JPS61259295A - Voice conversion system for sentence - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] This invention relates to a device that converts text into speech and transmits it, which is being studied for the purpose of helping visually impaired people understand text.

[Conventional technology]

Traditionally, Braille has been used as a method for visually impaired people to understand and decipher a single text.

There is a way to understand by touch, but in recent years, as shown in the conceptual diagram in Figure 6, research on devices that convert text into full sound and listen to it with the ears has become active at research institutes such as universities. Ta. Figure 6 shows the concept, CPU7
The reading device 70 controlled by the computer 3 reads the text on the document and stores it in the storage device 71 as digital information. The information in the storage device 71 is transferred to the audio signal conversion device 72 according to a command from the CPU 73, where it is converted into an audio signal, and the audio signal is transmitted through the amplifier 6 and the speaker 7.

[Problem that the invention seeks to solve]

In such a device, the voice is transmitted as a single voice.1For example, when there are multiple characters in a novel, etc.,
Even if all of this is read by the above-mentioned device, converted into voice, and transmitted, it is worrisome that multiple people can only be expressed with the same single voice, making it monotonous.This point should be completely improved. Ta.

[Means for solving problems]

In order to solve the above-mentioned problems, the present invention attaches a plurality of identification codes to the text read by a reading device, reads these, converts them into identification codes, judges the text according to these codes, and makes a plurality of identification codes. This is something that can be expressed vocally.

That is, 1. In a method of reading all the sentences on a document with a reading device, converting the read data into an audio signal, and transmitting the sentences by voice, the present invention provides an identification code in advance to the sentences to be read and divides the sentences into a plurality of sentences. This text is read by a reading device, encoded with an identification code, and stored in a storage device, and all data in the storage device are classified by identification code and sent to a plurality of audio signal converters.

The present invention provides a complete text-to-speech conversion method characterized by transmitting the text in a plurality of different voices.

[Effect]

The text voice conversion method of this invention attaches an identification code to the text to be read in advance, divides the text into multiple parts, reads the text with a reading device, converts it into all identification codes, and then classifies the text according to each identification code. and sends it to multiple audio signal converters3.
The text is conveyed using multiple different voices, eliminating monotony and making it easier to understand the text.

〔Example〕

FIG. 1 is a block diagram showing the configuration of a speech conversion device for one sentence according to an embodiment of the present invention, and FIG. 2 shows an example of a sentence to be read by this device.

FIG. 3 is an example showing the storage arrangement of information in the storage device after being read by the present device.

The configuration of this device is as follows: 1. A reading device that is controlled by a CPU 3 and reads text from a document. A storage device 2 for storing all the digital information read by the reading device 1; a buffer 4-1 for temporarily holding sentences determined by the CPU 3; 4-2. 4-
3. An audio signal converter 5-1 corresponding to each buffer 4-1, 4-2, and 4-3 converts information in the buffer into an audio analog signal. 5-2. 5-3. It is comprised of multiple amplifiers 6.degree. for amplifying the entire audio signal converted by these audio signal converters, and a speaker 7 for evoking audio with the output of the amplifier 6. The text read by this device is shown in FIG.
3. ...It is entered in the column of 30 (this number does not need to be entered, but will be explained with a code for convenience of explanation). Each character 12 is divided and displayed with 1, for example, a sentence 'k(), (), <> corresponding to a character's words. For example, character A's words are [].

The word B is (), and the word C is <>.

It is assumed that the text used in this device is given such an identification code. It goes without saying that this identification may be made by other indications, such as the numbers 1.2.3 or other symbols, instead of these parentheses.

FIG. 3 shows the storage state of information in the storage device 2 after reading the text in FIG. 2 with the reading device 1 in FIG. (Displayed as m for convenience of explanation).

m of 20-2 = 2, m of 20-3 = 3゜2
m=4.0-4. . . . The column for m=x (x is the maximum number of locations) is blank, and coded digital information indicated by 24 is stored. The last area of each location is marked with an identification code 22 (displayed as Q for convenience of explanation). In this example, the location corresponding to [ ] shown in Figure 2 is marked 23a. The part corresponding to frlJ and 0 is marked with r2J, shown as 22b, and the part corresponding to <> is marked with "3", shown as 22C.

Next, the operation of the embodiment of the present device will be explained according to FIGS. 4 and 5.

FIG. 4 is a flowchart showing the function of reading a text after the CPU 3 of FIG. 1 commands Kanayama to read the text. In FIG. 4, 30 is a functional part of the CPU 3 in FIG. First, as shown in step 31, the CPU 3 selects n =
1 f: set, 32 (n indicates the number of rows) m = 1
Do 2 sections. (m is the location for storing all the data) Next, in step 33, the reading device 1 is instructed to read n=1 line of text on the sheet 10 of FIG. 2.

The read data is then taken into the CPU 3 in step 34. The imported data is judged in step 35 as to whether or not there is [ ] in the sentence. If there is data corresponding to [ ], the process advances to step 36 and an identification code of "1" is added to it.
It is transferred to the storage device 2 together with the data contained within. Storage device 2
Then, as shown at 20-1 in Figure 3, the location m
= Data is stored in 1, and 2 is stored in area Q indicated by 22.
An identification code "1" indicated by 3a is stored. (This process is not specified in the flowchart of Figure 4 and is omitted.) Next, if there is no [ ] in the imported n = 1 line of text, proceed to the second step 37, and next time performs all tests to determine whether or not () exists. If there is ), the process advances to step 38, and as in step 36, this time an identification code "2" is added and the data is transferred to the storage device 2 together with the data in parentheses. In this case, the storage device 2 is the same as described above, but the identification code "2" is stored in the Q area indicated by 22.

If () does not exist, proceed to the next step 39, and this time perform all the determinations as to whether or not > exists. Similarly, this time, the data is transferred to the storage device 2 together with the data in the file with the identification code "3" attached. This concludes the judgment of the text, but if the text shown in Figure 2 to be read is mistakenly marked with any identification number (i.e., C], (), <> in this example)
etc.) is not included, as shown in step 41, the specific identification code ``rl'' or ``
2" or "3", or a special audio signal converter may be provided and an identification code corresponding to this may be attached. The sentences judged in this way are 20 in Figure 3.
-1 as indicated by the content at location m = 1, and 2
For example, Q21 indicated by 23a is stored in Q indicated by 2. Next, the process moves to step 42, where m=m+1, the location shown in FIG.

If everything has not been checked yet, the process moves to step 35 and the same procedure is repeated again. After checking everything on that line, the process proceeds to step 44 n = n
+1 and move on to the next line 6 Next, in step 45, if the first line is not the last line, the process jumps to step 33 and sends a reading command to the reading device 1 to read the text on the first line, and in step 34 the CPU The read data is taken in, and the same procedure as described above is repeated, and when all n, ie, deposit checks are completed in step 45, the process ends.

In this way, the text written on the form 1o shown in FIG.
is attached and stored.

Next, referring to FIG. 5, the operation of converting the data in the storage device 2 into an audio signal by the audio signal converting section and emitting audio from the speaker will be explained.

First, 50 indicates a functional part of the CPU 3.

The CPU 3 sets m=1 in step 51.

In step 52, location m = 1 in storage device 2
Read the data. First, in step 53, "1" is subtracted from the specified area Q of location m = 1, and if the result is Q = 1 and "0" as shown in FIG. 3, the process moves to step 55, and location m = The contents of step 1 are transferred to the buffer (1) shown at 4-1 in FIG. Then, the audio signal converter (1) converts these contents into audio signals, and these signals are transmitted as audio through the amplifier 6 and the speaker 7. If it does not become "0" in step 54, step 57
Proceed to , and this time subtract "2" from the value of Q. This result is 1
For example, the value of Q for m = 2 shown in Figure 3 is ``2''.
”, then it becomes rOJ, so the process goes to step 58, and the contents of that location are changed to 4- as in step 55 above.
The signal is transferred to a buffer (2) indicated by 2, and all commands for conversion into an audio signal are sent to an audio signal converter (2) indicated by 5-2. Next, the process advances to step 59, where the audio signal converter (2) converts these contents into audio signals, and these signals are transmitted as audio through the amplifier 6 and speaker 7.

Next, if rOJ is not achieved in step 60, the process advances to step 61, and this time "3" is subtracted from the value of Q. If this result is 1, for example, ``3'' like the value of Q for m = 3 shown in FIG. The signal is transferred to a buffer (3) indicated by -3, and an instruction is given to an audio signal converter (3) indicated by 5-3 to convert it into an audio signal.

Next, the process proceeds to step 64, where the audio signal converter (3) converts these contents into audio signals, and these signals are sent to the amplifier 6.
, is transmitted as audio through the speaker 7. If Q
If [1 Sat "2j, r3J" is not attached, the process proceeds to step 65, where the specific buffer (1), (
2) or (3), and issue a conversion command to the corresponding audio signal converter to convert it into an audio signal.9 In this example, some kind of signal must be sent in step 41 in Figure 4. Although it is impossible to proceed to this step because an identification code is attached, this function was added just in case, and is not necessarily necessary. Next, when the discrimination of the identification code is completed in this way, the process advances to step 67 and moves to m=m+1 to check all locations in the next storage device, but in step 68 it is selected whether or not all locations have been checked. If there are still locations remaining, proceed to step 52.

Data at the next location is read from the storage device 2 to the CPU, and the same procedure as described above is repeated. Then, once all locations in the storage device 2 have been checked in step 68, the process ends. As described above, sentences are identified by characters, and the sentences are read and coded according to identification codes corresponding to the characters, and stored in the storage device 2. and.

Data is sent to a buffer for each of these identification codes, and a dedicated audio signal converter is provided for each identification code.
Audio signals assigned the same identification code are converted into audio signals by the same audio signal converter. Each of these audio signal converters.

For example, if you change the frequency components of male voice 1 female voice, high voice 1 voice, low voice 1 etc. of the same male voice,
In this example, 5-
1. 5-2. 5-3, each of which is configured to emit three types of voices, but is not necessarily limited to three types. Therefore, according to this embodiment, a sentence in which multiple characters appear in the first novel is written as (), as shown in FIG.
These sentences are identified in advance with (), <>, etc., and these sentences are read by this device and converted into multiple voices by different audio signal converters depending on the character. This eliminates the monotony of conventional methods of communication.

In addition, although Figure 2 shows an example in which all sentences corresponding to the characters are identified using parentheses, they are not identified using parentheses.

For example 1, 2, 3. Needless to say, it is also possible to use numbers such as at the beginning or end of the sentence to identify the sentence.

〔Effect of the invention〕

According to the text-speech conversion method of the present invention described above, the voices of multiple characters in a novel etc. are transmitted in correspondence with different voices, so it is not possible to simply convert the voices of a plurality of characters in a novel etc. Compared to the monotonous transmission method using a single voice, it is now possible to transmit multiple voices, making it a more effective speech conversion device and an effective aid for visually impaired people to understand text.

[Brief explanation of drawings]

FIG. 1 is a block diagram of a text speech conversion method according to an embodiment of the present invention, FIG. 2 is an explanatory diagram showing an example of a text used in the method of the present invention, and FIG. 4 is a flowchart of the text reading method according to the method of the present invention, and FIG. 5 is a flowchart of the function of converting data in the storage device into an audio signal according to the method of the present invention. , FIG. 6 is a block diagram showing the general concept of a conventional speech conversion device. 1...Reading device, 2...Storage device, 3...CPU
. 4-1. 4-2. 4-3...Buffer, 5-1°5-2.5-3...Audio signal converter, 6...Amplifier. 7-2 Heater, 13a, 13b, 14a,
14b, 15a, 15b identification brackets, 23
a, 23b, 23c・in! Separate coat. Figure 2

Claims (1)

[Claims] In this method, a reading device reads the text on a document, converts the read data into an audio signal, and transmits the text audibly.In this method, the text to be read is divided into multiple parts by providing identification codes in advance, and then the text is read. The device reads this sentence, encodes the identification code, and stores it in a storage device.The data in the storage device is divided by identification code and sent to multiple audio signal converters, which converts the sentence into multiple different voices. A text-to-speech conversion method that is characterized by the transmission of text.