JPH10301595A - Voice recognition and response device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To output a voice response at easy-to-hear level which is not affected by the line quality of a telephone line by detecting the extent of the line quality from a received voice and adjusting the response level of the voice according to the detected extent of the line quality. SOLUTION: The telephone voice recognition and response device 8 is connected to a telephone set 1 through the telephone line 2. The voice from the telephone set 1 is inputted to a line quality detection part 3 and a voice recognition part 6 and voice recognition is performed here. The recognized voice is processed by a control part 4, whose process result is synthesized by a voice response part 5 into a voice signal, which is outputted as an answer voice to the telephone line 1. At this time, a line quality detection part 3 detects the line quality from the input voice. According to the detected line quality, an answer level adjustment part 7 adjusts the answer level of the output of the voice response part 5. Therefore, the response voice is heard at the response level which is not affected by the quality of the telephone line 2 and a man- machine is made smooth.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a voice recognition response device, and more particularly to a voice recognition response device that changes a voice response level in accordance with line quality.

[0002]

2. Description of the Related Art Conventionally, various types of guidance systems which recognize a voice input via a telephone line and provide a response according to the voice have been generalized as a voice recognition response device. In such a system, the response level is fixedly set at the time of manufacturing the device or semi-fixed at the time of system construction, assuming an average line quality. Japanese Patent Application Laid-Open (JP-A) No. 63-301 describes an example of a speech recognition response device that performs high-level control.
No. 998 is disclosed. This speech recognition response device will be described with reference to the block diagram of FIG.
The telephone set 1 is connected to the voice recognition response device 9 via the telephone line 2, the voice recognition response device 9 calculates an S / N ratio of the signal level of the input signal and the S / N ratio of the noise level,
The control unit 11 determines the ratio between the signal level and the noise level of the response voice based on the result of the S / N calculation unit 10, a voice response unit 12, and a voice recognition unit 13.

Next, the operation of the conventional speech recognition response device 9 will be described. The uttered voice is input from the telephone 1 to the voice recognition response device 9 via the telephone line 2. The S / N calculator 10 in the speech recognition responder 9 calculates the S / N ratio (hereinafter, referred to as ISN) between the signal level of the input signal and the noise level, and the speech recognition unit 13 performs speech recognition processing. Next, the control unit 11 selects a word that prompts the speaker to make the next utterance, and outputs this to the voice response unit 12. Control unit 1
1 is the S / N of the response voice signal of the voice response unit 12 at the same time.
The ratio (hereinafter referred to as OSN) is calculated by the IS from the S / N calculator 10.
N is determined according to N, and this is output to the voice response unit 12.
For example, OSN = α · ISN (where α is a constant). The voice response unit 12 is configured to hold the OSN determined by the control unit 11 on the telephone line 2 and output a response voice to notify the speaker of the wording. Therefore, the S / N ratio of the response voice heard by the speaker at this time is proportional to the S / N ratio of the voice previously uttered by oneself. Therefore, the speaker who hears the response voice, for example, if the S / N ratio of the response voice is low, judges that there is ambient noise or the voice is low, and naturally the next speaker utters a loud voice. , Or ambient noise.

Another conventional speech recognition response device is disclosed in Japanese Patent Publication No. 3-52638. Referring to the block diagram of FIG. 3, this conventional voice recognition response device will be described. The telephone 1 is connected to a voice recognition response device 16 via an exchange 14 and a network control circuit 15. The voice recognition and response device 16 includes a voice recognition unit 18 that includes a voice level detection unit 17 that detects a voice level of an input voice, a voice response unit 20 that includes a response level adjustment unit 19 that adjusts an output level of a response voice, The information processing apparatus 21 controls these units and the network control circuit 15. Note that the audio level detection unit 17 and the response level adjustment unit 1
9, a level adjustment loop 22 for transferring information for level adjustment is provided.

Next, the operation of the conventional voice recognition response device 16 will be described. The uttered voice is input from the telephone 1 to the voice recognition responder 16 via the exchange 14 and the network control circuit 15. The level of the uttered voice is detected by the voice level detection unit 17 in the voice recognition unit 18,
According to the detected voice level, the voice level adjusting unit 19 in the voice response device 20 is controlled through the level adjusting loop 22. The content of the control at this time is such that when the detected voice level is low, the response voice level is low, and when the detected voice level is high, the response voice level is high.

[0006]

All of the above-mentioned conventional voice recognition and response devices have some problems. For example, if the level shown first is fixed, the response sound may be loud depending on the state of the line. There is a problem that the level of the response sound is different each time the sound is used, and the sound level is small. In particular, when the line quality is poor, the response sound becomes difficult to hear, and thus there is a problem that the usability of the user deteriorates. In the case of calculating the S / N ratio between the signal level of the input signal and the noise level, and the method of detecting the voice level of the input voice, use the human psychological action when the line condition is poor. Therefore, in order to secure a large input level, the response level is reduced. For this reason, if the device user recognizes the voice recognition response device as a machine, there is no guarantee that it will make a loud voice because it is difficult to hear, and it may be more difficult to hear than a device with a fixed level. There is a problem that there is also.

[0007] In particular, unless the voice of the device user is too low to cause any trouble on the voice recognition / response device side, the voice of the voice recognition / response device is not affected by the voice of the device user. Even if output at a level considered appropriate from the beginning, there is no problem, and it is judged that the serviceability is higher.

An object of the present invention is to detect a line quality from a voice input via a telephone line and adjust a response level, thereby outputting a voice response at an easy-to-listen level which is not affected by the line quality to a device user. It is an object of the present invention to provide a telephone voice recognition response device which enables the user to perform a voice recognition.

[0009]

According to the present invention, there is provided a voice recognition and response apparatus connected to a telephone via a telephone line and configured to recognize a voice to be received, and a response to the voice recognized by the voice recognition section. A voice recognition response device comprising: a voice response unit that creates and outputs the telephone line to the telephone; a line quality detection unit that detects a degree of line quality of the telephone line from the received voice; and a line that the line quality detection unit detects. A response level adjusting unit that adjusts a response level of a voice output from the voice responding unit according to a degree of quality.

[0010] In the voice recognition response device of the present invention, the line quality detecting unit may detect a line quality of the telephone line from the received voice level and a signal-to-noise ratio.

In the voice recognition response apparatus according to the present invention, the response level adjusting unit may adjust the response level of the voice if the degree of the line quality of the telephone line is lower than a standard value, and lower the level if the level is good.

[0012] In the voice recognition response apparatus according to the present invention, the response level adjusting unit raises the voice response level if the line quality of the telephone line is lower than a standard value, and maintains the standard value if the line quality is good. May be.

[Operation] The line quality is detected from the voice input via the telephone line, and the response level is adjusted to a level that is easy for the user to hear from the detection result. For this reason,
There is no need to set response levels.

[0014]

Next, embodiments of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing an embodiment of the present invention.

The telephone 1 is connected to a telephone voice recognition / response device 8 via a telephone line 2. The telephone voice recognition device 8 includes a line quality detection unit 3 that detects the line quality of voice input from the telephone 1 and a voice recognition unit 6 that recognizes input voice.
The control unit 4 processes the recognition result and outputs a response signal to the voice response unit 5, the voice response unit 5 outputs the response signal from the control unit 4 as a voice signal, and the line quality detection unit 3 detects the response signal. A response level adjusting unit 7 reduces the response level when the line quality is good according to the line quality, and increases the response level when the line quality is bad.

Next, the operation will be described.

The telephone voice recognition and response device 8 is connected to the telephone 1 via the telephone line 2. The voice from the telephone 1 is input to the line quality detecting unit 3 and the voice recognizing unit 6, where the voice is recognized. The recognized voice is processed by the control unit 4, and the processing result is synthesized into a voice signal by the voice response unit 5 and output to the telephone 1 as a response voice. In the apparatus according to the present invention, the line quality detecting unit 3 detects the line quality from the input voice. The response level adjustment unit 7 adjusts the response level of the output of the voice response unit 5 according to the detected line quality.

The above-mentioned line quality detector 3 uses the level of the input voice and the S / N ratio between the input level and the noise level as parameters for detecting the line quality. For example, LC (line quality) = α (constant), IL (input audio level), SN
(S / N ratio).

The response level adjusting unit 7 adjusts the response level so that the response voice level is increased when the line quality is poor, and the response level is decreased when the line quality is good. For example, OL ′ (response level) = α (constant) · OL (response level before adjustment) / LC (line quality).

In the above-described embodiment, the case where the line quality is detected based on the input voice level and the S / N ratio has been described. However, depending on the use situation, either one of them may be used. In addition, if it is not necessary to reduce the response level when the line quality is good in adjusting the response level, the adjustment for reducing the response level may be omitted.

[0022]

As described above, according to the present invention, a user of a device can hear a response sound at a response level that is not affected by the quality of a telephone line, and can achieve a smooth man-machine interface. It has an effect.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention.

FIG. 2 is a block diagram of a conventional voice recognition response device.

FIG. 3 is a block diagram of another conventional voice recognition response device.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Telephone 2 Telephone line 3 Line quality detection part 4 Control part 5 Voice response part 6 Voice recognition part 7 Response level adjustment part 8 Telephone voice recognition response device

 ──────────────────────────────────────────────────の Continuing from the front page (72) Inventor Satoshi Tsukada Inside NEC Corporation 5-7-1 Shiba, Minato-ku, Tokyo

Claims (4)

[Claims] A voice recognition unit that connects to a telephone via a telephone line and recognizes voice to be received; and a voice response unit that generates a response to the voice recognized by the voice recognition unit and outputs the response to the telephone. In the voice recognition response device, a line quality detection unit that detects the level of the line quality of the telephone line from the received voice, and a voice output from the voice response unit based on the level of the line quality detected by the line quality detection unit. A speech recognition response device, comprising: a response level adjustment unit that adjusts a response level. 2. The speech recognition response device according to claim 1, wherein said line quality detection unit detects a degree of line quality of said telephone line from a level of said received voice and a signal noise ratio. 3. The voice according to claim 1, wherein the response level adjusting unit adjusts the response level of the voice if the level of the line quality of the telephone line is lower than a standard value, and lowers the response level if the level is good. Recognition response device. 4. The system according to claim 1, wherein the response level adjusting unit increases the response level of the voice if the line quality of the telephone line is lower than a standard value, and adjusts the response level to the standard value if the level is good. Item 1. The voice recognition response device according to Item 1.