JP3122097B2 - Document reading control device and document reading method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Summary] The present invention relates to a text-to-speech control device and a text-to-speech method for synthesizing speech in accordance with coded Japanese character data, and to a punctuation mark of the next line when a line break of a text is not separated by a punctuation mark. For the purpose of synthesizing an accurate voice by processing as a sentence, if the punctuation is not delimited in the line feed of the sentence, the corresponding final character string data is temporarily stored, and the first punctuation that appears next By combining the final character string data with the character string data up to the data to form character string data between punctuation marks, and performing speech synthesis in accordance with the character string data between the punctuation marks, it is possible to prevent inappropriate character string data breaks. It is configured to perform accurate text reading aloud.

The present invention relates to a text-to-speech control device and a text-to-speech method for synthesizing speech in accordance with coded Japanese character data.

A text-to-speech device converts digital data of a text created by a device such as a so-called Japanese word processor (hereinafter referred to as a word processor) or another computer system having a character input function into a speech signal, and as if it were a text. This is a device that has the function of reading aloud, and is a very excellent function for people who are blind, to quickly understand a large number of documents, or to transmit information using audio equipment without using real voice. Therefore, further enhancement of functions is desired.

[Prior Art] FIG. 4 shows a configuration of a conventional text-to-speech apparatus.

In FIG. 1, after character string data is input to a character string input unit 2, it is transferred to an analysis unit dividing unit 4 and divided by delimiter data such as punctuation marks “.” And “,” in a sentence.
The character string data divided between the punctuation marks is transferred to the reading conversion unit 6. The reading conversion unit 6 identifies the word to be used by comparing each of the divided character string data with the Japanese notation stored in the dictionary unit 8, and outputs pronunciation data corresponding to the word. As a method of this word identification, for example, a technique such as the longest match method is applied. The utterance data output here is transferred to the speech synthesizer 10, and the speech is synthesized according to the utterance data. For example, a technology such as a VCV method is applied as the voice synthesis method.

[Means to be Solved by the Invention] However, in such a conventional text-to-speech apparatus, character string data is divided by delimiter data such as punctuation marks "." Although a method of performing speech synthesis processing for each character string data is adopted, for example, as shown in FIG. 5, a sentence continues over a plurality of lines, and the line feed portion includes punctuation marks “.” And “,”. Even if they are not separated by a delimiter, the line feed part of the character string is determined to be a text delimiter, so the text is cut at places that should not be separated, and speech synthesis is not performed with correct character string data There was a problem. In the case shown in Fig. 5, the text-to-speech device is divided into "From Japanese character string", "Clear voice" and "Synthesize", and each character string is identified by words. The line where the speech should be synthesized is broken on the way, so it is divided into "sentence reading device", "from Japanese character string", "clear sound", "voice" and "synthesize". , "Clear speech" is erroneously identified as a word, and erroneous speech synthesis is performed.

The present invention has been made in view of such a problem, and an object of the present invention is to provide a document reading control device and a document reading method that can always perform character synthesis by dividing character string data between punctuation marks. I do.

[Means for Solving the Problems] FIG. 1 is an explanatory view of the principle of the present invention.

First, the present invention provides an analysis unit dividing unit that divides, for each punctuation mark data, coded character string data in which the maximum number of characters is divided for each predetermined line and continues over a plurality of lines,
The present invention is directed to a text-to-speech control device that includes a conversion unit that converts each divided character string data into generated data and that outputs the generated data as voice.

For such a device, the final character string data that is not divided by punctuation data in a certain line is retained, and the character string data up to the first punctuation data in the character string data of the next line following the final character string data is stored. A correction unit is provided that combines the final character string data to form character string data between punctuation marks and converts the data into generated data.

Also, the present invention divides coded character string data in which the maximum number of characters is divided for each predetermined line and continues over a plurality of lines for each punctuation mark data, and converts each divided character string data into generated data. It is intended for a text-to-speech method that outputs as a sound by conversion.

In such a method, the last character string data which is not divided by the punctuation data in a certain line is held, and the last character string data up to the first punctuation data in the character string data of the next line following the last character string data is stored. The method is characterized in that the string data is combined to form character string data between punctuation marks and converted to generated data.

[Operation] In the text-to-speech apparatus of the present invention having such a configuration, since character string data is always divided into character string data between punctuation marks, correct speech synthesis can be performed as written.

[Embodiment] Fig. 2 is an embodiment configuration diagram showing one embodiment of the present invention.

In FIG. 2, reference numeral 12 denotes a character input unit for inputting coded character string data. Reference numeral 14 denotes an analysis unit division unit, which divides the input character string data by delimiter data such as punctuation marks “.” And “,”. Reference numeral 16 identifies a word to be used by comparing the divided character string data with Japanese notation data stored in the dictionary unit 18 and outputs pronunciation data corresponding to the word. As a method of this word identification, for example, a technique such as the longest match method is applied. Reference numeral 20 denotes a speech synthesis unit that performs speech synthesis according to the utterance data. For example, a technology such as a VCV method is applied as the voice synthesis method.

Reference numeral 22 denotes a correction unit, which includes a memory 24 and a combining unit 26.If character string data between punctuation marks extends over a plurality of unit blocks due to line feed or the like, the final character data string not divided by punctuation mark data is temporarily stored. When the analysis unit dividing unit 14 divides the character string data up to the first punctuation data appearing in the next character data string, the combining unit 26 combines the final character data string with the character string data. Thus, a character data string between punctuation marks is formed, and the character data string between punctuation marks is transferred to the reading conversion unit 16.

Next, the operation of the embodiment having such a configuration will be described with reference to FIG.

As shown in FIG. 5, the case where the maximum number of characters per line (per unit block) is determined, and speech synthesis is performed on a sentence that continues over a plurality of lines without being cut off by punctuation marks will be described. The analysis unit dividing unit 14 divides the character string data up to the “sentence reading device” according to the character string data, transfers the character string data to the reading conversion unit 16, and performs speech synthesis by the speech synthesis unit 20.

Next, the analysis unit division unit 14 divides the character string data from “Japanese character string” to transfer it to the conversion unit 16, and the speech synthesis unit 20 performs speech synthesis.

Next, the analysis unit dividing unit 14 divides the character string data up to “clear sound”. However, since this character string data is not broken by the punctuation of "." And ",",
The processing is transferred to 24, and the processing moves to the next line (unit block).

In the processing of the next unit block, the analysis unit division unit
14 divides the character string data up to “voice”. The “clear sound” and “voice” stored in the memory 24
Are combined to form character string data between punctuation marks of “clear voice”, and are transferred to the conversion unit 16, and the speech synthesis unit 20 performs speech synthesis.

And finally, the analysis unit division unit 14 “composes”
Is divided, and the speech synthesis unit 20 performs speech synthesis.

As described above, according to this embodiment, even when a sentence to be read is cut off due to a line break or the like, speech is always divided for each character string between punctuation marks, so that the sentence can be read correctly.

[Effects of the Invention] As described above, according to the present invention, when character string data between punctuation marks extends over a plurality of unit blocks due to line feed or the like, the correction unit corrects the last character data string that is not divided by the punctuation mark data. And further form the character string data between punctuation marks by combining the final character string data with the character string data up to the first punctuation data appearing in the next character string data following the final character string data, Since speech synthesis is performed, sentences can always be read correctly.

[Brief description of the drawings]

 1 is a diagram illustrating the principle of the present invention; FIG. 2 is a diagram illustrating the configuration of an embodiment of the present invention; FIG. 3 is a diagram illustrating the operation of the embodiment; FIG. FIG. 5 is an explanatory diagram for explaining the operation of FIG. In the figure, 12: character string input section 14: analysis unit division section 16: reading conversion section 18: dictionary section 20: speech synthesis section 22: correction section 24: memory 26: connection section

 ──────────────────────────────────────────────────続 き Continuation of the front page (72) Inventor Mitsuko Kaseta 1015 Ueodanaka, Nakahara-ku, Kawasaki-shi, Kanagawa Fujitsu Limited (56) References JP-A-63-106040 (JP, A) JP-A-60-246436 (JP, A)

Claims (2)

(57) [Claims] An analysis unit dividing unit that divides coded character string data for each punctuation mark data in which a maximum number of characters is divided for each predetermined line and continues over a plurality of lines; A document reading control device including a conversion unit for converting column data into generated data for outputting the generated data as voice, wherein a final character string data which is not divided by the punctuation data in a certain line is held, A correction unit that combines the character string data up to the first punctuation data appearing in the character string data of the next line following the data with the final character string data to form character string data between punctuation marks and converts the data into the generated data. A text-to-speech control device, comprising: 2. A method according to claim 1, wherein the coded character string data in which the maximum number of characters is divided for each predetermined line and which continues over a plurality of lines is divided for each punctuation mark data, and each divided character string data is generated as generated data. In the text-to-speech method of outputting as a voice by converting to the following, the final character string data not divided by the punctuation data in a certain line is held, and the first character string data in the next line following the final character string data is stored. A method of reading a document, comprising combining character string data up to appearing punctuation data with the final character string data to form character string data between punctuation marks and converting the data into the generated data.