JPH0516604B2 - - Google Patents

Description

[Detailed description of the invention]

[Field of Application of the Invention] The present invention provides, for example, information on various types of information (e.g., bank account balances and deposits) over a telephone line.
In a voice response/notification service system that automatically responds (provides) desired information by voice in response to a request (inquiry) from a user, or automatically notifies the user of predetermined information that has occurred, the user The present invention relates to a speech recognition control method for improving the recognition rate of a speech recognition device used to identify input speech from a telephone. [Prior Art] The voice response/notification service system as described above was initially applied only to users' telephones of the push-button dial type (PB type).
Therefore, the type of service request for this system, etc. will be determined after the user's PB telephone is connected to the system and after it is input to the system by the PB signal (combination of specific frequencies within the voice band). The process was beginning to proceed. Incidentally, recently, there has been a growing demand for this type of service (for example, services such as deposit balance inquiries and deposit notifications in banking services) to be widely provided. However, the user's phone is not necessarily
Not a PB type, but rather a dial type (DP type)
With DP telephones, it is not possible to input service request types using PB signals as described above. Furthermore, even if the user's telephone is a PB type, the user may wish to receive services through voice input rather than mechanical operation. In such a case, it is necessary to identify the voice input from the user's telephone using a voice recognition device, input the service request, etc. into the system, and provide the desired service. In this case, the input voice is transmitted through a telephone line, and a user who is unfamiliar with the voice recognition device (hereinafter referred to as a beginner, whereas a user who is accustomed to it is referred to as an expert) may have difficulty in inputting and speaking. Despite the uncertainty, conventional speech recognition devices of this type, for example, only provide one type of voice input guidance (input reminder message), and especially beginners' input voice may be inappropriate or inappropriate. As a result, the recognition rate inevitably becomes low, and also causes a decrease in the overall recognition rate. [Object of the Invention] An object of the present invention is to provide a speech recognition control method that eliminates the drawbacks of the above-mentioned conventional techniques and can obtain a high recognition rate, especially for input speech from beginners. [Summary of the Invention] In the voice recognition control system according to the present invention, predetermined types with different contents are set as input reminder messages, and in the case of an information response service, a telephone number corresponding to the desired input reminder message type is set. By inducing incoming service requests by
Alternatively, in the case of an information notification service, by selecting and storing each desired input reminder message type according to each notifier who should perform the information notification service, the required input reminder message type can be selected and stored according to the content of the desired input reminder message type. This controls the voice input. [Embodiments of the Invention] Hereinafter, embodiments of the present invention will be described based on the drawings. FIG. 1 is a block diagram of an embodiment of a speech recognition apparatus using a speech recognition control method according to the present invention, and FIG. 2 is a flowchart thereof. Here, 1A and 1B are telephone lines related to the voice input from the caller (user) and the synthesized voice output from the recognition device, which have different phone numbers.
2 is an input unit that performs gain adjustment, band limitation, and other necessary preprocessing on the input audio signal, and then converts the input audio signal into digital data, and also performs the necessary functions for incoming and outgoing calls on the telephone lines 1A and 1B;
3 is an analysis unit that performs audio analysis of the input audio based on the input digital audio signal and extracts characteristic data thereof; 4 is a detector that detects audio sections of the input audio according to a predetermined threshold for detecting audio sections; 5 is a speech recognition section that performs pattern matching processing (similarity calculation processing) between the input speech and each standard speech pattern; 6 is a speech recognition section that performs pattern matching processing (similarity calculation processing) between the input speech and each standard speech pattern; 6 is a speech recognition section that performs pattern matching processing (similarity calculation processing) between the input speech and each standard speech pattern; 7 is a standard speech pattern memory that stores (memorizes) multiple sets of standard speech pattern data for each word to be recognized; 8 is a standard speech pattern that controls the selection; A selection section 9 is a speech synthesis section 10 for displaying/confirming input voice analysis results/recognition results, voice input instructions, and other necessary display/instructions.
There are two types, one for beginners and one for experts, depending on the speaker.
11 is a notifier memory for the user; 12 is a control unit for controlling the above-mentioned units and other necessary processing; 13;
is a host device that performs desired service processing based on voice recognition results. First, in the service system where this speech recognition device is used, the information inquiry (response) is performed.
The control and operation of the service will be explained based on FIG. 1 and FIG. 2a. First, two types of input reminder messages are prepared, for example, for beginners and for experts, and it is assumed that these correspond to telephone lines 1A and 1B, respectively. In addition, the above-mentioned types may be provided with a larger number of contents depending on the practical case. For example, when a user attempts to receive an information response service, he/she may call the telephone line 1 depending on whether he/she wishes to receive an input reminder message for beginners or experts regarding input instructions for the content request code (request code).
Dial the phone number corresponding to A or 1B. As a result, the user's telephone is connected to the telephone line 1A or 1B, and the input unit 2, after detecting which of them receives the call, transmits the information to the control unit 12 (see FIG. 2a). Processing 20A). The control unit 12 determines the type of subsequent input reminder message (either one for beginners or an input reminder message for experts) based on the information (processing 21).
A) Instructs the input section 2, analysis section 3, and speech section detection section 4 to prepare for speech input. In addition, all sets of standard voice patterns for the classification of words to be recognized at that time (for example, classifications of numbers, service type names, product names, place names, etc.) are stored in the standard voice pattern memory 7.
The standard voice pattern selection unit 8 is instructed to select from among the following (process 22). When these preparations are completed, an input reminder message (either for beginners or experts) to prompt the speaker to input voice is sent from the telephone line 1A or 1B via the voice synthesis unit 9 (same process 23). . This allows the speaker to connect to telephone line 1A or 1B.
When inputting audio from (same process 24), input section 2
performs digital conversion, etc., and the analysis unit 3 performs audio analysis on the digital audio signal and extracts the characteristic data, etc. (processing 25). The speech recognition section 5 performs pattern matching processing between the feature data of the input speech and the selected standard speech pattern data, and transmits the degree of similarity of each standard speech pattern to the input speech to the determination section 6 (same as Processing 26). The determination unit 6 transmits the one with the highest degree of similarity (most likely) to the control unit 12 as a recognition result (processing 26). In the case of a reject whose most probable similarity value to the input voice is low and it is doubtful to output it as a recognition result, the control unit 12
Instructs the standard voice pattern selection unit 8 to select the same pattern as before (same process 30), and also connects the telephone line 1A via the voice synthesis unit 9.
Alternatively, a re-input reminder message is sent from 1B (same process 31). In addition, if it is not a reject, the control unit 12
The recognition request message is sent from the telephone line 1A or 1B via the speech synthesis section 9 as a display for the speaker to confirm whether the recognition result is correct or not (process 28). The speaker listens to this to know whether his or her input voice was recognized correctly or incorrectly, and inputs the confirmation result through the telephone line 1A or 1B (process 29). If the recognition candidate is correct based on the confirmation information, the control unit 12 sends it as a recognition result to the host device 13, ends the processing for one input voice, and prepares for the next input. On the other hand, if you receive confirmation information that it was misrecognized, process 3 as in the case of reject above.
0 and 31, and this is repeated until correct recognition is obtained. When correct recognition is obtained, the recognition result is sent to the host device 13 in the same manner as described above, and the series of processing ends. Here, examples of the contents of the input reminder messages for beginners and experts are shown in the table below.

〔Effect of the invention〕

As described in detail above, according to the present invention, it is possible to improve the overall recognition rate of the speech recognition device by ensuring voice input, especially from beginners. Significant effects can be achieved in improving system reliability and serviceability, and in turn, increasing adoption.

[Brief explanation of drawings]

FIG. 1 is a block diagram of an embodiment of a speech recognition apparatus using a speech recognition control method according to the present invention, and FIG. 2 is a flowchart thereof. 1...Telephone line, 2...Input section, 3...Analysis section, 4...Speech section detection section, 5...Speech recognition section,
6... Judgment unit, 7... Standard voice pattern memory, 8
...Standard speech pattern selection section, 9...Speech synthesis section,
10...Message selection unit, 11...Notifier memory, 12...Control unit, 13...Host device.

Claims (1)

[Claims] 1. In a voice response service system that includes a voice recognition device that sends an input reminder message through a telephone line and recognizes the input voice input sequentially in accordance with the message, the speakers using the service system are at least beginners and experienced speakers. A plurality of telephone lines are set up corresponding to each of the above-mentioned predetermined types for use according to which of the predetermined types they belong to, including persons, and a prompt for the input of the speaker corresponding to each telephone line is provided. means for storing an input reminder message for a telephone line, and a means for reading out and transmitting the stored input reminder message corresponding to one of the telephone lines to a speaker making a call through one of the telephone lines; 1. A voice recognition control method, comprising means for controlling a user to input a desired voice.