JPH077335B2 - Conversational text-to-speech device - Google Patents

Description

DETAILED DESCRIPTION [Overview] For example, the invention is for causing a machine to read the dialogue of a script in which a plurality of persons appear. For this purpose, a character discrimination unit for discriminating the characters written in the script, a character information table in which the vocal characteristics of the characters are registered, and a plurality of phoneme files are provided. If it is written in the script that the character A is read aloud as "XXX", the vocal characteristics of the character A are extracted from the character information table and the phoneme file is selected according to this feature. Then, the phoneme parameters corresponding to "XXXXX" are extracted from the selected phoneme file, and the sentence "XXX" is spoken according to the other characteristics of the character A (volume, range, vocalization speed, accent, etc.). Output.

[Industrial application field]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a conversational text-to-speech device that automatically reads a dialogue in a script in which a plurality of persons appear. The field of use of the conversational text-to-speech device of the present invention is to read a novel, a field such as a theater or a movie, in which each actor reads a scenario while reading a sentence printed in Japanese with a character's name. Due to this, there are installations in facilities such as libraries for the blind and those with low vision.

[Conventional technology]

FIG. 4 is a diagram showing the configuration of a conventional Japanese reading-aloud device. In the figure, 1 is a Japanese sentence analysis unit, 2 is a Japanese dictionary, 3 is a voice parameter setting unit, 4 is a voice synthesis unit, 7 is a phoneme file, 8 is a speaker, and 9 is a manual operation panel. .

The Japanese sentence analysis unit 1 divides the Japanese sentence into word strings by referring to the Japanese dictionary 2, divides the word strings into voicing units of an appropriate size, and reads the voicing units in order. It is sent to the parameter setting unit 3. In the Japanese dictionary 2, words, word readings, accent information, etc. are registered. Although only one phoneme file 7 is shown, there are actually a plurality of phoneme files such as a male phoneme file and a female phoneme file. The manual operation panel 9 is used to set an instruction as to which phoneme file 7 to use, volume, tone color, tone range, utterance speed, and the like. Voice parameter setting section 7
Retrieves the phoneme parameter corresponding to the reading of each voice unit from the phoneme file 7 designated by the manual operation panel 9,
The tone range, tone color, utterance speed, volume, etc. are set in accordance with the control device from the manual operation panel 9, and these are sent to the voice synthesizer 4. The voice synthesis unit 4 performs voice synthesis based on the data from the voice parameter setting unit 3. The electric signal output from the voice synthesizer 4 is converted into voice by the speaker 8.

[Problems to be solved]

In the conventional Japanese reading device, the Japanese sentence is read out based on the setting information of the manual operation panel, or the Japanese sentence is read out with a certain tone color, range and utterance speed. However, recently, the application fields for reading out Japanese sentences have been expanding, and the field that requires text reading by a plurality of persons cannot be dealt with by the conventional technology.

The present invention was created in view of this point, and the conversational text-to-speech device capable of pre-registering the utterance characteristics of the characters as data in advance and distinguishing the text-to-speech by a plurality of persons. Is intended to provide.

[Means for solving problems]

FIG. 1 is a principle diagram of the present invention. In the figure, 1 is a Japanese sentence analysis unit, 2 is a Japanese dictionary, 3 is a voice parameter setting unit, 4 is a voice synthesis unit, 5 is a character identification unit, 6 is a character information table, 7 is a phoneme file, Reference numerals 8 denote speakers, respectively.

Generally, in the script, a plurality of characters represented as “A” and “B” are displayed. The Japanese dictionary analysis unit 1
The Japanese sentence is analyzed while referring to the Japanese dictionary 2 and divided into word strings. When the character name is detected, it is passed to the character determination unit 5. For the reading text in the brackets, the reading sequence of words is divided into utterance units, and the readings in utterance units are sequentially passed to the voice parameter setting unit 3. At this time, word accent information is also passed. As shown in FIG. 2, in the character information table 6, character names such as A, B, ... And voice feature information that can specify the utterance feature of the character are registered in advance. . The utterance feature information means age, sex, place of origin in a role, phoneme file number, and the like. The contents of the character information table 6 can be rewritten. When the character determination unit 5 receives a character name (for example, A) from the Japanese sentence analysis unit 1, the character determination unit 5 reads the content of the character information table 6 and the character name called A in the character information table 6. Check if exists. When the character exists, the character determination unit 5 notifies the character information table 6 that the character is A. Then, the utterance characteristic information of the character A is read from the character information table 6. Although only one phoneme file 7 is shown in FIG.
Is a phoneme file for men and a plurality of phoneme files for women. As shown in FIG. 3, the phoneme file 7 stores phonemes and parameters corresponding thereto. The phoneme file 7 is selected according to the phoneme file number read from the character information table 6. The age information and the birthplace information in the read utterance feature information are passed to the voice parameter setting unit 3. The voice parameter setting unit 3 takes out the parameter corresponding to the reading passed from the Japanese sentence analysis unit 1 from the selected phoneme file 7, and according to the received age information and birthplace information, the volume, range, vocalization speed, etc. And correct the accent. Then, the voice parameter setting unit 3 uses the phoneme parameter,
Volume information, range information, utterance speed information, accent information, etc. are passed to the voice synthesizer 4. The voice synthesizer 4 synthesizes voices based on these data and outputs a voice synthesis signal. The voice synthesizing unit 4 is, for example, a Pcall type.
The electrical voice synthesis signal output from the voice synthesis unit 4 is
It is converted into voice by the speaker 8. Note that the utterance order is automatically instructed by analyzing the characters in the input Japanese sentence "XXX".

When changing the overall sentence structure, it is necessary to have editing functions such as moving, inserting, changing, and replacing the Japanese sentences that are input to the reading device. A connection form of the reading device is considered.

〔The invention's effect〕

Although a reading device having a function of reading a conversational sentence by a plurality of people has not existed in the past, a plurality of people can be provided by providing a character determination unit, a character information table, and a plurality of phoneme files as in the present invention. It becomes possible to read conversational sentences in, for example, automatic reading of scenarios and automatic reading according to the characters of a novel.

[Brief description of drawings]

FIG. 1 is a principle diagram of the present invention, FIG. 2 is a diagram showing a configuration example of a character information table, FIG. 3 is a diagram showing a configuration example of a phoneme file, and FIG. 4 is a diagram showing a configuration of a conventional example. is there. 1 ... Japanese sentence analysis unit, 2 ... Japanese dictionary, 3 ... Voice parameter setting unit, 4 ... Voice synthesis unit, 5 ... Character discrimination unit, 7 ... Character information table, 7 ... Phoneme file, 8 ... Speaker, 9 ... Manual operation panel.

Claims (1)

[Claims] 1. A speech synthesis unit (4), a plurality of phoneme files (7), a Japanese dictionary (2), a Japanese sentence analysis unit (1), and a Japanese sentence to be read aloud. Character discriminating unit (5) for discriminating a character who is a voiced speaker
And a character information table (6) that registers vocal characteristic information that affects the vocal volume, tone color, range, vocal speed, accent condition such as gender, age, birthplace of the character, and character discrimination The volume, the range, the accent, etc. are determined based on the utterance characteristic information on the character information table (6) corresponding to the persons identified by the section (5), and the phoneme file ( 7) is selected, the phoneme parameters corresponding to the reading sent from the Japanese analysis unit (1) are extracted from the selected phoneme file (7), and the phoneme parameters, the volume, the range, and the accent are extracted from the speech synthesis unit (4). ), And a voice parameter setting unit (3) for sending to the conversation type sentence reading device.