JPH07219583A - Method and device for speech processing - Google Patents

Abstract

PURPOSE:To enable smooth operation matching human sensations by considering a speech recognition result to be confirmed and outputting corresponding action information when the speech recognition result is displayed and no speech is inputted within a certain permissible time. CONSTITUTION:A speech recognizing process part 2 performs a speech recognizing process for a speech 'henshuu' (editing in English) inputted from a microphone 1 by using a word dictionary 3, holds 'henshuu' in KANJI(Chinese character) as its result in a speech recognition result holding part 4, and displays it at a state display part 6. A recognition result control part 5 confirms whether or not a time previously set by a timer 7 is elapsed and when some speech, e.g. 'chigau' (wrong) is inputted within the time, it is considered that the result displayed at the display part 6 is canceled, so that an input wait state is entered. When the previously set allowable timer is elapsed and no other speech is inputted, it is confirmed that the displayed result is correct; and action information corresponding to 'henshuu' in KANJI stored in the speech recognition result holding part 4 is outputted to an information output part 8.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a voice processing method and apparatus capable of performing various operations, which are generally called commands, for operating an application such as a computer by voice. It is a thing.

[0002]

2. Description of the Related Art Generally, in an application such as a computer, the operation of the application is generally designated by a keyboard input or a mouse. Furthermore, recently, voice input has begun to appear.

[0003]

However, in the conventional and current speech recognition techniques, correct recognition with 100% accuracy has not been realized yet. Also, it will be difficult to realize it in the near future. Therefore, if the application is erroneously recognized, the application once executes the operation resulting from the erroneous recognition, so that the user needs to return to the original state and then perform the operation again. There is a drawback that it may increase more than in the case of. Therefore, it is also considered to make the user confirm the voice recognition result with an input device other than the voice such as a keyboard each time, and then perform the operation. Since it is necessary, there is a drawback that it is troublesome, and as a result, the psychological state of the user may be deteriorated, the smooth operation of the user may be hindered, and distrust of the computer may be unnecessarily increased.

[0004]

The present invention eliminates the above-mentioned drawbacks and presents a voice recognition result to a situation presenting means,
If no other control information (voice, keyboard, etc.) is input within a certain grace period measured by the time measuring means, it is determined that the confirmation has been made by the user, and the action corresponding to the voice recognition result is determined. Output information,
On the other hand, some other control information (voice, keyboard, etc.)
When is input, a recognition result control means for invalidating the previous speech recognition result is provided, and as a result, a smooth operation that matches the human sense can be performed.

[0005]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described in detail below with reference to the drawings. FIG. 1 is a block diagram showing the configuration of an apparatus according to an embodiment of the present invention. In the figure, 1 is a microphone for inputting voice, 2 is a voice recognition processing unit that recognizes the input voice, 3 is a word dictionary used in the voice recognition processing unit 2, and 4 is recognized by the voice recognition processing unit 2. The voice recognition result holding unit 5 which holds the result is referred to the information from the timer 7 and presents the result held in the voice recognition holding unit to the situation presenting unit 6 with reference to the word dictionary 3. Alternatively, the recognition result control unit 6 for controlling the invalidation, the determination, etc.
A situation presenting unit that presents a situation such as a result held in the voice recognition holding unit 4, a timer 7 that measures elapsed time,
An information output unit 8 refers to the word dictionary 3 and outputs action information corresponding to the recognition result finally determined by the recognition result control unit 5.

FIG. 2 is a flow chart showing the processing procedure of the operation in the apparatus shown in FIG. With reference to this figure, the operation procedure of the present invention will be described by taking as an example the operation by the input voice "Henschu" shown in FIGS.

In FIG. 2, in step S0, initialization such as timer reset is performed, and the process proceeds to step S1.

In step S1, a voice input is waited for from the microphone 1 on the screen for waiting for voice input as shown in FIG. 4A. For example, when "Henshu" is input,
Go to step S2.

In step S2, the voice recognition processing unit 2 performs voice recognition processing on the input voice "Henschu" using the word dictionary 3. During the voice recognition process, that effect is displayed as shown in FIG. The result "edit" is held in the voice recognition result holding unit 4, and the process proceeds to step S3.

Next, in step S3, in the recognition result control unit 5, the result "edit" stored in the voice recognition result holding unit 4 is displayed in the situation presenting unit 6 as shown in FIG. 4 (c). Present and move to step S4.

In step S4, the recognition result control section 5 resets the timer 7 and enters a preset mode for confirming the grace time. Here, the time measured by the timer 7 may be notified to the user as indicated by 41 in FIG. 4C. Here, an example is shown in which "." Is displayed every 0.5 seconds.

In step S5, the recognition result control section 5 confirms at predetermined time intervals whether or not a preset time has elapsed. If the set time has not elapsed, the process proceeds to step S6, and if the set time has elapsed, the process proceeds to step S7.

In step S6, it is confirmed whether or not any voice, for example, "chigau" is input, and if the voice is input, it is determined that the one presented in the situation presenting section 6 is canceled. The content (that is, the message in the input waiting state) to that effect is presented to the situation presentation unit 6 as shown in FIG. 4A, and the process proceeds to step S0. If no voice is input, the process proceeds to step S5.

On the other hand, when the preset grace time of 2 seconds has elapsed and no other voice is input during that time, the situation presenting section 6 of FIG. 4C is presented in step S7. It is the action information corresponding to the "edit" stored in the voice recognition result holding unit 4 from the information stored in the word dictionary 3 by interpreting that the object has been accepted (confirmed as correct) by the user. "Comma
"nd-e" is output to the information output unit 8, and as a result, the application executes the action to end the processing related to the input voice this time, and the display is also returned to the same state as (a) as in FIG. 4 (d). .

FIG. 3 shows an example of the contents of the word dictionary 3.

In this embodiment, the recognition result control means has explained the method of canceling when another voice is input within the preset grace time. However, the keyboard and mouse are canceled within the preset grace time. The same processing may be applied to the input of control information from other input devices such as.

In this embodiment, the so-called Macondo of the application has been described as a word to be stored in the word dictionary, but the present invention is not limited to this, and an arbitrary word and its information may be stored. Good.

In this embodiment, the recognition result control means is described as an example in which the recognition result control target is all the words stored in the word dictionary, but the invention is not limited to this. It is also possible to add a mark to each word stored in the word dictionary as to whether or not to be a target of the recognition result control, and to perform the recognition result control only for the marked word.

In the present embodiment, the recognition result control means has been described with the preset grace time being 2 seconds, but the present invention is not limited to this, and the user may set the grace time at any time. Good.

In the present embodiment, the recognition result control means has explained the method of canceling when another voice is input within the grace time that has been set, but it is not canceled but first. The next candidate of the result recognized by may be displayed in order.

In this embodiment, an example in which one word dictionary is processed has been described, but the present invention is not limited to this. For example, a word dictionary in which a so-called command of an application is registered and recognition result control are provided. In the means, a plurality of dictionaries, such as a control dictionary in which control words that can be input within a preset grace period are registered, may be used separately.

In the present embodiment, Japanese has been described as an example, but the present invention is not limited to this, and any language such as English or German can be applied.

In this embodiment, in the situation presenting section,
Although the example of presenting the situation by characters is described, the present invention is not limited to this, and the situation may be presented by other audiovisual information such as a picture.

[0024]

As described above, according to the present invention,
After presenting the voice recognition result to the situation presenting means, if no voice is input within a certain grace period measured by the time measuring means, the user has confirmed that the voice recognition result is correct. And output the action information corresponding to the voice recognition result, while if any voice is input, consider that the user wants to cancel the voice recognition result and provide a control to invalidate the voice recognition result. As a result, it is possible to obtain an effect that a smooth operation that matches the human sense can be performed.

[Brief description of drawings]

FIG. 1 is a functional block diagram according to an embodiment of the present invention.

FIG. 2 is a flowchart showing a processing procedure according to the embodiment of the present invention.

FIG. 3 is a diagram showing an example of contents of a word dictionary for explaining an embodiment of the present invention.

FIG. 4 is a diagram showing an example showing a process of specific processing for explaining an embodiment of the present invention.

Claims (4)

[Claims] 1. An input unit for inputting a voice, a recognition unit for recognizing the input voice, a result holding unit for holding the recognized result, and a presenting unit for presenting the held voice recognition result. A time measuring means for measuring the passage of a predetermined time, and a control means for controlling to execute a process according to the voice recognition result held in the result holding means after the passage of the predetermined time by the time measuring means A voice processing device comprising: 2. The voice according to claim 1, wherein, when an invalidation instruction is input before the predetermined time is measured by the time measuring means, the contents held in the result holding means are invalidated. Processing equipment. 3. Inputting a voice, recognizing the input voice, holding the recognized result, presenting the held information, and starting measurement of a lapse of a predetermined time, lapse of the predetermined time Is measured, and then control is performed so as to execute processing according to the held information. 4. The information processing method according to claim 3, wherein if the invalidation instruction is input before the predetermined time is measured, the held information is invalidated.