JPH04199420A - Document lead-aloud device - Google Patents

Abstract

PURPOSE:To obtain a highly natural voice output that can satisfactorily express the information contents of the document data by changing and controlling the conditions of the voice output of the document date based on the attributes of the emphasized characters, etc., included in the document date. CONSTITUTION:The document date read selectively out of a document data memory 1 are analyzed by a document analyzing part 6 in terms of a language. A voice synthesizing device 12 synthesizes and outputs the voices based on the voice date generated from the document data based on the analyzing result of the part 6. Then e controller 2 changes end controls the control data, e.g. at least one of the conversation speed, the sound volume, the sound quality, end the sound pitch of the voices that controls the voice synthesization output based on the attribute information like the emphasized expressions, the underlines, etc., added to the document data. Thus it is possible to output the highly natural document voices that have the rich expressions of contents where the atmosphere of the document data contents is reflected.

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Industrial Application Field) The present invention relates to a document reading device that can effectively output audible document data created on an information processing device.

(Prior Art) In recent years, with the development of information processing equipment, various documents have been frequently created in the form of text data using word processor functions. This type of document data is created as a string of character codes that can be handled by a predetermined information processing device, and is, for example, printed out via a printer or displayed on a predetermined display for monitoring. be done.

Recently, various speech synthesis techniques have been developed as methods for realizing excellent man-machine interfaces. This speech synthesis technology synthesizes and outputs document data given as a character code string, for example, in units of words, phrases, etc. based on predetermined rules.

Therefore, attempts have been made to use such voice synthesis technology to synthesize and output document data created on a predetermined information processing device as described above, thereby performing so-called reading of a document. Such a document reading device has various advantages, such as being able to provide document information in place of Braille for visually impaired people.

On the other hand, conventional document reading devices only linguistically analyze document data indicated by character code strings, generate voice data according to the analysis results, and perform voice synthesis. For this reason, even if an attribute such as highlighting is attached to document data, for example, there is a problem in that it is not possible to emphasize this and perform speech synthesis.

(Problem to be solved by the invention) In this way, in the past, speech was simply synthesized and output according to the document data indicated by the character code string, so if the character code string had an attribute indicating an emphasized expression, Even if information is attached, this is simply converted into speech in the same way as a normal character code string, so it tends to be synthesized and output speech, or what is called simple reading.

The present invention has been made in consideration of these circumstances, and its purpose is to change the synthesized and output speech according to attribute information indicating emphasized expressions attached to document data. An object of the present invention is to provide a document reading device capable of outputting a highly natural document voice with rich content expression.

[Structure of the Invention] (Means for Solving the Problems) A document reading device according to the present invention linguistically analyzes document data expressed in a format that can be processed on a predetermined information processing device, and reads the results of this analysis. control data for controlling the synthesized output of the voice, such as at least the speech rate, volume, quality, and pitch of the voice; One feature is that a means is provided for controlling changes in accordance with attribute information such as emphasized expressions and underlines added to the document data.

(Function) According to the present invention, control data that defines, for example, the speech rate, volume, sound quality, pitch, etc. of the voice, for controlling the synthesized output of the voice, can be added to the emphasized expression added to the document data. Since the changes are made according to the attribute information such as underline or underline, it is possible to synthesize and output the voice that has been changed or added according to the attribute information attached to the document data, allowing rich content expression that reflects the atmosphere of the document data content. This makes it possible to output highly natural document audio.

(Embodiment) Hereinafter, a document reading device according to an embodiment of the present invention will be described with reference to the drawings.

FIG. 1 is a schematic configuration diagram of the embodiment device, and FIG. 2 is a diagram showing the overall processing flow of the embodiment device.

In FIG. 1, a document data memory 1 stores a plurality of document data created in a format that can be handled and processed by a predetermined information processing device (computer). These document data are created by a document creation device such as a word processor, or are composed of information read by OCR or the like and converted into character codes.

The control device 2 starts a document reading process in response to an instruction from the input device 3, such as a keyboard, and reads data of the document instructed from the input device 3 from the document data memory 1 (step a). At this time, various types of control information indicating how to read out the document data are manually inputted from the human-powered device 3 (step b).

It is also possible to input such control information and specify the document using, for example, a command displayed on the display device 4 as a clue.

The control information input as described above is then stored in the setting buffer 5. In this state, the control device 2 supplies the document data selectively read out from the document data memory 1 to the document analysis section 6. Then, the document analysis unit 6 analyzes the document data morphologically, syntactically, and semantically according to the morphological information, pronunciation information, accent information, etc. of various words stored in advance in the word dictionary memory 7 ( Step C). Then, according to the analysis results, the document data is divided into words, for example, and information about each word is stored in the analysis result buffer 8. At this time, if attribute information such as emphasized characters or underlines is added to the document data, the document analysis section 6 stores this information in the analysis result memory 8 by attaching the information to the corresponding word.

When the analysis process for the document data is completed in this way, the control device 2 then activates the audio data generation section 9. Then, the voice data generation unit 9 generates voice data for voice synthesis from the document data according to the generation rules stored in the voice data generation rule memory 10, and stores this in the voice data memory 11 (step d). ). When generating this audio data, if attribute information such as emphasized characters and underlining described above is added to the word, the audio data generation unit 9 adds the emphasis to the word according to the conditions set in the setting buffer 5 described above. Add information on how to emphasize text parts for speech synthesis (steps e, f, g)
).

After the voice data is created as described above, the voice synthesizer 12 is activated, and the information on the voice synthesis setting conditions stored in the setting bar 5 is given to the voice synthesizer 12 (step h ). At this time, if time information regarding reading is set in the setting buffer 5, the reading speed setting unit 3 is activated, and a calculation is performed to determine at what speed the audio data should be output by voice synthesis. (steps i, j). This calculation is basically performed by calculating the reading speed from the amount of document data and the specified reading time.

When the reading speed is calculated in this way, the speech synthesis device 12 executes speech synthesis processing at the calculated reading speed, converts the audio data into speech, and outputs the speech output device (
Output from speaker) 14.

When synthesizing and outputting such audio data, the display data creation unit 15 creates display data based on the document data, adding information indicating, for example, the breaks and accent positions of the output audio, and stores this in the display data memory. 16 (step k).

Then, the display data stored in the display data memory 16 is displayed on the display device 4 in conjunction with the synthesized output of the audio data (step Ω).

Next, we will discuss the specific processing operations of this device configured as described above, as shown in Figure 3, using an example in which document data such as ``I read a book today'' is read out by voice synthesis. Explain. In this example, the word [Today is underlined and
Information is given that the word should be emphasized.

In this case, for example, according to an instruction command displayed on the display device 4, conditions for reading out the document data are inputted from the input device 3.

These reading conditions include items such as reading speed, sound quality, height (1) strength, time to finish reading (1) presence or absence of special readings for emphasized characters, changes in special readings, presence or absence of reading, length of breaks, etc. For example, the settings are registered in the setting buffer 5 in the format shown in FIG.

On the other hand, the designated document data is given to the document analysis section 6. The document analysis unit 6 refers to a word dictionary that describes the part of speech, pronunciation, accent type, meaning, etc. for the heading, as shown in FIG. 5, for example, and analyzes the document data morphologically, syntactically, and semantically. To analyze.

Then, the document data is divided into words, and the analysis results are stored in the analysis result buffer 8 as a data structure as shown in FIG.

At this time, the document analysis unit 6 detects emphasized characters such as enlarged characters, reversed characters, and underlined characters in the document data, and indicates to the analysis result buffer 8 that the characters are to be emphasized. Describe attributes.

The attributes of this emphasized character are expressed by inserting a predetermined control code or the character immediately before the target character.

When the analysis of the document data is completed, the document data generation section 9 is activated. The document data generation unit 9 generates audio data from the document data by referring to the audio data generation rules as shown in FIG. 7, for example, and this audio data is generated in the format shown in FIG. It will be done. In other words, the voice data generation rule shown in FIG. 7 is: [If the accent type is not 0 for a three-stage verb, and the conjugation t or unnatural form is present, the accent type is set to 0 and voice data is generated.] It shows the rule.

The audio data is created in a format of [speed; sound quality; height; character string to be read] as shown in FIG. 8 while referring to such rules. In the case of the document data ``I read a book today. In Figure 8, [゛] shows the position of the accent, and [・ko shows the position of the rest.

However, if the data stored in the analysis result buffer 8 has an attribute that indicates emphasized characters, for example, control data for instructing changes in speed, tone quality, or intensity at the word break may be included. inserted. Such control data changes the conditions for speech synthesis at that point. Note that if there is no change in the control data, speech synthesis will proceed as is according to the initially set control data.

Thereafter, the speech synthesizer 12 described above is activated.

At this time, if the above-mentioned reading end time is set, the reading speed setting section 13 first calculates the total number of beats in the sentence from the audio data. Then, the unit time for speech synthesis required for each beat is calculated by taking into account the number of breaks in the document, and the speech rate for speech synthesis is calculated.

The beat here is the smallest unit of pronunciation, for example [School (
This shows that the word ``cakko'' has 4 beats, and the word ``shakai'' has 3 beats. Therefore, in the case of the above-mentioned molecule, there are 14 beats in total.

However, for a document with 1500 beats, the reading time is 5
If the duration of the 100 breaks in the above document is set as 10 milliseconds, then the reading speed of the sentence is (60X5- (1oOXO
,ol))÷1500-0.199,
It is required that speech synthesis be performed in a time of 0.199 seconds per beat.

According to the speech rate calculated in this way, the speech synthesizer 12 synthesizes the speech data according to a predetermined synthesis rule.
Various control data (sound quality, etc.) mentioned above.

The voice is synthesized according to the height, strength, etc.).

As a result, if there are no emphasized characters, speech synthesis is performed as is according to the initially set control data, and if emphasized characters are specified, the speech speed, sound quality, height, strength, etc. The changes are made and speech synthesis is performed.

When the document data is read aloud by the voice synthesis output in this manner, the document data is displayed in audio symbol form as shown in FIG. This phonetic symbolization can be done by adding accent marks to the pronunciation, shifting the display position of the character in accordance with the audio output timing as shown in Figure 9(b), or changing the audio data. This is done while identifying and displaying the corresponding characters by coloring, inverting, underlining, etc. in accordance with the output.

Note that the setting buffer 5 contains [
If "None" is set, only the display as shown in FIG. 9 described above is performed. If the operator reads out the text based on the displayed information, it becomes possible to read out the document data with the correct accent within the set time as described above.

In this way, according to this device, instead of simply synthesizing document data into speech, it emphasizes words, etc. that should be emphasized according to attribute information added to document data by changing the conditions for outputting the speech. I can do it. As a result, it is possible to obtain audio output that emphasizes the output audio appropriately according to the document content, and it is possible to obtain highly natural document audio with rich content expression that reflects the atmosphere of the document data content. .

Note that the present invention is not limited to the embodiments described above. Emphasis on document data can be achieved not only by increasing the speaking speed, but also by changing the pitch and quality of the voice, and by combining these appropriately, it can also be achieved by changing the atmosphere of the voice output. good. Furthermore, the method of analyzing document data and the method of generating audio data can be changed as appropriate.

In addition, the present invention can be implemented with various modifications without departing from the gist thereof.

[Structure of the Invention] As explained above, according to the present invention, the conditions for audio output of the document data are changed and controlled according to the attributes such as emphasized characters appearing in the document data, so that the information content of the document data can be fully expressed. It is possible to obtain a highly natural voice output that can be expressed in a realistic manner, and it has a great practical effect on outputting various documents.

[Brief explanation of the drawing]

The figures show a document reading device according to an embodiment of the present invention. FIG. 1 is a schematic configuration diagram of the embodiment device, FIG. 2 is a diagram showing a schematic processing procedure flow of the embodiment device, and FIG. 3
The figure is a diagram showing an example of document data, and FIG. 4 is a diagram showing setting conditions for audio output. Furthermore, Fig. 5 is a diagram showing an example of the structure of a word dictionary used for document analysis, Fig. 6 is a diagram showing the analysis results of document data, and Fig. 7 is a diagram showing an example of the structure of a word dictionary used for document analysis.
8 is a diagram showing an example of a conversion rule for audio data, FIG. 8 is a diagram showing an example of a conversion result of audio data, and FIG. 9 is a diagram showing an example of a display of audio data. 1... Document data memory, 2... Control device, 3...
・Input device, 4...Display device, 5...Setting buffer,
6... Document analysis unit, 7... Word dictionary memory, 8-.
・Analysis result buffer, 9... audio data generation section, IO
... Voice data generation rule memory, 11... Voice data memory, 12... Speech synthesis device, 13... Reading speed setting section, 14... Display data creation section, 15...
Display data memory. Applicant's Representative Patent Attorney Takehiko Suzue Figure 1 Figure 2 Figure 3 Figure 4 Figure 6 Figure 7

Claims (2)

[Claims] (1) A means for linguistically analyzing document data expressed in a format that can be processed on a predetermined information processing device, a means for generating audio data from the document data according to the analysis results, and the generated audio data What is claimed is: 1. A document reading device comprising: means for synthesizing and outputting a voice based on the voice; and means for changing and controlling control data for controlling the synthesized voice based on attribute information added to the document data. (2) The document reading according to claim (1), wherein the control of the synthesized output of the voice is performed by varying at least one of the speech rate, volume, sound quality, and pitch of the voice. Device.