JP2680643B2 - Character display method of rule synthesizer - Google Patents

Description

The present invention relates to a character display method of a rule synthesizing device which is connected to a computer system or a word processor to generate a voice from a character string input.

(B) Conventional Technology Conventionally, for example, as shown in Japanese Patent Publication No. 62-23889,
Display data and audio output data are generated by a computer,
It has been made easy to confirm the display data by performing display by the display device, voice synthesis by the voice output device, and changing the display state corresponding to the output voice.

However, in such a conventional voice output device, when outputting a voice while displaying a character string of a long sentence such as when proofreading a created document, it is necessary to change the display state in units of punctuation marks. I was not able to know the character of the utterance accurately.

Also, if a sentence is divided into words and clauses for display and voice output, the voice output device cannot determine the phrase component over the whole sentence, resulting in a mechanical synthesized voice lacking naturalness. There was a drawback.

(C) Problems to be Solved by the Invention The present invention has been made in view of the above points, and changes the display state in units of characters, words, or phrases being displayed in correspondence with the voice being uttered. The present invention provides a character display method of a rule synthesizing device that makes it easy to understand the vocalization position.

(D) Means for Solving the Problem A character display method of a rule synthesizing device according to the present invention is a rule synthesizing device for generating a voice by inputting a character string, and the method for displaying characters displayed on a display device from an input character string is A character position table indicating the position and a parameter position table indicating the position in the parameter string of the voice uttered from the voice utterer based on the input character string are created respectively, and the character position table and the parameter position table are displayed on the display device. The character position information corresponding to the position in the parameter string of the voice being uttered is output by associating the position of the displayed character with the position in the parameter string of the voice uttered by the voice utterer, and the character position information is output. The display character position indicating the position of the character displayed on the display synchronized with the position in the parameter sequence of the voice being uttered is visually displayed in real time.

(E) Operation According to the present invention, the character position information corresponding to the parameter being uttered is output, and the display state of the character of the character position information is changed on a character-by-character or word-by-character basis. Yes, the correspondence between synthetic voice and display can be clearly confirmed.

(F) Embodiment FIG. 1 shows a block diagram of an embodiment of the device configuration for realizing the character display method of the rule synthesizing device of the present invention. FIG. 2 shows an example of processed text, and FIG. 3 shows an example of data in each table. The method of the present invention will be described in detail below based on these figures.

When a Japanese or Kana / Kana mixed sentence such as “I am a cat.” [FIG. 2 (a)] is input from a computer or word processor, the morphological analysis unit (1) divides the sentence into words. [Fig. 2 (b)]. At this time, the number of characters from the beginning of the sentence to the first character of each word is counted and stored in the character position table (2) as shown in FIG. 3 (a). The divided words are sent to the reading unit (3) to give readings to each word [Fig. 2 (c)] and correspond to each character position in the character position table (2). The reading position is calculated and stored in the reading position table (4) as shown in FIG. 3 (b). The word provided with the reading is sent to the accenting / phrase determining unit (5). Accent
In the phrase determination unit (5), each word is accented, accents are combined, that is, words are combined to form one accent phrase, and the position and depth of the phrase, that is, the breath is determined [Fig. 2 (d. )].

Next, the phoneme connecting unit (6) connects the phonemes for each reading (pronunciation) and generates a parameter string for voice synthesis [Fig. 2 (e)] together with the reading position table ( 4)
The parameter position corresponding to each reading position of is calculated,
It is stored in the parameter position table (7) as shown in FIG. 3 (c). Further, the prosody imparting unit (8) calculates pitch data serving as prosody information from the accent and phrase data, adds the pitch data to the parameter string, and the voice synthesis unit (9) generates a voice waveform, and a speaker (10). Output from.

Below, various tables (2), (4), (7) of the above-mentioned configuration
The method of creating will be described in detail below.

According to FIG. 5, character position table (2), reading position table (4)
Explain the creation process of.

When a sentence is input from a word processor or the like, it is stored in the sentence buffer (11). The sentence pointer (12) points to the first character of the sentence, and the morphological analysis unit (1)
The number of characters of the word stored in the word dictionary (13) is read from the sentence position pointed by the sentence pointer (12) and compared with the word dictionary (13). If they match, the word number of the corresponding word dictionary (13) is sent to the reading unit (3) and the match signal (14) is output. Initially, the position table pointer (15) points one position before the start position of the character position table (2) and reading position table (4), and is incremented by the coincidence signal (14) to point to the start position of each position table. It will be.

Further, the AND gate is opened by the coincidence signal (14), the value of the sentence pointer (12), that is, the sentence position is sent to the character position table (2), and the position table pointer (15) of the character position table (2) points to it. Stored in place.

Upon receiving the word number from the morpheme analysis unit (1), the reading unit (3) reads out the reading unit from the word dictionary (13) corresponding to this, and the accenting / phrase determination unit (5). , And the reading position table preparation section (16). This reading position table creation unit (16) counts the number of reading characters, accumulates the number of reading characters from the beginning of the sentence, and saves it in the position indicated by the position table pointer (15) of the reading position table (4). .

Further, the morphological analysis unit (1) advances the sentence pointer (12) by the number of characters of the matched word, moves to the morphological analysis of the next word, and continues this.

Next, the process of creating the parameter position table (7) will be described with reference to FIG.

The accent assigning / phrase deciding unit (5) decides the accent position and breathing position of each word, and inputs a phonetic string into the reading buffer (61).

The segment connecting unit (6) first sets the read pointer (62) and the parameter pointer (92) so as to point to the heads of the read buffer (61) and the parameter buffer (91), respectively. Next, the parameter string corresponding to the reading indicated by the reading pointer (62) of the reading buffer (67) is stored in the segment memory (6
It is read from 3) and stored in the position pointed to by the parameter pointer (92) of the parameter buffer (92).

Further, the above-mentioned segment connecting portion (6) has a reading pointer (62).
Is incremented, the parameter pointer (92) is advanced by the stored parameter string, and the segment parameters corresponding to the next reading are connected. A broken line portion in FIG. 6 is a parameter position table creation unit (63), and the comparison circuit (631) therein has a reading pointer (62) and a position table pointer (15) of the reading position table (4). Compare the reading position to point,
If they match, the AND gate is opened and the value of the parameter pointer (92) is stored in the location indicated by the position table pointer (15) of the parameter position table (7). After that, the position table pointer (15) is incremented and this is continued.

Next, a detailed block diagram of the voice synthesizer (9) is shown in FIG. 4 to explain the operation.

The parameters (LSP parameters, amplifier data, pitch data) output from the prosody imparting unit (8) are held in the parameter buffer (91). Parameter pointer (92)
Indicates the beginning of the parameter buffer (91),
Frame cycle from the tamming control circuit (93), eg 12.8
When the timing signal comes every msec, the parameter pointer (92) is incremented to point to the next parameter. The parameter pointed to by this parameter pointer (92) is sent to the synthesis filter (94), which produces a speech waveform and outputs it to the speaker (10).
Output. The comparison circuit (96) compares the contents of the parameter pointer (92) with the parameter position indicated by the counter (97) of the parameter position table (7), and outputs a coincidence signal when they match. The AND gate is opened by the coincidence signal, and the character position pointed by the counter (97) of the character position table (2) is output. Counters (97)
Indicates the same location in the parameter position table (7) and the character position table (2), and the parameter position and the character position correspond to each other. After that, the coincidence signal increments the counter (97) to indicate the next parameter position and character position. The character position information thus output is sent to the display device to change the display state of the corresponding display portion. For example,
The display color can be changed and blinking can be displayed.

In the above description, the character position is output for each word and the display state can be changed, but the character position may be output for each character or phrase.

(G) Effect of the Invention According to the character display method of the speech rule synthesizing device of the present invention, a display for displaying an input character string and a parameter in a parameter string corresponding to a display position of a character in the input character string are displayed. It has a correspondence table with the position, outputs the character position information corresponding to the parameter being uttered based on the correspondence table, and displays the display character position synchronized with the uttered voice in real time according to the character position information. Since it is displayed, it is possible to change the display state of the character position corresponding to the voice being uttered in units of characters, words, or phrases, and it is possible to improve the accuracy of correspondence between synthetic voice and display.

Therefore, according to the method of the present invention, it is possible to contribute to the realization of a text reviewing device that can proofread a created text very easily.

[Brief description of the drawings]

FIG. 1 is a block diagram of a device configuration for realizing a character display method of a speech rule synthesizing device of the present invention, FIG. 2 is a diagram showing processing for an input sentence, and FIG. 3 is a state of each table in FIG. FIG. 4, FIG. 4 is a detailed block diagram of the voice synthesizing unit of FIG. 1, and FIGS. 5 and 6 are diagrams showing states of respective tables of FIG. (2) …… Character position table, (4) …… Reading position table, (7) …… Parameter position table.

 ─────────────────────────────────────────────────── ─── Continued Front Page (72) Toru Kitamura Toru Kitamura 2-18 Keihan Hondori, Moriguchi City, Osaka Sanyo Electric Co., Ltd. (56) Reference JP-A-60-79427 (JP, A) 57-114434 (JP, U)

Claims (1)

(57) [Claims] 1. A rule synthesizing device for generating a voice by inputting a character string, comprising a character position table indicating the position of a character displayed on a display from the input character string, and a voice utterer based on the input character string. Create a parameter position table showing the position in the parameter sequence of the voice uttered from, respectively, the position of the character displayed on the display from the character position table and the parameter position table and the parameters of the voice uttered from the voice utterer The character position information corresponding to the position in the parameter string of the voice being uttered is output in association with the position in the string,
Character display of a rule synthesizing device characterized by displaying in real time a display character position indicating a position of a character displayed on a display synchronized with a position in a parameter string of a voice being uttered according to the character position information Method.