JPH03236099A - Document read-aloud device - Google Patents

Abstract

PURPOSE:To accurately display a read-aloud position by storing analytic results of plural phrases in a storage means and outputting the analytic results to a synthesizing means corresponding to a command specifying the read-aloud position in a document, and outputting a voice. CONSTITUTION:A task control part 21 reads a Japanese document out of a memory 25 through an access part 23 and outputting it to the document read- aloud device 33 through a transmission/reception part 29, and a reception part 41 stores individual texts in a buffer 43 together with added text numbers. The document is analyzed by a word identification part 47 and an analysis part 49 and stored as phoneme and rhythm data in a buffer 51, and they are outputted from a speaker 35 through generation parts 55 and 57. An input text number and position information are inputted from the generation part 55 to a control part 21 through the transmission/reception part 29 and the control part 21 sends the text number and position to a display device 31 to invert and displays the position where a read-aloud is currently made. The control part 21 outputs control commands corresponding to respective keys to the device 33 and the buffer 51 sends the phrase at the position specified with a control command to a voice synthesizing mechanism, which synthesizes a voice.

Description

[Detailed Description of the Invention] [Summary] Regarding a text-to-speech device that reads Japanese sentences containing kanji and kana in a natural voice, this device realizes the functions of repeating, backtracking, and forwarding specified phrases, as well as accurately displaying the reading position. With the aim of achieving this, we have developed an analysis means that performs linguistic analysis on input sentences containing kanji and kana, and outputs the analysis results of phonological and prosodic information in phrase units, and a synthesis means that synthesizes and outputs speech based on the analysis results. In the text reading device, storage means sequentially stores analysis results of a plurality of phrases and outputs analysis results of corresponding phrases to a synthesis means in response to a control command specifying a reading position in an input sentence; and display means for displaying the position in the sentence corresponding to the phrase output from the storage means to the synthesis means.

[Industrial application field]

The present invention relates to a text reading device that reads out Japanese sentences (text) containing Kanji and Kana in a natural voice.

A text reading device can be used as an information transmission means when visual information transmission is not possible. For example, reading aloud from a database using a telephone line, a terminal for the blind, proofreading text input into a word processor, voice guidance, etc. can be considered.

[Conventional technology]

In order to read aloud a sentence containing kanji and kana, the words in the sentence are correctly identified (hereinafter referred to as ``Identify J''), the correct phonetic (reading) and accent are automatically assigned to the word, and the sentence is given a natural accent, pause, etc. Language analysis technology is needed to automatically add intonation, etc.

Therefore, the structure of a text reading device that reads out sentences containing kanji and kana input from an input device (keyboard, etc.) is based on the language processing section that performs this language analysis and the phonological and prosody information obtained here. It can be broadly divided into a speech synthesis section that synthesizes corresponding speech.

In other words, the language processing unit breaks down the input kanji-kana mixed sentence into predetermined units (words, word strings), refers to the word dictionary, and performs word identification processing that adds pronunciation, accent information, and grammatical information to each of them. After doing so, try compound words, clauses,
Processing is performed to reconstruct the relationships between clauses (accent phrases, exhalation paragraphs) into higher-level linguistic units in the order of the phrases, correct phonological information as compound words or sentences, determine prosodic information such as accent and intonation, and The configuration is such that the information is passed to the speech synthesis section.

By the way, it takes a relatively long time to analyze Japanese sentences because it searches a large word dictionary (Japanese sentences 20
(approx. 300 milliseconds for text), the audio output from the text reading device was sometimes interrupted.

Therefore, in conventional text-to-speech devices, buffers are installed before and after the analysis processing (word identification processing and prosody analysis processing) to perform the analysis processing and speech synthesis processing in parallel, so that the audio output will not be interrupted even when the analysis takes time. I was doing it.

[Problem to be solved by the invention]

The above-mentioned text reading device simply reads the text, but does not have a function that allows the reading position of the text to be moved from the outside, such as repeating a specified phrase, going back, or moving forward.

In addition, text-to-speech devices display the input Japanese text on the display device and highlight the position currently being read, but when the capacity of the buffer provided to prevent audio interruptions is large, this There may be a time lag between display output and audio output.

SUMMARY OF THE INVENTION An object of the present invention is to provide a text reading device that can realize the functions of repeating, backtracking, and forwarding specified phrases, and can accurately display the reading position.

[Means to solve the problem]

FIG. 1 is a block diagram of the principle of the present invention.

In the figure, the text reading device includes an analysis means 11 and a synthesis means 13.

The analysis means 11 performs linguistic analysis of the input sentence containing kanji and kana, and outputs the analysis results of reading and other phonological information, as well as inter-word pauses, accent combinations and other prosodic information, in units of phrases.

The synthesizing means 13 synthesizes and outputs speech based on the analysis result.

The present invention includes storage means 15 and display means 17.

The storage means 15 sequentially stores the analysis results of a plurality of phrases, and in response to a control command that specifies the reading position in the input sentence, the analysis results of the corresponding phrases are stored in the synthesis means 1.
Output to 3.

The display means 17 displays the position in the sentence corresponding to the phrase output from the storage means 15 to the synthesis means 13.

[For production]

In the present invention, the analysis results of a plurality of phrases are sequentially stored in the storage means 15, and the analysis results of the corresponding phrases are stored in the synthesis means 1 in response to a control command that specifies the reading position in the sentence.
3 to perform audio output. That is, by retaining the analysis results of a plurality of phrases before and after the phrase currently being read out, it is possible to repeat the phrase being read out, and to specify backward or forward reading positions in units of phrases.

Furthermore, since the position in the sentence corresponding to the phrase output from the storage means 15 to the synthesis means 13 is displayed, even when the reading position is moved, there is no difference between the audio output and the display output.

[Example] Hereinafter, the operation of the example will be described in detail based on the drawings.

FIG. 2 is a block diagram showing the configuration of a system using the text reading device of the present invention.

In the figure, a memory 25 is connected to the task management section 21 via a file access section 23, and a keyboard 27, a command transmission/reception section 29, and a display device 31 are also connected. A text reading device 33 is connected to the command transmitting/receiving section 29 . Sentence reading device 3
A speaker 35 is connected to 3.

FIG. 3 is a block diagram showing the configuration of an embodiment of the text reading device.

In the figure, a Japanese data storage buffer 43 is connected to the receiving section 41. A word identification section 47 to which a word dictionary 45 is connected is connected to the Japanese data storage buffer 43 . A phoneme/prosody analysis section 49 is connected to the word identification section 47 . A phoneme/prosody data storage buffer 51 is connected to the phoneme/prosody analysis section 49 . A synthesis parameter creation section 55 to which the audio file 53 is connected is connected to the phoneme/prosody data storage buffer 51 . A synthesized speech generator 57 is connected to the synthesis parameter generator. A speaker 35 is connected to the synthesized speech creation section 57 . A command transmitting/receiving section 29 is connected to the receiving section 41, the Japanese data storage buffer 43, and the phoneme/prosody data storage buffer 51.

The word identification unit 47 identifies words or word strings while referring to the word dictionary 45, and extracts the pronunciation, accent information, and grammatical information of each word. Note that the word dictionary 45 contains notation for word identification, grammatical attributes representing connection relationships between words, and other word attributes, as well as phonology (reading), accent, and accent combination used for next-stage phonology and prosodic processing. Attributes are registered respectively.

The phoneme/prosody analysis unit 49 performs a process of reconstructing into higher-level linguistic units in the order of compound words, clauses, and relationships between clauses. That is, based on the pronunciation of the word obtained by the word identification section 47,
Speech synthesis is performed according to phonological rules that obtain rendaku or phoneticization information generated by word concatenation, and prosodic rules that determine accent phrases, exhalation paragraphs, and other tonal control information based on grammatical information of word and clause boundaries. Create various control information (phonological information, prosody information) necessary for

The synthesis parameter creation unit 55 refers to the audio file 53 in which basic Japanese syllables are stored in parameter format, reads out necessary parameters based on the phonetic symbols, synthesizes a continuous parameter string, and synthesizes this. Audio creation section 5
Output to 7.

The synthesized speech creation unit 57 synthesizes a speech waveform from the synthesis parameters and causes the speaker 35 to output continuous speech.

The Japanese data storage buffer 43 and the phoneme/prosody data storage buffer 51 are for processing language analysis and speech synthesis in parallel, and when language analysis takes time, the audio output from the speaker 35 is interrupted. It has a function to prevent this.

Here, the correspondence between FIG. 1 and FIGS. 2 and 3 will be shown.

The analysis means 11 corresponds to a word dictionary 45, a word identification section 47, and a phoneme/prosody analysis section 49.

The synthesis means 13 corresponds to an audio file 53, a synthesis parameter generation section 55, a synthesized speech generation section 57, and a speaker 35.

The storage means 15 corresponds to a Japanese data storage buffer 43 and a phoneme/prosody data storage buffer 51.

The display means 17 corresponds to the display device [31].

The receiving unit 41 is a standard component of a text reading device.

The operation of the text reading device will be described below with reference to FIGS. 2 and 3.

The task management unit 21 reads a Japanese text from the memory 25 via the file access unit 23 and sends the Japanese text to the text reading device 33 via the command transmission/reception unit 29.
Output to.

The Japanese text input to the text reading device 33 is received by the receiving unit 41, a text number is assigned to each text in order to identify each text that makes up the text, and the text is stored in the Japanese data storage buffer 43. Stored.

The Japanese sentences stored in the Japanese data storage buffer 43 are processed by the word identification section 47 and the phoneme/prosody analysis section 49.
It is analyzed and stored in the phoneme/prosody data storage buffer 51 as phoneme/prosody data for each phrase. For example, the input text "Today's weather is sunny" is "
The phrases are broken down and stored into three phrases: ``Kyouno'', ``Tenkiwa'', and ``Haredes''.

The phoneme/prosodic data storage buffer 51 is configured to store phoneme/prosodic data of several phrases or dozens of phrases, and the stored phoneme/prosodic data is collected by the synthesis parameter creation unit 55 and the synthetic speech creation unit 57 on a phrase-by-phrase basis. The signal is output from the speaker 35 via. In addition, the phoneme and prosody data stored in the phoneme and prosody data storage buffer 51 are as follows:
For each phrase, input text number and position (e.g. "
``Kiyouno'' holds the starting position and length of a word, such as the first to fourth characters of the first text, and this information of the phrase output to the synthesis parameter creation unit 55 is used by the command It is output to the task management section 21 via the transmission/reception section 29.

The task management unit 21 outputs the text number and reading position currently being read out to the display device 31 based on the information supplied from the command transmitting/receiving unit 29, and reverses the position currently being read out in the Japanese text being displayed. Display.

The task management unit 21 also scans the keyboard 27, and when a key on the keyboard 27, such as repeating a phrase being read out, backtracking the reading position, or forwarding the reading position, is pressed, the task management unit 21 controls the control corresponding to each key. The command is output to the text reading device 33.

In the text reading device 33, the control command is received by the Japanese data storage buffer 43 and the phoneme/prosody data storage buffer 51 via the command transmission/reception unit 29.

When the phoneme/prosody data storage buffer 51 receives a control command, the phoneme/prosody data storage buffer 51 is configured to perform a speech synthesis mechanism (synthesis parameter creation section 5).
5. The voice synthesis process in the synthesized voice creation unit 57) is temporarily stopped, and the phrase at the position specified by the control command is sent to the voice synthesis mechanism to perform voice synthesis.

For example, as shown in FIG. 4, the phoneme/prosody data storage buffer 51 stores six phrases numbered sequentially from the beginning of the text and outputs the third phrase. In response to the control command "Read out the next phrase," the fifth phrase is sent to the speech synthesis mechanism and output from the speaker 35. Similarly, it is possible to set the reading position according to the control command, such as repeating the phrase that is being output, or going backwards such as "reading the previous phrase two times."

In addition, when the reading position is advanced beyond the range stored in the phonetic/prosody data storage buffer 51 (for example, to read out a phrase four or more phrases after the phrase currently being output), according to the control command, After the text stored in the Japanese data storage buffer 43 is output to the word identification section 47 and analyzed, it is stored in the phonological/prosodic data storage buffer 51 and then sent to the synthesis parameter creation section 5.
5 is output.

〔Effect of the invention〕

As described above, according to the present invention, by making full use of the buffering effect of the text reading device, it is possible to not only simply read Japanese text, but also to avoid repeating phrases being read, backtracking to the reading position, and postponing the reading position. Since it can be controlled, it is possible to read the same passage repeatedly, go back and listen to a passage that you have missed, and skip over unnecessary passages.

In addition, since the display output is performed using the position in the Japanese text of the phrase output to the speech synthesis mechanism, even when the buffer capacity is large or the reading position moves, the display indicating the position being read is displayed. Output and audio output can be synchronized.

[Brief explanation of drawings]

FIG. 1 is a block diagram of the principle of the present invention, FIG. 2 is a block diagram showing the configuration of the system, FIG. 3 is a block diagram showing the configuration of an embodiment, and FIG. 4 is a diagram explaining the operation of the buffer. In the figure, 11 is an analysis means, 13 is a synthesis means, 15 is a storage means, 17 is a display means, 21 is a task management section, 23 is a file access section, 25 is a memory, 27 is a keyboard, 29 is a command transmission/reception section, 31 is a display device, 33 is a text reading device, 35 is a speaker, 41 is a receiving unit, 43 is a Japanese data storage buffer, 45 is a word dictionary, 47 is a word identification unit, 49 is a phonological/prosody analysis unit, 51 is a phonological/prosody analysis unit 53 is a prosody data storage buffer, 53 is an audio file, and 55 is a synthesis parameter creation unit. 57 is a synthesized speech creation section. Figure 1 is a block diagram showing the system configuration. Figure 2 is a block diagram showing the configuration of the 1175-embodiment. Figure 1 is a diagram explaining the operation of the buffer.

Claims (1)

[Claims] (1) Perform linguistic analysis of input sentences containing kanji and kana,
A text reading device comprising an analysis means (11) for outputting analysis results of phonetic information and prosody information in units of phrases, and a synthesis means (13) for outputting synthesized speech based on the analysis results, a storage means (15) for sequentially storing the analysis results and outputting the analysis results of the corresponding phrase to the synthesis means (13) in response to a control command specifying a reading position in an input sentence; 15) to the synthesis means (13); display means (17) for displaying a position in the sentence corresponding to the phrase output from the synthesis means (13).