JPH0690628B2 - Sentence / voice conversion method - Google Patents

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a sentence / speech conversion method for converting a kanji / kana mixed sentence into speech.

2. Description of the Related Art FIG. 3 shows an outline of a sentence / speech conversion method. Third
In the figure, a kanji / kana mixed sentence input from the keyboard is added with reading and accent in the kanji / kana conversion unit 3 based on the data stored in the kanji / kana conversion dictionary storage unit 2, and a voice synthesis unit 4 is formed as a kana character string. Passed to. The voice synthesizing unit 4 generates voice parameters from the kana character string and the accent based on the data stored in the monosyllabic file storage unit 5, and synthesizes the voice. The kanji / kana mixed sentence input to the kanji / kana conversion unit 3 is subjected to generation of an accent and conversion into a kana character string based on the reading / accent / part-of-speech information of the dictionary of the sentence / speech conversion device. For the character strings of numbers and classifiers, accent generation and conversion to kana character strings are performed based on the reading and accent information of the classifier dictionary. The voice signal synthesized by the voice synthesizer 4 is amplified by the amplifier 6 and output from the speaker 7.

DISCLOSURE OF THE INVENTION Problems to be Solved by the Invention However, in the above-mentioned conventional sentence / speech conversion method, one type of digit reading is Ichi, Ni, San, Yon, Go, Roku, Nana, Hachi, Kyu, Juu ... Can only be read. For example, the kanji-kana mixed sentence in the first sentence example was converted into the kana character string in the second sentence example.

(First sentence example) One book.

(Second sentence example) Ichisatsunohon.

As shown in the first sentence example and the second sentence example, in the conventional sentence-speech conversion method, since proper reading is not given to the number,
There was a problem that the intelligibility of sentences converted into speech was poor.

The present invention eliminates the above-mentioned drawbacks of the conventional example, provides an excellent sentence-speech conversion method with a high degree of intelligibility of a sentence converted into a voice by increasing the reading accuracy rate for a kana-kanji mixed input sentence. The purpose is to do.

Means for Solving the Problems In order to achieve the above object, the present invention provides different rules depending on the type of phoneme at the beginning of a classifier so that the correct reading of the number can be obtained. It is the one.

Operation Therefore, according to the present invention, it is possible to increase the accuracy of reading and improve the intelligibility of a sentence converted into a voice by giving correct readings to character strings of numbers and classifiers. It has the effect that

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

Tables 1 and 2 show tables used in the present embodiment, respectively. Table 1 shows a phoneme transformation rule table of numbers when a Chinese syllabary is attached, and Table 2 shows numbers when a foreign syllabary is attached. It is a phoneme transformation rule table.

FIG. 1 and FIG. 2 show the processing flows according to the phoneme transformation rules shown in Tables 1 and 2, respectively, and by performing processing according to this processing flow, the reading of the number is changed. An example in which the first sentence example is converted by the processing according to the present embodiment is shown in the third sentence example.

(Third sentence example) Issatsunohon.

As described above, according to the present embodiment, correct reading can be given to the character strings of numbers and classifiers, and the accuracy of reading the input sentence is improved, so that the intelligibility of the sentence converted into speech is improved. There is an advantage of improving.

Effect of the Invention As described above, the present invention provides different phoneme transformation rules depending on the type of phoneme at the beginning of the classifier, and changes the reading of the number so that the correct reading can be obtained. Since correct reading can be added to the character string of the and classifier, it is possible to increase the reading accuracy rate for the input sentence and improve the intelligibility of the sentence converted into the voice.

[Brief description of drawings]

FIG. 1 and FIG. 2 are diagrams showing a phoneme transformation processing flow in a sentence / speech conversion method according to an embodiment of the present invention, respectively.
FIG. 3 is a block diagram of the sentence / speech converter. 1 ... keyboard, 2 ... kanji-kana conversion dictionary storage,
3 ... Kanji / Kana conversion unit, 4 ... Voice synthesis unit, 5 ... Single syllable file storage unit, 6 ... Amplification unit, 7 ... Speaker.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (56) References Japanese Pronunciation Accent Dictionary Revised New Edition by Japan Broadcasting Corporation Explanation / Appendix P. 121-127

Claims (1)

[Claims] 1. A sentence characterized by changing the reading of numbers in accordance with different phoneme transformation rules depending on the type of phoneme at the beginning of the classifier when the reading of numbers with a Chinese or foreign classifier is added. -Voice conversion method.