JPH0228826A - Sound recognizing device - Google Patents

Abstract

PURPOSE:To improve operability and input efficiency by moving a cursor to a head character position, to which a sound is inputted just before move indicating operation, when the move operation of the cursor is executed and the cursor is in a next display position. CONSTITUTION:A sound recognizing device 1 is composed of a microphone 2 and a CPU part 5, etc., and a display part 9 and a keyboard 10 are connected. For a sound signal to be collected by this microphone 2, a feature parameter is computed by a feature extracting part 4. Then, the CPU part 5 determines a syllable block and prepares the sound pattern of the block on the basis of this parameter and the block and pattern are stored to an input pattern storing part 8. For this sound pattern and the plural sound patterns of a standard pattern storing part 6, distance computation is executed in a matching part 7. Then, a syllable candidate is prepared according to a computed result and stored in a memory 5a and displayed on a display part 9. Further, the cursor move, etc., at the time of display is automatically executed to the next display position with a sound input.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to a speech recognition device suitable for use in Japanese word processors, personal computers, and the like.

Prior Art FIG. 4 is a diagram showing a method of moving cursor A in a typical prior art speech recognition device. For example, in a Japanese word processor, etc., you can say "Akai" by voice.
When the recognition result is displayed as shown in Fig. 4 (1), the recognition result of the first syllable is incorrect, and therefore the fourth syllable is indicated by diagonal lines in Fig. 4. In Figure (1), it is necessary to move and correct the cursor A at the next input position. Furthermore, this figure 4 (1)
The characters displayed before the first syllable "ba" have already been converted into kana-kanji.

To make a correction, first operate the "-" key repeatedly, go through the states shown in Figure 4 (2) and Figure 4 (3), and then move the cursor as shown in Figure 4 (4). Move A to the first syllable. With the cursor A moved in this way, for example, by calling up the next candidate based on the recognition result, or by manually pressing the key, as shown in FIG. 4 (5),
Modify this first syllable and use the "→" key to make the fourth syllable.
After moving the cursor A to the next input position shown in FIG. 6, operations such as kana-kanji conversion are performed.

Problems to be Solved by the Invention In the prior art as described above, when a voice is input, the
As shown in 1), when cursor A is at the next input position and it is necessary to correct syllable candidates near the beginning position such as the first syllable or second syllable, the number of characters input is large. This method has the drawback that the "←" key must be operated repeatedly, and therefore the operation is complicated.

An object of the present invention is to provide a speech recognition device with improved operability when correcting syllable candidates, which is necessary when input speech is misrecognized.

Means for Solving the Problems The present invention recognizes input speech syllable by syllable, creates one or more syllable candidates for each syllable, visually displays the syllable candidates as a recognition result, and also displays the syllable candidates visually as a recognition result. In a speech recognition device that moves to the next display position and visually displays the recognition result to be recognized in response to the next voice input, the cursor is instructed to move in the direction l\ according to the display order of the recognition results, or in the opposite direction. When the displayed recognition result is corrected by moving the cursor to the correction location, the cursor is at the next display position immediately before the cursor movement instruction operation, and the cursor is moved in the direction according to the display order of the recognition results. When an instruction is given to do something, the cursor is moved to the position of the first syllable candidate that was voice input immediately before the instruction, and thereafter, each time the cursor movement instruction operation is performed, the cursor is moved to the position of the first syllable candidate that was inputted by voice immediately before the instruction, and thereafter, each time the cursor movement instruction operation is performed, the cursor is moved to the position of the first syllable candidate that was inputted by voice immediately before the instruction. The speech recognition box is characterized by having a cursor movement control means for sequentially moving a cursor to syllable candidates.

In addition, the present invention recognizes input speech syllable by syllable, creates one or more syllable candidates for each syllable, visually displays the syllable candidates as a recognition result, and moves the cursor to recognize the next speech input. In a speech recognition device that moves to a next display position to visually display the recognition results to be displayed, instructing the cursor to move in the direction according to the display order of the recognition results or in the opposite direction to move the cursor to the correction location;
When modifying the displayed recognition results, the cursor must be at the position of the first syllable candidate displayed immediately before the cursor movement instruction operation, and the cursor must be moved in the opposite direction to the display order of the recognition results. cursor movement control means for moving the cursor to the next display position when instructed to do something, and then sequentially moving the cursor toward the first syllable candidate each time the cursor movement instruction operation is performed; This is a speech recognition device characterized by having the following features:

Effect According to the present invention, input speech is recognized syllable by syllable,
As a result of the recognition, one or more syllable candidates are visually displayed. Along with displaying this syllable candidate, a cursor is displayed by the cursor movement control means at the next display position where the syllable candidate of the next input voice is displayed. Therefore, the cursor is automatically moved to the next display position in response to voice input. Further, the cursor is moved in the direction according to the display order of the recognition results, or in the opposite direction, and the position to which the cursor is moved becomes the location where the recognition results can be corrected.

When moving the cursor, move in the direction according to the display order with the cursor at the next display position.
When the cursor is instructed to move, the cursor is moved by the cursor movement control means to the position of the first syllable candidate that was input by voice immediately before the instruction, and thereafter, each time the movement instruction operation is performed, the cursor is moved to the position of the first syllable candidate. are sequentially moved to the syllable candidates that follow.

Further, according to the present invention, when moving the cursor,
When the cursor is located at the position of the displayed first syllable candidate and is instructed to move in the opposite direction to the display order, the cursor is moved to the next display position by the cursor movement control means. Thereafter, each time a movement instruction operation is performed, the syllable candidate is sequentially moved toward the first syllable candidate.

Embodiment FIG. 1 is a block diagram showing the configuration of a speech recognition device 1 according to an embodiment of the present invention. The audio signal picked up by the micropon 2 passes only the audio band and is given to the feature extraction unit 4 via the amplification unit 3 which performs amplification. Feature extraction unit 4
calculates feature parameters used for extracting syllable sections and calculating distances for pattern matching, which will be described later, from the input speech waveform.

The 010 section 5 determines a syllable section based on the characteristic parameters, creates a speech pattern for the syllable section, and stores it in the input cover pattern storage section 8. The matching section 7 calculates the distance between the input voice pattern and a plurality of voice patterns stored in the standard pattern storage section 6.

The matching results calculated in this way are sent again to the 010 unit 5, where syllable candidates are created and stored in the memory 5a in the 020 unit 5. This syllable candidate is also displayed on the display section 9.

Display section 9 and keyboard 1 for instructing cursor movement
0 is an external device, and a device to which the speech recognition device 1 is connected, such as a word processor, is used. The CPU section 5 controls the movement of the cursor in response to input from the keyboard 10 and the like.

In the speech recognition device 1 configured as described above, when "Akai" is input from the microphone 2, 020 part 5
The recognition results stored in the memory 5a are displayed on the display screen of the display section 9, for example, as shown in FIG. 2(1). Immediately after voice input is performed in this way,
The cursor indicated by reference numeral 11 is automatically moved to the next display position where the result recognized by the next voice input +"- is displayed. In this figure 2 (1), the cursor is the first character. The characters displayed before ``ba'' have already been converted into kana-kanji.

Also, since this first character "ba" is misrecognized, it needs to be corrected. By pressing the "→" key on the keyboard 10, the cursor 11 is moved to the position shown in FIG. 2 (2). You can move to the first character position. In this state, the next candidate of the recognition result stored in the memory 5a in the 020 unit 5 can be called up, or the keyboard 1o can be called up.
This first syllable is corrected by the input operation from , as shown in FIG. 2 (3).

When the first character has been corrected in this way, by pressing the "←" key on the keyboard 10, the second
As shown in Figure (4), the cursor 11 can be moved to the next display position.

In this state, by operating the keyboard 10, for example, kana-kanji conversion is performed.

FIG. 3 is a flowchart for explaining the operation. In step n1, it is determined whether the [→] key on the keyboard 10 has been operated, and if so, it is determined in step n2 whether the cursor 11 is at the next display position, and if so, the process moves to step n3. The cursor 11 is moved to the position of the first character and the process returns to step n1.

In step n1, if the "→J" key has not been operated, the process moves to step n4, and it is determined whether the "←" key has been operated. If so, the process moves to step n5, and the cursor 11 is positioned at the position of the first character. It is determined whether there is or not, and if so, the cursor 11 is moved to the next display position in step n6 and the process returns to step n1.

If the cursor 11 is not at the next display position in step n2, and if the cursor 11 is not at the first character position in step n5, the process moves to step n7 and the "→" button is pressed.
A normal movement operation of the cursor 11 is performed in which the cursor 11 is moved syllable by syllable in response to one operation of the key or the "←" key, and the process returns to step n1.

If the "←" key is not operated in step n4, then in step n8 other key processing operations corresponding to human power, such as kana-kanji conversion and calling up syllable candidates, are performed, and the process returns to step n1.

In the cursor movement method described above, when the "→" key is operated while the cursor 11 is at the next display position, the cursor 11 is moved to the position of the first character input by voice immediately before the operation. , and on the other hand, cursor 1
When the "←" key is operated with 1 at the first character position, the cursor 11 is moved to the next display position. Therefore, when a large number of characters have been input, it is not necessary to repeatedly operate cursor movement keys such as "→" and "←" keys to move the cursor 11, as described in the prior art, and input voice It is possible to significantly improve the operability when correcting syllable candidates, which is necessary when syllable candidates are misrecognized, and to dramatically improve input efficiency.

Effects of the Invention As described above, according to the present invention, when the cursor is instructed to move in the direction according to the display order of characters, when the cursor is at the next display position, the cursor is moved by the voice immediately before the movement instruction operation. Since the cursor moves to the position of the first character that was input, there is no need to repeatedly move the cursor, and the cursor can be moved to the position of the first character with a single movement, thereby preventing misrecognition of the input voice. In this case, the operability and input efficiency when correcting syllable candidates can be significantly improved.

Further, according to the present invention, when the cursor is directed to move in the opposite direction to the character display order, and the cursor is at the position of the first character of the input syllable candidate, the cursor moves to the next display position. Since the cursor can be moved, there is no need to repeatedly move the cursor, and the cursor can be moved to the next display position with a single movement operation. The operability and input efficiency during correction can be significantly improved.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of a speech recognition device 1 according to an embodiment of the present invention, and FIG. 2 shows a cursor 11 of the speech recognition device 1.
FIG. 3 is a flowchart for explaining the operation, and FIG. 4 is a diagram showing a method of moving cursor A in a conventional speech recognition device. 1.-Voice recognition device, 2.. Microphone, 3..
Amplifying section, 4... Feature extraction section, 5... CPU section, 6.
...Standard pattern storage section, 7...Matching section, 8.
...Input cover pattern storage section, 9-...Display section, 10...
Keyboard agent patent attorney number of strokes

Claims (2)

[Claims] (1) Recognize input speech syllable by syllable, create one or more syllable candidates for each syllable, visually display the syllable candidates as recognition results, and recognize the cursor with the next speech input. in a speech recognition device that moves to a next display position to visually display the cursor, instructing the cursor to move in the direction according to the display order of the recognition results, or in the opposite direction to move the cursor to the correction location;
When modifying the displayed recognition results, if the cursor is at the next display position immediately before the cursor movement instruction operation, and the cursor is instructed to move in the direction according to the display order of the recognition results; , move the cursor to the position of the first syllable candidate that was voice input immediately before the instruction, and then, each time the cursor movement instruction operation is performed, the cursor is sequentially moved to the syllable candidates following the first syllable candidate. A speech recognition device characterized by having a cursor movement control means for moving the cursor. (2) Recognizes the input speech syllable by syllable, creates one or more syllable candidates for each syllable, visually displays the syllable candidates as the recognition result, and recognizes the cursor with the next speech input. In a speech recognition device that moves to a next display position to display the results visually, the cursor is directed to move in the direction according to the display order of the recognition results, or in the opposite direction, and moved to the correction location;
When modifying the displayed recognition results, the cursor must be at the position of the first syllable candidate displayed immediately before the cursor movement instruction operation, and the cursor must be moved in the opposite direction to the display order of the recognition results. cursor movement control means for moving the cursor to the next display position when instructed to do something, and then sequentially moving the cursor toward the first syllable candidate each time the cursor movement instruction operation is performed; A voice recognition device comprising: