JP2000352990A - Foreign language speech synthesizer - Google Patents

Abstract

(57) [Summary] [PROBLEMS] The present invention is directed to a case where when a Japanese text contains words in a foreign language such as English, the speaker is the same as the Japanese voice, and is close to Japanese. The purpose is to synthesize speech with easy-to-understand pronunciation. An apparatus for converting words written in a foreign language into speech, comprising: a language discriminating unit that discriminates and sorts a foreign language part in an input character sequence; It includes a conversion unit that converts a part into a phonetic symbol sequence and a processing unit that converts a phonetic symbol sequence into a kana sequence.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a foreign language speech synthesizer for converting a sentence written in characters into a foreign language speech.

[0002]

2. Description of the Related Art Conventional speech synthesis has been developed for each specific language such as Japanese, English or German, and has been used as a device for outputting the specific language speech. This is because the basic units that form the speech corresponding to the phonetic symbols, the prosody such as intonation and rhythm differ depending on the language, and it is necessary to develop a synthesis method for each language.

On the other hand, the speech synthesis technology has been used in various fields such as reading out e-mails and reading newspaper articles. In the case of Japanese, it is usually written in a sentence mixed with kanji or kana, but recently, foreign words such as English,
It has been used in Japanese textbooks.

Conventionally, when such a text is converted into speech, a Japanese speech synthesizer is used for Japanese text, and a speech is synthesized using an English speech synthesizer when an English text appears. Can be continuously output.

[0005]

FIG. 1 shows a configuration diagram of a conventional multi-language speech synthesis. In the drawing, reference numeral 1 denotes a language discriminating unit, 2-i denotes a language synthesizer, and 3 denotes a mixing unit.

However, this method has the following two problems. (1) Since the Japanese voice and the foreign language voice are developed based on voices of different persons, the voice quality of the synthesized voice differs between the voice of the Japanese language part and the voice of the foreign language part. As a result, voices of different voice qualities are switched, making it very difficult to hear.

(2) The voice of the foreign language part is synthesized such that a native speaker of the country speaks. However, this makes it difficult for people who listen to the whole Japanese to hear it, and it is hard to understand. Rather, for English, Japanese English
It will be easier to hear if you combine them in English). In particular, when providing information to elderly people by voice, it is considered that it is more acceptable to pronounce in the form of Japanese foreign language notation.

[0008] The present invention has been made to solve the above problems.
If Japanese text contains words written in foreign languages, such as English, make sure that the voice is synthesized by the same speaker as the Japanese voice, with a pronunciation similar to Japanese and easy to understand. The purpose is.

In the present specification, the content of the invention will be described using English as an example, but it goes without saying that other foreign languages can be realized by the same method as in English.

[0010]

[Means to solve the problem] The language of each country usually includes
There is a notation called orthography. However, the notation method does not reflect the actual pronunciation, and there is usually a gap between the two. Pronunciation is international phonetic alphabet (IP
Although it is described by a phonetic symbol such as A), in principle, pronunciation in many languages can be expressed by a common symbol regardless of the type of language. Once a foreign language can be represented by phonetic symbols, it may be converted into similar phonetic symbols in the native language, and this may be used as an input symbol string for speech synthesis in the native language such as Japanese.

The present invention makes use of this point. In the case of the synthesis of English appearing in Japanese, the number of vowels is a, i in Japanese except for long vowels and continuous vowels. , U,
five vowels, e and o, in English

[0012]

[Outside 1]

These are the seven vowels. Also, consonants are

[0014]

[Outside 2]

Some are not used in Japanese. In other words, English has more types of phonetic symbols than Japanese, so it is necessary to assign specific vowels and consonants in English to similar phonetic symbols in Japanese. For example,

[0016]

[Outside 3]

And so on.

For example, an English word (English notation → phonetic symbol →
When expressed in the order of Japanese romaji notation → katakana notation), it is as follows. In this case, when converting from English phonetic symbols to Japanese Roman alphabet notation, it is necessary to replace phonetic symbols.

[0019]

[Outside 4]

As described above, once an English word is converted into a series of phonetic symbols, it can be easily converted to Katakana, depending on the correspondence with Japanese phonetic symbols.

[0021]

FIG. 2 shows a block diagram of the present invention.
In the figure, reference numeral 1 denotes a language discriminating unit, 4 denotes a morphological analysis unit, 5 denotes a reading sequence, 6 denotes a conversion unit to phonetic symbol strings, 7 denotes a kana conversion unit, and 8 denotes a Japanese speech synthesis unit.

Foreign language (English) in Japanese character strings
The unit is determined by the language determining unit 1. The reading of the kanji is determined by the processing of the morphological analysis unit 4, the reading of the kanji is determined by the processing of the morphological analysis unit 4, and the Japanese voice synthesis unit 8 converts the Japanese portion into the continuous speech.

The portion determined to be English is sent to the phonetic symbol string converter 6 to obtain the phonetic symbol string of the corresponding word. In the case of English, the derivation of phonetic symbols includes a group of rules well known as letter-to-sound rules, and can also be realized by a dictionary of phonetic symbols and indexing this dictionary.

When the phonetic symbol string is obtained, the kana conversion unit 7 converts the phonetic symbol into kana according to the correspondence with the Japanese phonetic symbols, and the Japanese voice synthesis unit 8 converts it into continuous speech.

[0025]

As described above, according to the present invention,
When a Japanese text contains a foreign language such as English, and synthesizes the speech from the text, the foreign language speech synthesis has the same voice quality as the Japanese speaker who is the ground sentence, and the pronunciation of the foreign language itself Rather, in order to be realized by foreign language pronunciation,
This has the effect of producing a synthesized sound that is extremely easy for Japanese to hear.

Today, as various information providing services are entering homes, it is expected that text information will be easily converted to voice and heard. Elderly people and those who are not proficient in English, and if you realize foreign languages with Japanese pronunciation, you will be able to hear Sooms and help to understand the whole sentence easily.

[Brief description of the drawings]

FIG. 1 is a configuration diagram showing a conventional foreign language speech synthesis system.

FIG. 2 is a configuration diagram showing an apparatus according to the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Language discrimination part 4 Morphological analysis part 5 Reading sequence 6 Conversion part to phonetic symbol string 7 Kana conversion part 8 Japanese speech synthesis part

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Osamu Mizuno Nippon Telegraph and Telephone Corporation 3-9-1-2 Nishishinjuku, Shinjuku-ku, Tokyo (72) Inventor Yamato Sato 1-3-1 Gotenyama, Musashino City, Tokyo NTT Advanced Technology Co., Ltd. (72) Inventor Masashi Sawada 1-3-1 Gotenyama, Musashino-shi, Tokyo F-term in NTT Advanced Technology Co., Ltd. (reference) 5D045 AA07 AA09 9A001 BB02 BB03 BB04 CZ02 DD11 EE02 HH12 HH14 HH18 HH33 JJ14 JJ72 KK37 KK46 KK62

Claims (1)

[Claims] 1. A device for converting words written in a foreign language into speech, comprising: a language discriminator for discriminating and sorting a foreign language portion in an input character sequence; A foreign language speech synthesizer, comprising: a conversion unit that converts a phonetic symbol sequence into a phonetic symbol sequence; and a processing unit that converts a phonetic symbol sequence into a kana sequence.