JPH09265380A - Method and device for synthesizing voice - Google Patents

Abstract

PROBLEM TO BE SOLVED: To classify the respective text strings of document data to be voice synthesized and to change and read aloud output voice by voice synthesis for respective attributes. SOLUTION: A document structure analysis part 101 classifies the contents of the document data into a citing part and a text based on prescribed symbols (for instance > and ≫) included in the document data. A voice quality data holding part holds the at least two kinds of voice quality data for deciding the voice quality of synthetic voice in the voice synthesis. A voice synthesis part 102 acquires which one of the text and a citing sentence, character string data to be the object of a voice synthesis processing belong to from the analyzed result of the document structure analysis part 101, selects the voice quality data held in the voice quality data holding part 103 corresponding to the attribute and performs the voice synthesis for the character string data based on the selected voice quality data.

Description

Detailed Description of the Invention

[0001]

TECHNICAL FIELD The present invention relates to a speech synthesis method and apparatus. In particular, the present invention relates to a voice synthesizing method and device suitable for converting a text string of e-mail, net news, etc. into voice and reading the voice.

[0002]

2. Description of the Related Art With the spread of computers and word processors,
The digitization of documents is progressing. Further, a document reading technique has been developed in which such a computerized document is processed by a computer, converted into a voice and output.

In recent years, computer networks such as the Internet and personal computer communication services of various companies are rapidly developing. As a result, electronic documents such as electronic mail and net news are becoming more important. E-mails and net news are often sent in the form of "replies" to mails and net news articles sent by others. In this case, it is often the case that a part of the contents of another person's mail or news article is quoted and used.

[0004]

When automatically reading aloud such an electronic mail or netnews by voice synthesis, if the text and the quoted part are read aloud without distinction, it is only necessary to hear the voice. , There is a problem that it is difficult to grasp the contents. In order to solve this problem, it is necessary to change the output voice in the text and the quoted part, but there was no such technology in the past.

The present invention has been made in view of the above-mentioned prior art, and a speech synthesis method capable of classifying each text string of document data to be speech-synthesized by an attribute and changing an output speech for each attribute. And to provide a device.

By the voice synthesizing method and apparatus for achieving the above object, for example, when reading an e-mail or the like, it is possible to read the document by changing the output voice depending on the text of the document and the quoted portion.

[0007]

A speech synthesis apparatus of the present invention for achieving the above object has the following configuration. That is, based on a predetermined symbol included in the document data, a classifying unit that classifies the content of the document data into at least two types of attributes, and a voice quality of synthesized speech in speech synthesis are determined.
Holding means for holding at least two types of voice quality data,
The present invention further includes a selection unit that selects the voice quality data held by the holding unit according to the attribute classified by the classification unit, and a synthesizing unit that performs voice synthesis based on the voice quality data selected by the selection unit.

Further, preferably, the classification of the attribute of the document data by the classifying means is performed for each line of the document data.

Preferably, the classification means classifies the attributes of each line of the document data based on whether or not a character string matching a predetermined regular expression including a predetermined symbol exists in the line. .

Further, preferably, the classification means classifies the attributes of each line of the document data based on whether or not a character string having a predetermined symbol at the end is a predetermined number of characters or less.

Further, preferably, the selecting means selects the voice quality data based on the attribute classified by the classifying means and the character string near the predetermined symbol. For example, when a character string indicating a citation source is arranged in the vicinity of a predetermined symbol (for example, immediately before the predetermined symbol), voice output is performed with a different voice quality for each citation source, further facilitating understanding of the document content. Become.

Further, preferably, the classification means classifies the contents of the document data into a citation sentence and a body text based on a predetermined symbol included in the document data.

Further, preferably, based on the information included in a predetermined line of the document data, whether the voice synthesis of the content classified into the predetermined attribute of the document data is performed with a male voice quality or a female voice quality. It further comprises a determining means for determining.

[0014]

DETAILED DESCRIPTION OF THE INVENTION An embodiment of the present invention will be described below with reference to the accompanying drawings.

<First Embodiment> FIG. 1 is a block diagram showing the functional arrangement of a document reading device according to the first embodiment. In the figure, reference numeral 101 denotes a document structure analysis unit, which distinguishes a text body and a quoted portion in document data. Here, the document structure analysis unit 101 distinguishes the text from the quoted part based on the quote symbol “>” often used in electronic mail and netnews. A voice synthesis unit 102 performs voice synthesis with a voice quality based on voice quality data. 1
A voice quality data holding unit 03 holds a plurality of types of voice quality data used by the voice synthesis unit 102. Here, the voice synthesizing unit 102 performs voice synthesis by switching the voice quality data to be used between the main body and the quoted sentence distinguished by the document structure analyzing unit 101, and outputs synthesized voices having different voice qualities.

As a result, when the contents of electronic mail or net news are read aloud by voice synthesis, the voice quality can be changed between the body text and the quoted portion, and the body text and the quoted portion can be distinguished only by listening to the output voice. It is possible,
It becomes possible to grasp the contents more accurately.

FIG. 2 is a block diagram showing the configuration of the document reading device according to the first embodiment. In the figure, 2
Reference numeral 1 denotes a control memory, which stores a control program according to a control procedure as shown in the flowchart of FIG. 22
Is a central processing unit for making judgments and calculations according to the control procedure held in the control memory 21. Reference numeral 23 is a memory, and 24 is a disk device, which holds voice quality data, an input document, and the like. 25 is a bus.

Next, the operation of this apparatus will be described with reference to the flow chart shown in FIG. FIG. 3 is a flow chart for explaining the reading procedure of the document data in the first embodiment. Further, FIG. 4 is a diagram showing an example of document data as a target of document synthesis.

First, each variable is initialized (step S3).
01). In this process, a buffer S for holding one line read from the input document, a buffer S ′ for holding the preceding one line, and a buffer B for holding a part of the document contents are used. Then, each buffer is initialized to an empty string (indicated by Φ). In this process, the flag indicating whether the line S is a part of a book or a part of a quote is I (S), and the preceding line S'is a part of the text or a part of the quote. Let I (S ') be a flag indicating that. In step S301, I (S ') is initialized to 0.

Next, the line character string currently held in the buffer S is copied to S ', and a new line from the input document is read into the buffer S (step S302). Then, it is checked whether the line held in the buffer S indicates the end of the file of the input document (step S303).

If it is not the end of the file, it is determined whether the buffer S is a part of the text or a quoted part (step S304). In this example, if the buffer S satisfies the following condition, it is determined to be a part of the quoted portion, and if not, it is determined to be a part of the text. The condition is that it matches the regular expression "^. * >>" or "^. * >>" and that the character string before the quotation marks ">>>" or ">" is N characters or less. is there. For example, as shown in FIG.
The line starting with ag >> will match this condition.

Here, for example, "^. * >>" means that an arbitrary character string exists from the beginning of the character string and ends with a quotation mark >>. In addition to the above, preferably, the AND condition does not start with “<<”, (<
If it starts with <, it is not a quoted part).

If S is judged to be part of the quoted portion, then I
(S) = 1 is set (step S305), and if it is determined to be a part of the text, I (S) = 0 is set (step S306).
Then, the value of I (S) is checked (step S30
7), if I (S) ≠ I (S ′), the contents of the buffer B are sent to the voice synthesizer 102. In the voice synthesis unit 102,
In accordance with the value of I (S), the voice quality data of the main body and the voice quality data of the quoted portion are switched, the contents of the buffer B are converted into voice and output (step S308).

Here, in step S308, I
When (S) = 1, I (S ′) = 0 and it can be seen that the data held in the buffer B up to that point is a part of the text. Therefore, when I (S) = 1, the voice quality data of the body is selected and the voice synthesis is executed. Similarly, I (S) =
In the case of 0, I (S ′) = 1, and it can be seen that the data held in the buffer B by that time is a part of the quoted text. Therefore, when I (S) = 0, the voice quality data of the quoted sentence is selected and the voice synthesis is executed.

When the voice synthesis is executed, the character string in the buffer B is made an empty character string (Φ), and the process proceeds to step S309.
In step S309, the contents of row S are added to buffer B, and the process returns to step S302.

As described above, according to the first embodiment, document data can be read aloud with different voice qualities between the quoted text and the text. In particular, since the quoted text and the text are distinguished based on the quoted symbols generally used in electronic mail and netnews, the reading device is suitable for reading this kind of document data.

<Second Embodiment> In the above first embodiment, an example in which two types of voice quality data, that is, the body text and the quoted portion, are switched has been described. However, in reality, there are cases where the contents transmitted by a plurality of callers are quoted at the same time. In such a case, a character string indicating the sender of the quotation is often described immediately before the quotation mark (herein, such a character string is referred to as a quotation source mark). In FIG. 4, “yag” is a character string (quotation source symbol) indicating the sender. Therefore, a plurality of voice quality data for a quoted portion are prepared in advance in consideration of automatically reading aloud a document in which the transmission contents of a plurality of callers are simultaneously quoted. Then, the document structure analysis unit 1
In 01, the quotation part may be distinguished for each caller based on the quotation mark, and appropriate voice quality data may be assigned to each to synthesize the voice.

For example, the document structure analysis unit 101 identifies the citation source according to the character string before the quotation mark ">", and registers the voice quality data and the citation source each time a new citation source appears. To generate. By referring to this registration table, it becomes possible to adopt the same voice quality data for the same citation source. When a citation source not registered in the registration table appears, voice quality data not registered in the registration table is selected from the voice quality data holding unit 103, assigned to the citation source, and additionally registered in the registration table.

The male voice and the female voice may be distinguished. The gender of the sender of the mail or news article or the sender of the quoted part is estimated from the header information of the email or net news article or the quoting source symbol (e
In the case of -mail, the e-mail address described in the "From" line of the mail header represents the sender), and the male voice and the female voice may be switched according to this estimation result. Alternatively, a character string indicating the sender of a mail or a news article may be registered in association with a gender, and a male voice and a female voice may be distinguished by referring to this.

Even if the present invention is applied to a system composed of a plurality of devices (for example, host computer, interface device, reader, printer, etc.), a device composed of one device (for example, copying machine, facsimile) Device).

Further, an object of the present invention is to supply a storage medium having a program code of software for realizing the functions of the above-described embodiment to a system or apparatus, and to supply a computer (or CPU) of the system or apparatus.
And MPU) read and execute the program code stored in the storage medium.

In this case, the program code itself read from the storage medium realizes the functions of the above-described embodiment, and the storage medium storing the program code constitutes the present invention.

A storage medium for supplying the program code is, for example, a floppy disk, a hard disk, an optical disk, a magneto-optical disk, a CD-ROM, a CD.
-R, a magnetic tape, a nonvolatile memory card, a ROM, or the like can be used.

Further, by executing the program code read by the computer, not only the functions of the above-described embodiment are realized, but also the OS (operating system) running on the computer based on the instruction of the program code. It is needless to say that this also includes a case where the above) performs a part or all of the actual processing and the processing realizes the functions of the above-described embodiments.

Further, after the program code read from the storage medium is written in the memory provided in the function expansion board inserted into the computer or the function expansion unit connected to the computer, based on the instruction of the program code, It goes without saying that a case where the CPU or the like included in the function expansion board or the function expansion unit performs some or all of the actual processing and the processing realizes the functions of the above-described embodiments is also included.

When the present invention is applied to the above-mentioned storage medium, the storage medium stores the program code corresponding to the above-mentioned flow chart. Briefly, the memory map example of FIG. Each module shown will be stored in the storage medium.

That is, at least the program code of each of the “classification processing module”, the “selection processing module” and the “synthesis processing module” may be stored in the storage medium.

Here, the classification processing module is a program module that performs classification processing for classifying the contents of the document data into at least two types of attributes based on a predetermined symbol included in the document data. The selection processing module includes a holding unit that holds at least two types of voice quality data for determining the voice quality of the synthesized voice in the voice synthesis according to the attribute classified by the classification process of the character string data to be subjected to the voice synthesis. It is a module that realizes a selection process for selecting any voice quality data. The synthesis processing module is a module for performing synthesis processing for performing voice synthesis based on the voice quality data selected in the selection processing.

[0039]

As described above, according to the present invention,
It is possible to classify each text string of the document data to be speech-synthesized by the attribute and change the output speech for each attribute. Therefore, for example, when reading an e-mail or the like, it is possible to read the document by changing the output voice depending on the text and the quoted part of the document, and the distinction between the text and the quoted part becomes clear only by listening to the output voice. The effect is that the content can be grasped more accurately.

[0040]

[Brief description of drawings]

FIG. 1 is a block diagram showing a functional configuration of a document reading device according to a first embodiment.

FIG. 2 is a block diagram showing a configuration of a document reading device according to the first embodiment.

FIG. 3 is a flowchart illustrating a reading procedure of document data according to the first embodiment.

FIG. 4 is a diagram showing an example of document data as a target of document synthesis.

FIG. 5 is a diagram showing an example of a memory map of a storage medium that stores a control program according to the present invention.

[Explanation of symbols]

 21 control memory 22 central processing unit 23 memory 24 disk device 25 bus 101 document structure analysis unit 102 speech synthesis unit 103 voice quality data holding unit

Front page continuation (72) Inventor Tsuyoshi Yagisawa 3-30-2 Shimomaruko, Ota-ku, Tokyo Canon Inc. (72) Inventor Minoru Fujita 3-30-2 Shimomaruko, Ota-ku, Tokyo Canon Inc. Within

Claims (15)

[Claims] 1. A classification means for classifying the content of the document data into at least two types of attributes based on a predetermined symbol included in the document data, and at least two types of voice qualities for determining the voice quality of synthesized speech in voice synthesis. Holding means for holding the data, selecting means for selecting the voice quality data held by the holding means in accordance with the attribute of the character string data to be synthesized which is classified by the classifying means, and the selecting means A speech synthesizing device, comprising: a synthesizing unit for synthesizing speech with respect to the character string data based on selected voice quality data. 2. The speech synthesizer according to claim 1, wherein the classification of the attribute of the document data by the classifying unit is performed on a line-by-line basis of the document data. 3. The classifying means classifies the attribute of each line of each line of the document data based on whether or not a character string matching a predetermined regular expression including a predetermined symbol exists in the line. The speech synthesizer according to claim 2. 4. The classifying means classifies the attribute of each line of each line of the document data based on whether or not a character string ending with a predetermined symbol is a predetermined number of characters or less. The speech synthesizer according to 2. 5. The voice synthesis according to claim 1, wherein the selecting unit selects the voice quality data based on the attribute classified by the classifying unit and a character string near the predetermined symbol. apparatus. 6. The voice synthesizing apparatus according to claim 1, wherein the classification unit classifies the content of the document data into a quoted text and a text based on a predetermined symbol included in the document data. 7. The apparatus further comprises a deciding means for deciding a voice quality to be used at the time of speech synthesis of contents classified into a predetermined attribute of the document data, based on information included in a specific line of the document data. The speech synthesizer according to claim 1. 8. A classifying step of classifying the contents of the document data into at least two types of attributes based on a predetermined symbol included in the document data, and a classifying step of the character string data to be combined processing. A selection step of selecting one of the voice quality data from a holding means for holding at least two types of voice quality data for determining the voice quality of the synthesized voice in the voice synthesis according to the attribute, and the voice quality selected in the selection step. And a synthesizing step of synthesizing the voice with respect to the character string data based on the data. 9. The speech synthesis method according to claim 8, wherein the classification of the attribute of the document data in the classifying step is performed for each line of the document data. 10. The classifying step classifies the attributes of each line of the document data based on whether or not a character string that matches a predetermined regular expression including a predetermined symbol exists in each line of the document data. The speech synthesis method according to claim 9. 11. The classifying step, for each line of the document data, classifies the attributes of each line based on whether or not a character string ending with a predetermined symbol is a predetermined number of characters or less. 9. The voice synthesis method according to item 9. 12. The voice synthesis according to claim 8, wherein the selecting step selects voice quality data based on the attribute classified in the classifying step and a character string near the predetermined symbol. Method. 13. The voice synthesizing method according to claim 8, wherein the classifying step classifies the content of the document data into a quoted text and a text based on a predetermined symbol included in the document data. 14. The method according to claim 14, further comprising a determining step of determining a voice quality to be used at the time of voice synthesis of a content classified into a predetermined attribute of the document data based on information included in a specific line of the document data. The speech synthesis method according to claim 8. 15. A computer-readable memory that stores a control program for performing voice synthesis output based on document data, wherein the content of the document data is at least two types based on a predetermined symbol included in the document data. At least two types of voice quality data for determining the voice quality of the synthesized voice in the voice synthesis according to the code of the classification process for classifying into attributes and the attribute of the character string data to be subjected to the synthesis process classified in the classification process. A selection step code for selecting one of the voice quality data from a holding means for holding, and a synthesis step code for performing voice synthesis on the character string data based on the voice quality data selected in the selection step. A computer-readable memory characterized by.