easy-to-understand expression and outputs it, it becomes possible to correctly convey sentences that use a lot of written words, such as tables, to the writer.

[Brief explanation of the drawing]

FIG. 1 is a configuration diagram for explaining one embodiment of the present invention,
FIG. 2 is a detailed configuration diagram of the language analysis section shown in FIG. 1. 1... Language analysis section, 2... Phonetic symbol string storage section, 3.
...Speech synthesis unit, 11...Text conversion unit, 12...Morphological analysis unit, 13...Prosody generation unit, 14...Format information, 15...Format compatible dictionary.

Claims (1)

[Claims] 1. A language analysis unit that linguistically analyzes an arbitrary coded sentence and converts it into a string of phonetic symbols that includes prosodic information such as pronunciation, accent, and intonation, and converts the string of phonetic symbols into synthesized speech and outputs it. A sentence-to-speech conversion device consisting of a speech synthesis section is equipped with a mode for converting sentences written in a standard format into expressions that are easy for listeners to understand according to the format and outputting the sentences. A sentence-to-speech conversion device characterized by: