JPH01119822A - Sentence reader - Google Patents

Abstract

PURPOSE:To synthesize and express a natural voice by designating a range and adopting different reading with respect to a KANJI (Chinese character) of its part, in a sentence reader for outputting a KANJI and KANA (Japanese syllabary) mixed sentence in a voice. CONSTITUTION:A sentence by a KANJI and KANA mixed sentence inputted from an input/output display device 1 is sent to a sentence accumulating part 4 and stored in the sentence accumulating part 4 by an input/output control part 3 and a main control part 2. A language processing part 5 reads out the sentence from the sentence accumulating part 4, divides it into analytic units, executes a language analysis at every analytic word and replaces the language with the information of a phonetic symbol and a meter symbol. An acoustic processing part 6 synthesizes a sound signal from the phonetic symbol and the meter symbol, and outputs a voice from a loudspeaker 7. Also, with regard to the sentence concerned, as for the sentence whose reading or meter is incorrect, or the sentence being different from a user's intention, the original sentence is displayed on a display screen from the input/output display device 1, the range is designated by a cursor, and the phonetic symbol and the meter symbol which are intended are inputted in advance.

Description

[Detailed Description of the Invention] [Summary] The present invention relates to a device that performs language analysis on a sentence containing kanji and kana, converts it into phonetic symbols and prosodic symbols, and synthesizes and outputs speech based on this. The aim is to provide a device that can output natural speech for parts that would be unnatural when synthesized using the results of language analysis. means for specifying a range of arbitrary characters or character strings in It is constructed by providing a means for synthesizing and outputting.

[Industrial Field of Application] The present invention relates to a device that performs linguistic analysis on a sentence containing kanji and kana, converts it into phonetic symbols and prosodic symbols, and synthesizes and outputs speech based on this. This is a text-to-speech device that can output natural speech for parts that would be unnatural when synthesized using the analysis results, or intentionally output speech with a different prosody than normal according to specifications. Related.

[Prior Art] FIG. 3 is a block diagram showing the configuration of a conventional text reading device, in which 51 is an input device, 52 is a main control section, 53 is an input control section, 54 is a language processing section, and 55 is an audio device. Processing section, 5
6 represents a speaker.

In the figure, a sentence containing kanji and kana input from an input device 51 is analyzed by a language processing unit 54 and converted into a string of phonetic symbols and prosodic symbols (accents, pause symbols, etc.).

After converting these phonetic symbols and prosodic symbols into audio parameters, the audio processing unit 55 synthesizes audio and outputs it from the speaker 56.

[Problems to be Solved by the Invention] In a text reading device, it is desirable that the original characters are pronounced correctly and the meaning of the text can be accurately determined based on the output audio.

However, in reality, when there are homographs and allophones in a sentence containing kanji and kana, or when there are compound words made up of many words, the correct phonetic symbols and appropriate prosodic symbols are used for these words. It is not easy to grant.

If we were to make these highly practical, we would need to introduce semantic analysis or set up rules for individual words, so we would need to use a simple method like the one shown in Figure 3 above. It is impossible to achieve this with the configuration of the device.

Therefore, in the past, when there were homographs and compound words made up of many words, it was difficult to economically realize a text-to-speech device that could correctly pronounce these words. .

In view of these conventional problems, the present invention provides a text reading device that can correctly perform utterances in special cases such as those mentioned above, and that can be realized economically with a simple configuration. It is intended to.

[Means for Solving the Problems] According to the present invention, the above objects are achieved by the means described in the claims. That is, the present invention
A device that performs linguistic analysis on a sentence containing kanji and kana, converts it into phonetic symbols and prosodic symbols, and synthesizes and outputs speech based on this. A means for specifying a range of characters or character strings, and a means for synthesizing and outputting speech based on separately set symbol strings for the characters or character strings specified by the means, without using the results of the linguistic analysis. This is a text reading device equipped with

[Function] The text reading device according to the present invention can, for example, read a sentence such as ``I like Satsuki,'' which would be voice synthesized as ``I like Gogatsu'' using a normal language analysis method. , when you want to utter ``I love Satsuki'', specify the range for ``Satsuki'' and separately set the phonetic symbol and metrical symbol for voice synthesis with ``Satsukipa''.

When the text-to-speech device detects the part specified in the above range (May) in the sentence to be synthesized, it ignores the results of the language analysis and outputs the synthesized speech as "Pasatsuki". .

[Embodiment] FIG. 1 is a block diagram of an embodiment of the present invention, showing an example of the configuration of a text reading device, in which 1 is an input/output display device, 2 is a main control unit, and 3 is an input/output device. A control section, 4 is a text storage section, 5 is a language processing section, 6 is a sound processing section, and 7 is a speaker.

In the figure, a sentence containing kanji and kana input from the input/output display device 1 is sent to the input/output control unit 3 and the main control unit 2.
As a result, the text is sent to the text storage section 4 and stored therein.

The language processing unit 5 reads the text from the text storage unit 4, divides it into analysis units, performs language analysis on each analyzed word, and converts the language into information on phonetic symbols and prosodic symbols.

The sound processing section 6 synthesizes a sound signal from the phonetic symbols and the prosodic symbols, and outputs the signal from the speaker 7 as sound.

At this time, if the corresponding sentence is incorrect in pronunciation or prosody, or is different from the user's intention, the input/output display device 1 displays the original sentence on the display screen, and uses the cursor to select the appropriate range. is specified, and the phonetic symbols and prosodic symbols related to speech synthesis for the range are input. From now on, when performing speech synthesis on the text, language analysis will not be performed for the parts where the range has been specified.
Corresponding phonetic symbols and prosodic symbols are sent directly to the audio processing section, thereby performing speech synthesis.

FIG. 2 shows a flowchart of the above-mentioned control and operation, and (a) shows a method for specifying a different utterance for the synthesized speech resulting from language analysis. b) shows a speech synthesis method when there is an indication related to speech synthesis not based on the result of language analysis in the text.

[Effects of the Invention] As explained above, according to the present invention, a device that performs linguistic analysis on sentences containing kanji and kana, converts them into phonetic symbols and prosodic symbols, and synthesizes and outputs speech based on these symbols. , it is possible to output natural speech for parts that would be unnatural when synthesized using the results of normal language analysis, or to output speech with a prosody that is intentionally different from normal by specifying it. There is an advantage that the text reading device can be easily realized with a simple configuration.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of an embodiment of the present invention, FIG. 2 is a flowchart showing control and operation of an embodiment of the invention, and FIG. 3 is a block diagram showing the configuration of a conventional text reading device. 1... Input/output display device, 2... Main control section, 3... Input/output control section, 4... Text storage section, 5... ... Language processing section, 6 ... Sound processing section, 7 ... Speaker

Claims (1)

[Claims] In a device that performs language analysis on a sentence containing kanji and kana, converts it into phonetic symbols and prosodic symbols, and synthesizes and outputs speech based on this, the characters of the sentence containing kanji and kana to be synthesized into speech. A means for specifying a range of arbitrary characters or character strings in a string; and a means for specifying a range of arbitrary characters or character strings in a string; A text reading device characterized by comprising: means for synthesizing and outputting.