JPH0527790A - Voice input/output device - Google Patents

Abstract

PURPOSE:To decide the type of a user from the voice and change an answer message matching the type of the user. CONSTITUTION:This voice input/output device is provided with a frequency analysis device 2 which decides the sex of the user, a voice recognition device 3 which decides how much the user is in a hurry, an answer time measuring instrument 4 which decides how much the user is familiar with operation, a total decision device 5 which judges the type of the user by comparing data indicating the tendency of past users stored in a reference data memory 6 with the decision result data, and a voice synthesizer 8 which composes an answer message corresponding to the specified type of the user by using voice data stored in a voice data memory 7 and outputs the message.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a voice input / output device for operating a target device using voice.

[0002]

2. Description of the Related Art In recent years, information processing terminal devices such as cash deposit / withdrawal devices not only display characters on a display device but also display figures or operate by voice in order to guide the user's operation. Instructions are becoming multimedia. In particular, voice-based operation guidance is very effective for allowing a user who is unfamiliar with the apparatus to operate the machine without fail. Further, the voice recognition technology is capable of recognizing a simple response with a considerably high accuracy. Therefore, it is possible to recognize an information processing terminal device that can operate by voice while interacting with the information processing terminal device by voice. Expectations are rising.

By the way, in an information processing terminal device which advances an operation while responding to a conventional device, one kind of response message is output from the device side for one operation. .. And the content was often detailed, slow and polite, and rather redundant, to accommodate all types of users.

[0004]

As described above, the conventional uniform and redundant response message is unnecessary for a user who is accustomed to the operation and wants to perform a quick operation. On the contrary, there was a problem of increasing the burden. Further, for users who are unfamiliar with the operation, there is a problem that the explanation is insufficient, which causes a problem that the user may get confused or make an erroneous operation.

The present invention has been made to solve such a problem, and determines the type of the user from the voice of the user, and changes the response message to the optimum one that matches the type of the user. However, it is an object of the present invention to provide a voice input / output device that performs a fine-tuned response to the user.

[0006]

In order to achieve this object, the present invention provides a voice input / output device for operating an information processing device by voice while responding to a user.
In order to determine the gender of the user, a means for measuring the frequency of the input voice of the user, calculating the average value thereof and comparing it with the frequency used as the reference for the determination, and for determining the urgency of the user In addition, the method of measuring the speed of words from the time interval between words, the time required for response and the uniformity of the time interval between words to determine the user's familiarity with the operation, and the type of word Means for determining, a means for determining the type of user, means for pre-storing data indicating past user tendency analyzed from voice, and data for each determination result and reference data And a plurality of response messages for one operation corresponding to the user type to be determined, and a user identified from among them. To the type of In which and a means for outputting a response message response.

[0007]

According to the present invention having the above-mentioned configuration, the user is asked for a response by, for example, asking an appropriate question to the user, and the response is analyzed so that the type of the user can be determined at the same time as the purpose of the past. Estimated statistically in comparison with the trend. The result optimizes the content and speed of the response message.

That is, the sex can be determined from the average frequency of the voice of the user. For example, if the average frequency is higher than 400 hertz, the probability of being a female is high, and if it is 400 hertz or less, the probability of being a male is high. In addition, it is possible to determine whether the user is in a hurry based on the speed of words. For example, when recognizing the sentence "I want to withdraw cash", the time between "cash" and "want to withdraw" can be changed. If there is almost no such interval, you can judge that you are in a hurry.
On the other hand, if the interval is 0.5 seconds or more, it can be determined that the user is not in a hurry.

Furthermore, it is possible to determine fluency from the length of time required for a response, the time interval between words and the type of words used, and determine whether the user is accustomed to using the device. If the person has a time of 1 second or more before issuing a response, and if there is an interjection such as "Eto", this customer can be momentarily wondered how to respond. On the other hand, if an accurate response is immediately returned, it can be determined that this customer is familiar with how to respond. An optimum response message can be issued by classifying the user into several types based on these data and determining which type the user belongs to.

[0010]

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is a control block diagram of a voice input device showing an embodiment of the present invention. In the following, in the present embodiment, the one incorporated in a cash payment machine will be described as an example. In the figure, reference numeral 1 denotes an infrared detection device, which detects whether or not there is a person in front of a cash dispenser. 2 is a frequency analysis device, a microphone M
The frequency of the user's voice input from is measured, the average frequency thereof is calculated, and the average frequency is compared and used as a criterion for determining the sex of the user. For example, 400
If the average frequency is higher than hertz, it can be determined that the probability of being a female is high, and if it is 400 hertz or less, it is likely to be a male.

Reference numeral 3 denotes a voice recognition device, which recognizes the contents of the response and judges whether the speed of the word is "quick talk", "normal" or "slow" from the time interval between words, and Will be used as a criterion for deciding whether or not you are in a hurry. For example, when recognizing the sentence “I want to withdraw cash”, if there is almost no time interval between “I want to withdraw cash” and “I want to withdraw cash”, this user can be judged as “in a hurry”. On the contrary, if the interval is 0.5 seconds or more, this user can be judged as [not in a hurry].

Reference numeral 4 denotes a response time measuring device, which determines whether the user's response is "long" or "short" for the response, whether the time intervals between the words appearing in the response are uniform, and the type of the word. Whether the user is accustomed to using the device is determined by determining whether the user is “fluent”, “normal”, or “severe”. For example, if it takes a second or more for the user to make a response, and if there is an interjection such as “um”, this user is not accustomed to using the device and how to respond. It can be judged that he was lost for a moment. On the other hand, if an accurate response is immediately returned, it can be determined that this user is already using the device and is well aware of how to respond.

Reference numeral 5 is a comprehensive judgment device for judging the type of the user based on the judgment results by the frequency analysis device 2, the voice recognition device 3 and the response time measurement device 4, and 6 is the comprehensive judgment device 5 for the user. It will be the standard when judging the type,
This is a reference data memory in which data indicating past user trends analyzed from voice is recorded in advance.

Reference numeral 7 is a voice data memory, 8 is a voice synthesizer for synthesizing a predetermined response message from the voice data stored in the voice data memory 7, and outputting it via a speaker SP at a predetermined speed. The voice data memory 7 stores voice data so that a plurality of types of response messages can be output for one operation in order to output a response message corresponding to the type of user judged by the comprehensive judgment device 5. Has been done. Here, as a tendency of the past user analyzed from the voice stored in the reference data memory 6, for example, a quick and fluent male user has used the same device many times and operated. The data is stored that many people are accustomed to, and are in a hurry, and a quick response is desirable.

As described above, a user who has used the device in the past and is accustomed to operating the device and who is in a hurry is of type A, and responds to the user of type A with a simple response. The voice data for outputting the response message is stored in the voice data memory 7. Also,
Users who have used it in the past but are not in a hurry are of type B, and the voice data for outputting a response message that makes a normal response corresponds to this type B user. It is stored in the data memory 7. Furthermore, a user who is operating the device for the first time or is in a state close to the device is of type C, and the voice data for outputting a response message for giving a more detailed response to the user of this type C is voice data. It is stored in the memory 7.

As described above, FIG. 2 is a type determination table serving as a reference for the comprehensive determination device 5 to determine the type of user. First, of the users who are determined to have a high probability of being male by the frequency analysis device 2, the voice recognition device 3 determines that the user of type A is a quick mouth, and the response time measurement device. This is the case when it is determined in 4 that the user is responding fluently or normally.

Among the users who are determined to have a high probability of being male by the frequency analysis device 2, the type B user is determined to be normal or slow talking by the voice recognition device 3. Further, there is a case where it is determined that the response time measuring device 4 is performing fluent or normal response. Further, among the users who are determined to have a high probability of being male by the frequency analysis device 2, it is determined that the user is a type C user, and it is determined that the voice recognition device 3 is speaking quickly, normally or slowly, Further, it is a case where it is determined that the response time measuring device 4 is responding in a stuttering manner.

Among the users who are determined by the frequency analysis device 2 to have a high probability of being a woman, the type A user is determined to be fast by the voice recognition device 3, and the response time is measured. This is a case where it is determined that the device 4 is responding fluently. Further, among the users who are determined to have a high probability of being female by the frequency analysis device 2, it is determined by the voice recognition device 3 that the user of type B is quick.
When the response time measuring device 4 determines that the response is normal, the voice recognition device 3 determines that the response is normal, and when the response time measuring device 4 determines that the response time is fluent or normal, Further, there is a case where the voice recognition device 3 determines that the response is slow and the response time measurement device 4 determines that the response is fluent.

Further, among the users who are determined to have a high probability of being female by the frequency analysis device 2, it is determined by the voice recognition device 3 that the user of type C is quick.
When it is determined that the response time measuring device 4 is responding in a stuttering manner, it is determined that the voice recognition device 3 is normal, and when the response time measuring device 4 is determined to respond in a stuttering manner, Further, it is a case where the voice recognition device 3 determines that the response is slow, and the response time measurement device 4 determines that the response time measurement device 4 responds normally or in a stuttering manner.

FIG. 3 is an explanatory diagram showing the contents of the guidance voice in this embodiment. FIG. 3A shows the case where the user is the A type.
(B) shows the case where the user is a C type. In the case of the type A user, since it is determined that he is familiar with the operation method, the response message that merely prompts the operation is output as a whole. In addition, since the type C user determines that the device is being operated for the first time or is in a state close to it, what is used as an input means for various information, where is the device, etc. A response message that prompts the user to operate is output.

FIG. 4 is a flow chart of this embodiment, and the operation of this embodiment will be described below. First, the user stands in front of the cash dispenser. Then, the device recognizes that the user has appeared by the infrared detection device 1, prompts the user to perform an operation, and determines the type of the user.
The response message as in (1) is output to start the response with the user (S1, S2). At this time, it is assumed that the user's response is a slightly swollen male voice, and that the response is accurate and fluent as shown in FIG. When a voice is input from the user (S3), the voice recognition device 3 is activated in the device.
In addition to recognition of the voice input in step S1, the frequency analysis device 2, the voice recognition device 3, and the response time measurement device 4 determine the voice in order to determine the type of user. The frequency analysis device 2 indicates that the gender of the user is likely to be male (S4), and the voice recognition device 3 recognizes that the user is a little quicker (S4).
5) The response time measuring device 4 indicates that the response is fluent (S6).

On the other hand, as described above, the data showing the tendency of the past user analyzed from the voice is the reference data memory 6
The tendency is that, for example, a male user who is fast-paced and fluent is a user who has used the same processing device many times and is familiar with the operation, and is often in a hurry. Data is stored that it is desirable to take appropriate measures. At this time, the frequency analysis device 2 and the voice recognition device 3
And each determination result data of the user who is currently supported by the response time measuring device 4, and the reference data memory 6
Comparing with the reference data which is analyzed the tendency of the user in the past, the comprehensive judgment device 5 is accustomed to the operation because this user type has used this device in the past. It is determined that the user is a type A user who is present and is in a hurry (S7).

As a result, the voice synthesizer 8 uses the voice data stored in the voice data memory 7 as shown in FIG.
A response message that makes a simple response such as (7) to (7) is synthesized and output from the speaker SP at a speed slightly faster than usual (S8). Even with a simple response message as shown in FIGS. 3 (3) to (7), since the user is accustomed to the operation, he / she can quickly respond to the content and speed of the response message to promptly cash out. You can finish the process of withdrawing. On the other hand, it is assumed that the response to the response message in FIG. 3 (1) is an irregular female voice as shown in (8), which cannot be said to be accurate. At this time, as in the case described above, based on the determination results of the frequency analysis device 2, the voice identification device 3, and the response time measurement device 4, the comprehensive determination device 5 determines whether this user is using this device for the first time or It is determined that the user is a C type user in a close state.

As a result, the voice synthesizer 8 uses the voice data stored in the voice data memory 7 as shown in FIG.
A response message for making a more detailed response like (13) to (13) is synthesized and output from the speaker SP at a speed slightly slower than usual (S10). Thus, FIG.
By outputting a more detailed response message as shown in (9) to (13) at a slightly slower speed, even a user who is unfamiliar with the operation can be guided so as to surely achieve the purpose. You can

[0025]

As described above, according to the present invention, in the voice input / output device that operates the information processing device by voice while responding to the user, the input is made in order to determine the sex of the user. To measure the frequency of the user's voice, calculate the average value and compare it with the reference frequency, and to determine the user's urgency, measure the speed of words from the time interval between words. The time it takes to respond in order to determine how to measure and how familiar the user is with the operation,
In addition to measuring the uniformity of the time interval between words, it also stores in advance the data indicating the past user tendency analyzed from the voice, which serves as a reference for determining the type of word and the type of user. Means for comparing the determination result data with the reference data to determine the user type, and one operation corresponding to the determined user type. And a means for outputting a response message corresponding to the type of user specified from among the plurality of response messages.

Accordingly, in the information processing device operated by using the voice, the voice input / output device determines from the response by the voice whether the user is proficient in the operation, urgent or not, and the user It is possible to output different operation guidance voices depending on the type. Therefore, it is less likely that a user unfamiliar with the operation will be confused about the operation method or a user skilled in the operation will have a confused feeling due to the uniform guide voice from the device side. Can be expected. Further, when the present invention is applied to a device intended for many unspecified users such as a cash dispenser, it is possible to save the time of issuing unnecessary redundant messages depending on the operator, and it is efficient as a whole. Operation becomes possible, and as a result, the waiting time of the user can be shortened.

[Brief description of drawings]

FIG. 1 is a control block diagram of a voice input / output device showing an embodiment of the present invention.

FIG. 2 is a type determination table for determining the type of user.

FIG. 3 is an explanatory diagram showing an output example of a response message for each type.

FIG. 4 is a flowchart showing the operation of the present embodiment.

[Explanation of symbols]

 2 Frequency analysis device 3 Speech recognition device 4 Response time measurement device 5 Comprehensive judgment device 6 Reference data memory 7 Speech data memory 8 Speech synthesis device

Claims (1)

Claim: What is claimed is: 1. A voice input / output device for operating an information processing device by voice while responding to a user, in order to determine the gender of the user, an input voice of the user. And a means for calculating the average value and comparing it with the reference frequency, and a means for measuring the speed of words from the time interval between words in order to determine the urgency of the user. , To measure the user's familiarity with the operation, measure the time required for response, the uniformity of the time interval between words, and the means for determining the type of word, and the standard for determining the type of user. , Means for pre-storing data indicating past user tendency analyzed from voice, means for judging the type of user by comparing the data of each judgment result with reference data, Judgment Correspond to the type of user that is, 1
A voice input / output device comprising a plurality of response messages for one operation, and a means for outputting a response message corresponding to the type of user specified from among them.