JP2005234331A - Voice interaction device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a voice interaction device which is improved in convenience to a user by accurately estimating the user intelligibility and performing proper interaction control. <P>SOLUTION: An intelligibility calculation part 28 finds an intelligibility estimation parameter s1 calculated from the keyword appearance time from when voice input is requested to when a keyword is vocalized and an intelligibility estimation parameter s2 calculated from the rate of the number of phonemes of an important keyword to the total number of phonemes from the start to the end of user's utterance, and the intelligibility R of interaction by the user is calculated from the intelligibility estimation parameters s1 and s2. Then an interaction control part 29 performs interaction control based upon the intelligibility R, and decides the intelligibility by the user is low when the intelligibility R is 0 and outputs a message urging re-input. When the intelligibility R is 1, on the other hand, it is decided that the intelligibility is normal, input contents are confirmed, and an advance to a next step is made. When the intelligibility R is 2, it is decided that the intelligibility is high and an advance to the next state is made without confirming the input contents. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a speech dialogue apparatus used in a processing system that executes processing based on dialogue with a user.

  2. Description of the Related Art Conventionally, a voice dialogue apparatus used for dialogue with a user requires input request means for outputting a signal for requesting voice input, recognition means for recognizing input voice, and voice input, for example. A measuring means for measuring the time from when the voice input is detected and the duration of the voice input (speech time), and an output means for outputting a voice response signal corresponding to the voice recognition result, The time from when voice input is detected until the voice response signal is output, the response time of the voice response signal, or the expression format of the voice response signal is detected after the voice input is requested. There are those that variably control based on the time until the input is made and the duration of the voice input. With this device, it is possible to give an appropriate response to each user based on the reaction time of each user and the voice input time (see, for example, Patent Document 1).

On the other hand, the device that recognizes and responds to the user's voice in the same manner, the time from when the voice input is requested until the voice input is detected, the duration of the voice input (speech time), Some measure the number of utterances of the user to estimate the user's level of understanding (skill level), and based on the result, control the voice guidance scenario and utterance content by the response voice signal, as well as the utterance speed. This device can similarly give an appropriate response to each user (see, for example, Patent Document 2).
Japanese Patent Publication No. 5-18118 JP 2000-194386 A

  By the way, in the conventional apparatus, although the output timing of the voice response signal on the apparatus side can be changed by using the time until the user starts speaking, whether the user has spoken or not is inputted. There is a problem that it may not be able to respond correctly to the user's utterance because the content is not determined only by determination based on the frequency and strength of the signal. This is because even if the user responds immediately to the voice input request, the user has spoken a meaningful word such as the destination, or a preface to a meaningful word in Japanese, for example It is impossible to judge whether the user really understands the dialogue or speaks unless the words are uttered as “A ~”, “E ~”, etc. Because. That is, even if the user responds immediately to a meaningless word, it cannot be said that the user understands the dialogue, so even if a voice response signal is output based on this time, the user There was a problem that an appropriate response could not be given.

  In addition, when the number of input utterances and utterances is measured and compared with the number of utterances and utterances of standard users, the user's understanding (skill level) is estimated. There was a problem that it was difficult to deal with various utterances. Specifically, even words that have the same level of meaning as the destination, such as `` Kashiwa '' and `` Tokyo Amyuzu Mento Land '', compared with the standard number of utterances and utterances, There is a possibility that different understanding levels may be output due to the difference in the number of sounds or the number of words, or even if the level of understanding is the same, if you add words such as “Is” after the words and speak carefully, the level of understanding will be low There is a problem that even if a voice response signal is output based on the estimated degree of understanding, an appropriate response may not be given to the user.

  The present invention has been made in view of the above problems, and an object of the present invention is to provide a voice interaction device that accurately estimates a user's understanding and performs appropriate dialogue control to improve user convenience. And

  In order to solve the above-described problem, a voice interactive apparatus according to the invention of claim 1 includes a voice input means (for example, a microphone 1 of an embodiment described later) for inputting voice uttered by a user, and input voice. Voice recognition means (for example, a voice recognition section 22 in an embodiment described later) and keyword determination means (for example, a keyword determination section in an embodiment described later) for extracting a predetermined keyword from the recognized voice of the user. 23) and input sound number measuring means for measuring the number of sounds of the recognized user's voice (for example, a recognition word counting unit 26 and an understanding level calculating unit 28 of the embodiment described later, steps S11 to S12). And the degree of understanding of the user in the dialogue based on the ratio of the number of sounds of the keyword to the total number of sounds input by the user's utterance Degree control means (for example, the processes of steps S13 to S15 and the process of step S23 executed by the understanding degree calculation unit 28 of the embodiment described later) and the dialogue control for controlling the dialogue response according to the understanding degree of the user And a means (for example, a dialogue control unit 29 in an embodiment to be described later).

  The speech dialogue apparatus having the above-described configuration is a speech dialogue apparatus that includes voice input means for inputting voice uttered by a user, and voice recognition means for performing recognition processing of the input voice. The voice input by the user's utterance is such that the means is higher when the ratio of the number of meaningful words to the total number of uttered voices is higher. Since the user's level of understanding in dialogue is determined based on the ratio of the number of sounds of important keywords to the total number of tones, the dialogue control means must execute appropriate dialogue control according to the estimated level of understanding. Can do.

  According to a second aspect of the present invention, there is provided a voice interactive apparatus comprising: voice input means for inputting voice spoken by a user (for example, a microphone 1 in an embodiment described later); and voice recognition means for performing input voice recognition processing. (For example, a voice recognition unit 22 in an embodiment described later), keyword determination means (for example, a keyword determination unit 23 in an embodiment described later) for extracting a predetermined keyword from the recognized voice of the user, and the user Keyword appearance time measuring means for measuring the keyword appearance time from when the user requests utterance until the user utters the keyword (for example, executed by the time keyword combining unit 27 and the comprehension degree calculating unit 28 in the embodiment described later). Comprehension level determination for determining the level of understanding of the user in the dialogue based on the processing from step S1 to step S3) and the length of the keyword appearance time Dialog control means (for example, processing in steps S4 to S6 and processing in step S23 executed by the understanding level calculation unit 28 of an embodiment described later) and a dialog control means for controlling a dialog response according to the user's understanding level ( For example, a dialogue control unit 29) of an embodiment described later is provided.

  The speech dialogue apparatus having the above-described configuration is a speech dialogue apparatus that includes voice input means for inputting voice uttered by a user, and voice recognition means for performing recognition processing of the input voice. For example, the key word is important after requesting voice input, such that the meaning of words that have no meaning is short and the meaning of words that are spoken earlier is higher when the user speaks. Since the degree of understanding of the user in the dialogue is determined based on the length of the keyword appearance time until the utterance is spoken, the dialogue control means can execute appropriate dialogue control according to the estimated degree of understanding. .

  According to a third aspect of the present invention, there is provided a voice interaction apparatus comprising: voice input means for inputting voice spoken by a user (for example, a microphone 1 in an embodiment described later); and voice recognition means for performing processing for recognizing input voice. (For example, a voice recognition unit 22 in an embodiment described later), keyword determination means (for example, a keyword determination unit 23 in an embodiment described later) for extracting a predetermined keyword from the recognized voice of the user, and the user A total utterance time measuring means for measuring the total utterance time from the start to the end of the utterance (for example, the processing of step S31 executed by the time keyword combining unit 27 and the understanding degree calculating unit 28 in the embodiment described later); Keyword utterance duration measuring means for measuring the keyword utterance duration required for the user to utter the keyword itself (for example, a time keyword of an embodiment described later) A process of step S32 executed by the combination unit 27 and the understanding level calculation unit 28) and an understanding for determining the level of understanding of the user in the dialogue based on the ratio of the keyword utterance duration to the total utterance time Degree control means (for example, the processing from step S33 to step S35 and the processing of step S23 executed by the understanding level calculation unit 28 of the embodiment described later) and dialogue control for controlling the dialogue response according to the understanding level of the user And a means (for example, a dialogue control unit 29 in an embodiment to be described later).

  The speech dialogue apparatus having the above-described configuration is a speech dialogue apparatus that includes voice input means for inputting voice uttered by a user, and voice recognition means for performing recognition processing of the input voice. For example, if the percentage of the time during which a meaningful word is spoken is large relative to the total utterance time from when the user starts utterance to the end, the degree of understanding in the user's dialogue is higher In addition, based on the ratio of the keyword utterance duration during which important keywords were spoken relative to the total utterance time from the start to the end of voice input, the user's level of understanding in the dialogue is determined, so dialogue control The means can appropriately execute the dialogue control according to the estimated degree of understanding.

  According to a fourth aspect of the present invention, there is provided a voice interaction apparatus comprising: voice input means for inputting voice spoken by a user (for example, a microphone 1 in an embodiment described later); and voice recognition means for performing processing for recognizing input voice. (For example, a voice recognition unit 22 in an embodiment described later), keyword determination means (for example, a keyword determination unit 23 in an embodiment described later) for extracting a predetermined keyword from the recognized voice of the user, and uttering the keyword A keyword reference utterance time storage means (for example, a keyword database 25 in an embodiment to be described later) stored for each keyword as a keyword reference utterance time as a standard time required for the operation, and the user utters the keyword itself. Keyword utterance duration measuring means for measuring the keyword utterance duration required for (for example, the time key of the embodiment described later) The processing of step S42 executed by the link combining unit 27 and the understanding level calculation unit 28), and an understanding for determining the level of understanding of the user in the dialogue based on the ratio of the keyword utterance duration to the keyword reference utterance time Degree control means (for example, the processes in steps S43 to S45 and the process in step S23 executed by the understanding degree calculation unit 28 of the embodiment described later) and the dialogue control for controlling the dialogue response according to the understanding degree of the user And a means (for example, a dialogue control unit 29 in an embodiment to be described later).

  The speech dialogue apparatus having the above-described configuration is a speech dialogue apparatus that includes voice input means for inputting voice uttered by a user, and voice recognition means for performing recognition processing of the input voice. For example, if the user utters without confidence, the keyword utterance duration required for uttering important keywords is long, and if the user utters confidently, Necessary for the user to utter the keyword itself for the keyword-based utterance time that is normally required for the utterance of a given keyword, such as the length of the keyword utterance duration required is short and the understanding level of the user's dialogue is high. Since the user's level of understanding in the dialog is determined based on the ratio of the keyword utterance durations, the dialog control means is suitable for the estimated level of understanding. It is possible to perform an interactive control on.

  According to a fifth aspect of the present invention, there is provided a voice interactive apparatus comprising: voice input means for inputting voice spoken by a user (for example, a microphone 1 in an embodiment described later); and voice recognition means for performing processing for recognizing input voice. In addition to (for example, a voice recognition unit 22 in an embodiment described later) and keyword determination means (for example, a keyword determination unit 23 in an embodiment described later) for extracting a predetermined keyword from the recognized voice of the user, Input sound number measuring means for measuring the number of sounds of the recognized user's voice (for example, the processing from step S11 to step S12 executed by the recognized word counting unit 26 and the understanding level calculating unit 28 of the embodiment described later), Alternatively, keyword appearance time measuring means for measuring the keyword appearance time from when the user requests utterance until the user utters the keyword (for example, after The time keyword combining unit 27 and the understanding level calculating unit 28 of the embodiment to be executed), or the total utterance time from the start of the user to the end of the utterance. The utterance time measuring means (for example, the process of step S31 executed by the time keyword combining unit 27 and the comprehension degree calculating unit 28 in the embodiment described later) and the keyword utterance duration required for the user to utter the keyword itself Utterance duration measuring means for measuring the time (for example, the processing of step S32 executed by the time keyword combining unit 27 and the comprehension degree calculating unit 28 in the embodiment described later), or the standard time required to utter the keyword Is a keyword reference utterance time storage means (for example, described later). Keyword database 25) of the embodiment and keyword utterance duration measuring means for measuring a keyword utterance duration required for the user to utter the keyword itself (for example, a time keyword combining unit 27 of the embodiment described later) At least two or more means in the processing of step S42 executed by the comprehension level calculation unit 28, the ratio of the number of sounds of the keyword to the total number of sounds input by the user's utterance, and the keyword appearance time Based on a result of combining at least two of the length of the keyword utterance duration, the ratio of the keyword utterance duration to the total utterance time, and the ratio of the keyword utterance duration to the keyword reference utterance time. Understanding level determination means for determining the level of understanding of the user (e.g., understanding level calculation unit 28 in the embodiment described later) A combination of any two or more of steps S13 to S15, steps S4 to S6, steps S33 to S35, steps S43 to S45, and step S23) It is characterized by comprising dialogue control means (for example, dialogue control unit 29 in an embodiment to be described later) for controlling dialogue response according to the degree of understanding of the user.

  The speech dialogue apparatus having the above-described configuration is a speech dialogue apparatus that includes voice input means for inputting voice uttered by a user, and voice recognition means for performing recognition processing of the input voice. The ratio of the number of sounds of important keywords to the total number of sounds of the input speech, the length of the keyword appearance time until the important keywords are uttered after requesting the input of speech, or the input of speech The percentage of keyword utterance duration during which important keywords were uttered relative to the total utterance time from the start to the end of the utterance, or the user with respect to the keyword reference utterance time that is normally required for utterance of a given keyword Based on the result of combining at least two of the ratios of keyword utterance durations required to speak, Since determining the user's comprehension, dialogue control means can appropriately execute the dialogue control according to the estimated level of understanding.

  According to a sixth aspect of the present invention, there is provided a spoken dialogue apparatus according to any one of the first to fifth aspects, wherein a running environment determining unit (e.g., an embodiment to be described later) for judging a running environment of a mounted vehicle. Vehicle state detection device), and the understanding level determination means determines a threshold value for determining the level of understanding of the user in the dialogue according to the traveling environment of the vehicle determined by the traveling environment determination unit. It is characterized by changing.

  In the voice interaction device having the above configuration, the understanding level determination means changes the threshold value for determining the level of understanding of the user in the dialog according to the traveling environment of the vehicle determined by the traveling environment determination unit. Thus, for example, even when the user is driving, even if the utterance can be presumed to have been input with confidence, the user may be distracted by the driving and the language may be wrong. It is possible to accurately estimate the change in the understanding level of the user accompanying the change in the driving environment.

According to the spoken dialogue apparatus of claim 1, according to the degree of understanding of the user in the dialogue estimated by comparing the total number of sounds input by the user's utterance with the number of sounds of important keywords. Appropriate dialog control can be performed.
Therefore, users who have a high level of understanding perform dialogue control with a simple response, while users with a low level of understanding perform interactive control with a detailed and polite response, so that it is There is an effect that it is possible to realize a voice dialogue apparatus that executes dialogue control and improves user convenience. In addition, by allowing the input of words that do not have a meaning specific to free pronunciation, the effect of eliminating the pressure on the user's utterance as much as possible can be obtained. In addition, by estimating the degree of understanding by comparing the total number of sounds of the input speech with the number of sounds of important keywords, it is influenced by words that have a large range of sounds and polite words attached to the ending. Thus, it is possible to calculate the degree of understanding accurately without executing the appropriate dialogue control.

According to the spoken dialogue apparatus of the second aspect, appropriate dialogue control can be executed according to the degree of understanding estimated by the time until an important keyword is spoken.
Therefore, similarly to the first aspect, it is possible to realize an audio dialogue apparatus that executes appropriate dialogue control according to the level of the user and improves the convenience of the user. In addition, by allowing the input of words that do not have a meaning specific to free pronunciation, the effect of eliminating the pressure on the user's utterance as much as possible can be obtained. In addition, the degree of understanding is estimated based on the time it takes for an important keyword to be spoken, not simply whether or not the user has spoken. As a result, it is possible to execute appropriate dialogue control.

According to the voice dialogue apparatus according to the third aspect, the dialogue control can be appropriately executed according to the understanding level estimated from the time when the important keyword was spoken.
Therefore, similarly to the first aspect, it is possible to realize an audio dialogue apparatus that executes appropriate dialogue control according to the level of the user and improves the convenience of the user. In addition, by allowing the input of words that do not have a meaning specific to free pronunciation, the effect of eliminating the pressure on the user's utterance as much as possible can be obtained. In addition, it supports any language by estimating the level of understanding by comparing the total utterance time from the start of speech input to the end and the keyword utterance duration when the keyword was spoken, Even for words whose number of sounds is difficult to identify, it is possible to accurately calculate the degree of understanding from the ratio of meaningful words to spoken speech and execute appropriate dialogue control.

According to the voice dialogue apparatus according to the fourth aspect, the dialogue control can be appropriately executed according to the degree of understanding estimated by the time required for the user to speak the important keyword itself.
Therefore, similarly to the first aspect, it is possible to realize an audio dialogue apparatus that executes appropriate dialogue control according to the level of the user and improves the convenience of the user. In addition, by allowing the input of words that do not have a meaning specific to free pronunciation, the effect of eliminating the pressure on the user's utterance as much as possible can be obtained. In particular, the overall utterance is estimated by estimating the level of understanding by comparing the keyword-based utterance time that is typically required to utter a given keyword with the keyword utterance duration that is required for the user to utter an important keyword itself. There is no need to check the above, and it is possible to calculate the degree of understanding accurately and execute appropriate dialogue control only by confirming only important keywords.

According to the spoken dialogue apparatus of the fifth aspect, the dialogue control can be appropriately executed according to the degree of understanding estimated by the combined information regarding the important keyword.
Therefore, it is possible to improve the determination accuracy of the level of understanding in dialogue control, accurately calculate the level of understanding without being affected by the content and state of the input speech, and execute appropriate dialogue control. It is done.

According to the voice interactive apparatus according to the sixth aspect, it is possible to accurately estimate the change in the understanding level of the user accompanying the change in the traveling environment of the vehicle.
Therefore, even for a user who is burdened by driving the vehicle, the degree of understanding can be accurately calculated according to the state of the user at that time, and appropriate dialogue control can be executed. . Further, it is possible to improve the convenience for the user and to realize a voice interactive device suitable for mounting on a vehicle.

  Embodiments of the present invention will be described below with reference to the drawings.

  First, the first embodiment will be described. In the first embodiment, the keyword appearance time from the time when the user is requested to input voice (utterance) to the time when an important keyword is uttered, and the time from when the user starts a single utterance to the end. The case where the degree of understanding of the user in the dialogue is estimated based on the ratio of the number of important keyword sounds to the total number of input sounds (number of phonemes) is shown.

(Device configuration)
FIG. 1 is a block diagram showing the overall configuration of the voice interactive apparatus according to the first embodiment of the present invention.
In FIG. 1, the voice interaction apparatus according to the present embodiment includes a microphone 1 for inputting a user's voice, and the user's voice input from the microphone 1 is input to the signal processing unit 2.
The signal processing unit 2 performs speech recognition to convert the input speech into a recognized word, calculates an understanding level in the user's dialogue from the recognized word, and controls the dialogue based on the understanding level. A processing unit that generates a response sentence based on dialogue control, and the response sentence generated by the signal processing unit 2 is input to the speech synthesis unit 3 and the display 4. In addition, the speech synthesizer 3 outputs the response sentence generated in the signal processor 2 to the speaker 5. On the other hand, the display 4 displays the response sentence generated in the signal processing unit 2 on the screen.

(Details of signal processor)
Next, details of the signal processing unit 2 of the voice interactive apparatus according to the present embodiment will be described with reference to the drawings. FIG. 2 is a block diagram illustrating a configuration of the signal processing unit 2 of the voice interactive apparatus according to the present embodiment.
In FIG. 2, the voice input from the microphone 1 is first input to the utterance section detector 21, and the utterance section detector 21 starts the user's utterance section (start time and end time) based on the utterance of the voice interaction device. Is detected. Next, the input voice is input to the voice recognition unit 22, and voice recognition is executed in the voice recognition unit 22 to be converted into a recognized word (text).
In addition, the user's voice and utterance section information and the recognition word are input to the time recognition word combining unit 24, and the time recognition word combining unit 24 is time information when the recognition word is uttered for each recognition word. And the recognized word and time information corresponding to the recognized word are output to the time keyword combining unit 27 described later.

On the other hand, the speech output from the speech recognition unit 22 and its recognition word are input to the keyword determination unit 23, and the keyword determination unit 23 refers to the keyword database 25 in which keywords that are meaningful words in dialogue are stored. A predetermined keyword is extracted from the input recognition word, a tag is assigned to the keyword portion of the recognition word, and the recognized word having the identified keyword is output to the recognition word counting unit 26.
On the other hand, the recognized word counting unit 26 counts the number of phonemes of the keyword (number of sounds) and the number of phonemes of all recognized words including the keyword (total number of sounds), and the number of recognized words and the number of phonemes of the keyword. Is output to the time keyword combining unit 27 together with the recognized word in which the keyword is identified.

  In addition, the time keyword combining unit 27 includes a recognition word input from the time recognition word combining unit 24 and time information corresponding thereto, and a recognition word in which the keyword input from the recognition word counting unit 26 is identified. The time information at which the keyword is spoken is combined with the keyword, and the recognized word in which the keyword is identified and the corresponding time information are output to the comprehension calculator 28.

  In addition, the comprehension level calculation unit 28 uses the number of input recognition words and the number of phonemes of the keywords, the time information corresponding to the recognition words, and the information about the positions of the keywords of the recognition words, and the user in the dialogue. The comprehension level R is calculated as one of the following three states, for example, and is output to the dialogue control unit 29. The understanding level R will be described below. The understanding level R = 0 represents a case where “the user's understanding level is low and it is necessary to request re-input.” It is a normal understanding level, confirms the input content, and proceeds to the next step. " In addition, the understanding level R = 2 represents the case where “the user's understanding level is high, and the process immediately proceeds to the next step”. The method of calculating the understanding level R in the understanding level calculation unit 28 will be described later in detail.

On the other hand, the dialogue control unit 29 is a processing unit that controls the flow of dialogue based on the understanding level R calculated by the understanding level calculation unit 28. For example, for the three states of the above understanding level R, the understanding level If R = 0, a message prompting input again is output. When the understanding level R = 1, the recognition word (text) obtained from the input is confirmed, and then the process proceeds to the next step. Further, when the understanding level R = 2, the process proceeds to the next step without confirming the recognized word (text) obtained from the input. Details of the dialogue progress flow based on the understanding level R executed by the dialogue control unit 29 will be described later.
The response sentence generation unit 30 is a processing unit that generates and outputs a necessary response sentence in accordance with the control of the dialogue control unit 29.

  The keyword database 25 is a non-volatile memory such as a hard disk device, a magneto-optical disk device, or a flash memory, a readable recording medium such as a CD-ROM, or a volatile memory such as a RAM (Random Access Memory). It is assumed that the recording medium is configured by a memory or a computer-readable / writable recording medium.

  Further, the utterance section detection unit 21, the speech recognition unit 22, the keyword determination unit 23, the time recognition word combination unit 24, the recognition word count unit 26, the time keyword combination unit 27, the understanding level calculation unit 28, The dialogue control unit 29 and the response sentence generation unit 30 may be realized by dedicated hardware, and are configured by a memory and a CPU (central processing unit) to realize the functions of the above-described units. The function may be realized by loading the program for loading into the memory and executing the program.

(Understanding level estimation parameter)
Next, an understanding level estimation parameter for the understanding level calculation unit 28 to calculate the understanding level R in the present embodiment will be described.
FIG. 3 is a diagram illustrating a relationship between an utterance example and an element for calculating an understanding level estimation parameter. Note that FIG. 3 is a diagram in which the horizontal axis represents time and the vertical axis represents voice power, and shows both the speech by the speech synthesizer 3 and the user's speech in the speech dialogue apparatus.

  In FIG. 3, time t0 is the time when the speech by the speech synthesizer 3 of the voice interactive apparatus is finished. The time t1 is a time when an important keyword is uttered by the user. The time d indicates the keyword appearance time from when the user requests voice input (utterance) by ending the utterance by the voice synthesizing unit 3 of the voice interactive apparatus until an important keyword is uttered. . Further, the time T2 indicates the total utterance time from when the user starts a single utterance to when it ends. On the other hand, the time Tk indicates a keyword utterance duration required for the user to utter an important keyword itself. Note that these pieces of information correspond to the recognized word in which the keyword output from the time keyword combining unit 27 is identified and the corresponding time information.

  On the other hand, in FIG. 3, double circles indicate phonemes of input speech, and correspond to information regarding the number of recognized phonemes and keyword phonemes output from the recognized word counting unit 26. For example, a recognition word based on a voice uttered by a user, “Eh ~~, Maihama.” Is composed of 8 phonemes (number of sounds = 8) and is an important keyword “ The “maima” portion is composed of four phonemes (number of sounds = 4). In addition, the words “U ~~,” and “I.” indicated by bold underlines are words that have no meaning in the utterance.

  On the other hand, for each of the above elements, the understanding level estimation parameter s1 is shown as a function having the keyword appearance time d as a variable. The understanding level estimation parameter s2 is indicated by the ratio of the number of sounds of important keywords to the total number of sounds input from the start to the end of the single utterance by the user. For example, in the utterance example shown in FIG. 3, s2 = 4/8 = 0.5.

(Calculation procedure of understanding level estimation parameter s1)
Next, the calculation procedure of the understanding level estimation parameter s1 in the understanding level calculation unit 28 will be described with reference to the drawings. FIG. 4 is a flowchart showing a calculation procedure of the understanding level estimation parameter s1 in the understanding level calculation unit 28.
In FIG. 4, first, the comprehension calculator 28 obtains the time t0 when the utterance by the speech synthesizer 3 of the speech dialogue apparatus is finished (step S1).
Next, a time t1 at which the utterance of an important keyword is started by the user is acquired (step S2).

Then, the important keyword is obtained after subtracting the time t0 when the speech synthesizer 3 ends the utterance from the time t1 when the utterance of the important keyword is started by the user and requesting the user to input the voice (utterance). The keyword appearance time d (d = t1−t0) until the utterance is spoken is calculated. (Step S3).
If the keyword appearance time d can be calculated, it is determined whether the keyword appearance time d is shorter than the predetermined time Td (step S4).

If it is determined in step S4 that the keyword appearance time d is shorter than the predetermined time Td (d <Td) (YES in step S4), the understanding level of the user in the dialogue is estimated to be high, and the understanding level estimation parameter s1 is set to “ 1 ”is set (s1 = 1) (step S5).
If the keyword appearance time d is greater than or equal to the predetermined time Td in step S4 (d ≧ Td) (NO in step S4), the understanding level of the user in the dialogue is estimated to be low, and the understanding level estimation parameter s1 is set. “0” is set (s1 = 0) (step S6).

(Calculation procedure of understanding level estimation parameter s2)
Next, a procedure for calculating the understanding level estimation parameter s2 in the understanding level calculation unit 28 will be described with reference to the drawings. FIG. 5 is a flowchart showing a calculation procedure of the understanding level estimation parameter s2 in the understanding level calculation unit 28.
In FIG. 5, first, the comprehension calculator 28 obtains the total number of sounds p0 input from when the user starts a single utterance until it ends (step S11).
Next, the number of sounds p1 of an important keyword uttered by the user is acquired (step S12).
Then, it is determined whether or not the ratio (p1 / p0) of the keyword sound number p1 to the input total sound number p0 is 0.5 or more, for example (step S13).

If the ratio (p1 / p0) of the keyword sound number p1 to the input total sound number p0 is 0.5 or more in step S13 (YES in step S13), the user's level of understanding in the dialogue is Assuming that it is high, “1” is set to the understanding level estimation parameter s2 (s2 = 1) (step S14).
In addition, in step S13, when the ratio (p1 / p0) of the keyword sound number p1 to the input total sound number p0 is less than 0.5 (NO in step S13), the user's level of understanding in the dialogue is It is estimated that the value is low, and “0” is set to the understanding level estimation parameter s2 (s2 = 0) (step S15).

(Procedure for understanding R)
Next, the calculation procedure of the understanding level R of the understanding level calculation unit 28 in the present embodiment will be described. In the understanding level calculation unit 28, the understanding level R is calculated using the above-described understanding level estimation parameters s1 and s2. More specifically, in this embodiment, the understanding level R is calculated by the following equation (1) by combining the understanding level estimation parameters s1 and s2. Therefore, in this embodiment, the understanding level R is calculated as one of three states of R = 0, R = 1, and R = 2.

R = s1 + s2 (1)

(Dialogue progress flow)
Next, a dialogue progress flow based on the understanding level R of the voice dialogue apparatus of this embodiment will be described with reference to the drawings. FIG. 6 is a flowchart showing a dialogue progress flow based on the understanding level R of the voice dialogue apparatus of the present embodiment.
In FIG. 6, first, the signal processing unit 2 outputs a message for prompting voice input from the speaker 5 through the voice synthesis unit 3 during the dialogue, and asks the user for voice input (step S <b> 21).

On the other hand, when a sound is input from the microphone 1 (step S22), the signal processing unit 2 calculates an understanding level R in the understanding level calculation unit 28 (step S23).
Then, the signal processing unit 2 executes dialogue control based on the calculated understanding level R in the dialogue control unit 29, and determines the flow of dialogue (step S24).
Specifically, when the understanding level R = 0 (step S24: R = 0), it is estimated that the user's level of understanding is low and it is necessary to request re-input, so the signal processing unit 2 performs step S21. Return to, and output a message prompting input again.

  On the other hand, when the understanding level is R = 1 (step S24: R = 1), the user's level of understanding is estimated to be normal, and it is only necessary to confirm the input content and proceed to the next step. A confirmation message for the input recognition word (text) is output (step S25), and the user's confirmation voice input for the recognition word (text) obtained from the input is waited (step S26). It is determined whether or not the text is correct by the user (step S27).

As a result, if the user confirms that the recognized word (text) recognized by voice is correct (YES in step S27), the process proceeds to the next step.
If the recognized word (text) recognized by voice is not confirmed by the user (NO in step S27), the process returns to step S21, and a message prompting input is output again.
Furthermore, in the case of understanding level R = 2 (step S24: R = 2), it is estimated that the user's level of understanding is high, and it is only necessary to proceed to the next step, so the signal processing unit 2 is obtained from the input. Proceed to the next step without checking the recognized word (text).

  In the present embodiment, the signal processing unit 2 includes an input sound number measurement unit, an understanding level determination unit, and a keyword appearance time measurement unit. Specifically, the processing from step S11 to step S12 executed by the understanding level calculation unit 28 in addition to the recognized word counting unit 26 corresponds to the input sound number measurement means. Further, the processing from step S4 to step S6, the processing from step S13 to step S15, and the processing from step S23 executed by the understanding level calculation unit 28 correspond to the understanding level determination means. In addition to the time keyword combining unit 27, the processing from step S1 to step S3 executed by the understanding level calculating unit 28 corresponds to the keyword appearance time measuring means.

  As described above, according to the voice interactive apparatus of the present embodiment, an important keyword is obtained after the comprehension calculator 28 requests voice input (utterance) from the user for the voice input from the microphone 1. Important keywords that account for the understanding level estimation parameter s1 expressed as a function of the keyword appearance time d until the utterance and the total number of sounds p0 input from the start to the end of the single utterance by the user. The comprehension level estimation parameter s2 calculated from the ratio of the number of sounds p1 (p1 / p0) is obtained, and the comprehension level of the user in the dialog is calculated as the understanding level R from the understanding level estimation parameters s1 and s2. Then, the dialogue control unit 29 executes dialogue control based on the calculated understanding level R. For example, when the understanding level R = 0, the dialogue control unit 29 determines that the understanding level of the user is low and outputs a message prompting input again. To do. If the understanding level R = 1, it is determined that the user's level of understanding is normal, the input content is confirmed, and the process proceeds to the next step. Further, when the understanding level R = 2, it is determined that the user has a high level of understanding level, and the process proceeds to the next step without confirming the recognized word obtained from the input.

  Therefore, users who have a high level of understanding perform dialogue control with a simple response, while users with a low level of understanding perform interactive control with a detailed and polite response, so that it is There is an effect that it is possible to realize a voice dialogue apparatus that executes dialogue control and improves user convenience. In addition, by allowing the input of words that do not have a meaning specific to free pronunciation, the effect of eliminating the pressure on the user's utterance as much as possible can be obtained. In addition, the degree of understanding is estimated based on the time it takes for an important keyword to be spoken, not simply whether or not the user has spoken. In addition, by estimating the comprehension level by comparing the total number of input voices with the number of important keywords, the influence of words with a large range of sounds and polite words attached to the ending It is possible to obtain the effect that the degree of understanding can be accurately calculated without being received and appropriate dialogue control can be executed.

  Next, a second embodiment will be described. In the second embodiment, the keyword appearance time from the time when the user requests voice input (speech) to the time when an important keyword is uttered, and the time from when the user starts a single utterance until it ends. The case where the user's comprehension level in the dialogue is estimated by the ratio of the keyword utterance duration required for the user to utter the important keyword itself in the total utterance time is shown.

(Details of device configuration and signal processing unit)
Since the details of the apparatus configuration and the signal processing unit in this embodiment are the same as those in the first embodiment, description thereof is omitted here.

(Understanding level estimation parameter)
Next, an understanding level estimation parameter for the understanding level calculation unit 28 to calculate the understanding level R in the present embodiment will be described.
Specifically, for each element for calculating the understanding level estimation parameter shown in FIG. 3, the understanding level estimation parameter s1 is a function having the keyword appearance time d as a variable, as in the first embodiment. Indicated. On the other hand, the understanding level estimation parameter s3 is the keyword utterance duration time Tk required for the user to utter an important keyword itself in the total utterance time T2 from the start to the end of the single utterance. , Ie, a function having “Tk / T2” as a variable.

(Calculation procedure of comprehension level estimation parameter s3)
Next, the calculation procedure of the understanding level estimation parameter s3 in the understanding level calculation unit 28 will be described with reference to the drawings. FIG. 7 is a flowchart showing a calculation procedure of the understanding level estimation parameter s3 in the understanding level calculation unit 28.
In FIG. 7, first, the comprehension calculator 28 obtains a total utterance time T2 from when the user starts a single utterance to when it ends (step S31).
Next, keyword utterance duration Tk required for the user to utter an important keyword itself is acquired (step S32).
Then, it is determined whether or not the ratio (Tk / T2) of the keyword utterance duration Tk to the total utterance time T2 of the user is 0.5 or more, for example (step S33).

If the ratio of the keyword utterance duration Tk to the total utterance time T2 of the user (Tk / T2) is 0.5 or more in step S33 (YES in step S33), the user's level of understanding in the dialogue is Assuming that it is high, “1” is set to the understanding level estimation parameter s3 (s3 = 1) (step S34).
If the ratio of the keyword utterance duration Tk to the total utterance time T2 of the user (Tk / T2) is less than 0.5 (NO in step S33), the user's level of understanding in the dialogue is It is estimated that the value is low, and “0” is set to the understanding level estimation parameter s3 (s3 = 0) (step S35).

(Procedure for understanding R)
Next, the calculation procedure of the understanding level R of the understanding level calculation unit 28 in the present embodiment will be described. In the understanding level calculation unit 28, the understanding level R is calculated using the above-described understanding level estimation parameters s1 and s3. More specifically, in this embodiment, the understanding level R is calculated by the following equation (2) by combining the understanding level estimation parameters s1 and s3. Therefore, also in the present embodiment, the understanding level R is calculated as one of three states of R = 0, R = 1, and R = 2.

R = s1 + s3 (2)

(Dialogue progress flow)
In addition, when the understanding level R can be calculated, the voice interaction apparatus according to the present embodiment also executes dialogue control based on the dialogue progress flow based on the understanding level R described with reference to FIG. 6 in the first example. .

  In the present embodiment, the signal processing unit 2 includes total utterance time measuring means, keyword utterance duration measuring means, understanding level determining means, and keyword appearance time measuring means. Specifically, in addition to the time keyword combining unit 27, the process of step S31 executed by the understanding level calculating unit 28 corresponds to the total utterance time measuring means. In addition to the time keyword combining unit 27, the process of step S32 executed by the understanding level calculating unit 28 corresponds to a keyword utterance duration measuring unit. Further, the processing from step S33 to step S35, the processing from step S13 to step S15, and the processing from step S23 described in the first embodiment correspond to the understanding level determination unit. In addition to the time keyword combining unit 27 described in the first embodiment, the processing from step S1 to step S3 executed by the understanding level calculating unit 28 corresponds to the keyword appearance time measuring means.

  As described above, according to the voice interactive apparatus of the present embodiment, an important keyword is obtained after the comprehension calculator 28 requests voice input (utterance) from the user for the voice input from the microphone 1. Keywords that are important for the user in the total utterance time T2 from the start to the end of the single utterance by the understanding level estimation parameter s1 expressed as a function of the keyword appearance time d until the utterance A comprehension level estimation parameter s3 calculated from the ratio (Tk / T2) of the keyword utterance duration Tk required to speak itself is obtained, and the comprehension level of the user in the dialog is understood from the understanding level estimation parameters s1 and s3. Calculated as degree R. Then, the dialogue control unit 29 executes dialogue control based on the calculated understanding level R. For example, when the understanding level R = 0, the dialogue control unit 29 determines that the understanding level of the user is low and outputs a message prompting input again. To do. If the understanding level R = 1, it is determined that the user's level of understanding is normal, the input content is confirmed, and the process proceeds to the next step. Further, when the understanding level R = 2, it is determined that the user has a high level of understanding level, and the process proceeds to the next step without confirming the recognized word obtained from the input.

  Therefore, as in the first embodiment, a user with a high level of understanding performs interactive control with a simple response, while a user with a low level of understanding executes interactive control with a detailed and polite response. Thus, it is possible to realize an audio dialogue apparatus that executes appropriate dialogue control according to the level of the user and improves the convenience of the user. In addition, by allowing the input of words that do not have a meaning specific to free pronunciation, the effect of eliminating the pressure on the user's utterance as much as possible can be obtained. In addition, the degree of understanding is estimated based on the time it takes for an important keyword to be spoken, not simply whether or not the user has spoken. In addition, it can handle any language by estimating the degree of understanding by comparing the total utterance time from the start to the end of voice input with the keyword utterance duration when the keyword was spoken. Even for words whose number of sounds is difficult to identify, it is possible to accurately calculate the degree of understanding from the ratio of meaningful words to spoken speech and execute appropriate dialogue control.

  Next, a third embodiment will be described. In the third embodiment, the keyword appearance time from when the user requests voice input (speech) to the time when an important keyword is uttered, and the standard time required to utter a predetermined keyword. A case is shown in which the user's understanding level in the dialogue is estimated based on the ratio of the keyword utterance duration, which is the time required for the user to utter the keyword itself with respect to the reference utterance time.

(Details of device configuration and signal processing unit)
Since the details of the apparatus configuration and the signal processing unit in this embodiment are the same as those in the first embodiment, description thereof is omitted here.
However, in this embodiment, it is assumed that the keyword stored in the keyword database 25 stores a keyword reference utterance time, which is a standard time required to utter the keyword, for each keyword. In addition, the keyword determination unit 23 extracts a predetermined keyword from the input recognition word, adds a tag to the keyword portion of the recognition word, and uses the recognized word having the keyword identified as the keyword reference utterance time of the keyword. The keyword-based utterance time is further output to the comprehension level calculation unit 28 via the recognition word count unit 26 and the time keyword combination unit 27.

(Understanding level estimation parameter)
Next, an understanding level estimation parameter for the understanding level calculation unit 28 to calculate the understanding level R in the present embodiment will be described.
Specifically, for each element for calculating the understanding level estimation parameter shown in FIG. 3, the understanding level estimation parameter s1 is a function having the keyword appearance time d as a variable, as in the first embodiment. Indicated. On the other hand, the understanding level estimation parameter s4 is a keyword utterance duration time which is a time required for the user to utter the keyword itself with respect to the keyword reference utterance time Ds which is a standard time required to utter a predetermined keyword. It is shown as a function having the ratio of Tk, that is, “Tk / Ds” as a variable.

(Calculation procedure of understanding level estimation parameter s4)
Next, the calculation procedure of the understanding level estimation parameter s4 in the understanding level calculation unit 28 will be described with reference to the drawings. FIG. 8 is a flowchart showing a calculation procedure of the understanding level estimation parameter s4 in the understanding level calculation unit 28.
In FIG. 8, first, the comprehension calculator 28 obtains a keyword reference utterance time Ds that is a standard time required to utter an input important keyword (step S41).
Next, the keyword utterance duration time Tk required for the user to utter this important keyword itself is acquired (step S42).
Then, it is determined whether or not the ratio (Tk / Ds) of the keyword utterance duration Tk to the keyword reference utterance time Ds required to utter a keyword is 1.0 or less, for example (step S43).

If the ratio (Tk / Ds) of the keyword utterance duration Tk to the keyword reference utterance time Ds required to utter a keyword is 1.0 or less (YES in step S43), it is used in the dialog. The understanding level of the person is estimated to be high, and “1” is set to the understanding level estimation parameter s4 (s4 = 1) (step S44).
In addition, in step S43, when the ratio (Tk / Ds) of the keyword utterance duration Tk to the keyword reference utterance time Ds required to utter the keyword is larger than 1.0 (NO in step S43), the user in the dialogue Is understood to be low, and “0” is set to the understanding level estimation parameter s4 (s4 = 0) (step S45).

(Procedure for understanding R)
Next, the calculation procedure of the understanding level R of the understanding level calculation unit 28 in the present embodiment will be described. In the understanding level calculation unit 28, the understanding level R is calculated using the above-described understanding level estimation parameters s1 and s4. More specifically, in this embodiment, the understanding level R is calculated by the following equation (3) by combining the understanding level estimation parameters s1 and s4. Therefore, also in the present embodiment, the understanding level R is calculated as one of three states of R = 0, R = 1, and R = 2.

R = s1 + s4 (3)

(Dialogue progress flow)
In addition, when the understanding level R can be calculated, the voice interaction apparatus according to the present embodiment also executes dialogue control based on the dialogue progress flow based on the understanding level R described with reference to FIG. 6 in the first embodiment. .

  In the present embodiment, the signal processing unit 2 includes a keyword utterance duration measuring unit, an understanding level determining unit, and a keyword appearance time measuring unit. Specifically, in addition to the time keyword combining unit 27, the process of step S42 executed by the understanding level calculating unit 28 corresponds to a keyword utterance duration measuring unit. Further, the processing from step S43 to step S45 executed by the understanding level calculation unit 28, the processing from step S13 to step S15, and the processing from step S23 described in the first embodiment correspond to the understanding level determination means. In addition to the time keyword combining unit 27 described in the first embodiment, the processing from step S1 to step S3 executed by the understanding level calculating unit 28 corresponds to the keyword appearance time measuring means.

  As described above, according to the voice interactive apparatus of the present embodiment, an important keyword is obtained after the comprehension calculator 28 requests voice input (utterance) from the user for the voice input from the microphone 1. The user with respect to the understanding level estimation parameter s1 expressed as a function of the keyword appearance time d until the utterance and the keyword reference utterance time Ds, which is a standard time required to utter a predetermined keyword, is used by the user. The comprehension level estimation parameter s4 calculated from the ratio (Tk / Ds) of the keyword utterance duration Tk, which is the time required to speak, is obtained, and the comprehension level of the user in the dialogue is determined from the understanding level estimation parameters s1 and s4. Calculated as an understanding level R. Then, the dialogue control unit 29 executes dialogue control based on the calculated understanding level R. For example, when the understanding level R = 0, the dialogue control unit 29 determines that the understanding level of the user is low and outputs a message prompting input again. To do. If the understanding level R = 1, it is determined that the user's level of understanding is normal, the input content is confirmed, and the process proceeds to the next step. Further, when the understanding level R = 2, it is determined that the user has a high level of understanding level, and the process proceeds to the next step without confirming the recognized word obtained from the input.

  Therefore, as in the first embodiment, a user with a high level of understanding performs interactive control with a simple response, while a user with a low level of understanding executes interactive control with a detailed and polite response. Thus, it is possible to realize an audio dialogue apparatus that executes appropriate dialogue control according to the level of the user and improves the convenience of the user. In addition, by allowing the input of words that do not have a meaning specific to free pronunciation, the effect of eliminating the pressure on the user's utterance as much as possible can be obtained. In addition, the degree of understanding is estimated based on the time it takes for an important keyword to be spoken, not simply whether or not the user has spoken. In addition, the degree of understanding is estimated by comparing the keyword-based utterance time that is normally required for utterance of a predetermined keyword with the keyword utterance duration that is required for the user to utter an important keyword itself. Without checking the whole, it is possible to obtain an effect that an understanding level can be accurately calculated and appropriate dialogue control can be executed only by checking only important keywords.

  In the first to third embodiments described above, for example, in step S4, step S13, step S33, or step S43, threshold values for determining the understanding level estimation parameters s1 to s4 are fixed. However, the electronic apparatus control apparatus described in the first to third embodiments is particularly useful when mounted on a vehicle such as an automobile. As an example, the electronic apparatus control apparatus When the is installed in an automobile, the threshold values for determining these understanding level estimation parameters s1 to s4 may be varied based on the traveling environment of the vehicle.

  Specifically, for example, the signal processing unit 2 includes a navigation device equipped with a GPS (Global Positioning System), a speed sensor that detects the traveling speed of the vehicle, and a sensor provided in a steering, a brake, or the like. A vehicle state detection device that detects a vehicle traveling state and a driving state including a position, a traveling speed, a user's driving operation, and the like is connected.

  Then, the comprehension calculation unit 28 of the signal processing unit 2 acquires the position information of the vehicle, the traveling speed, the driving state, etc. as needed, so that it is important for the keyword appearance time d or the total number of sounds p0 spoken by the user. The ratio of the number of sounds p1 of a simple keyword (p1 / p0), or the ratio of the keyword utterance duration Tk required to utter an important keyword itself in the total utterance time T2 uttered by the user (Tk / T2), Alternatively, even if the ratio (Tk / Ds) of the keyword utterance duration Tk required for the user to utter the keyword itself with respect to the keyword reference utterance time Ds required to utter the predetermined keyword is the same, It is a burden on the user based on the driving conditions and driving conditions of the vehicle including the driving speed and the driving operation of the user, i.e. Is If a can be estimated depends, the threshold for determining the comprehension estimation parameter s1 to s4, the comprehension of the user for each of the determination results change as estimated low.

  As a result, while the user is driving the vehicle, even if the utterance can be presumed to have been input with confidence, the user may be distracted by the driving, and the language may be wrong. Do not confirm the input contents and proceed to the next step immediately, do not "re-input" or "confirm the input contents and proceed to the next step", etc. Accordingly, it is possible to execute an appropriate dialogue control.

  In the first embodiment, the understanding level R is obtained from the combination of the understanding level estimation parameter s1 and the understanding level estimation parameter s2. In the second example, the understanding level estimation parameter s1 and the understanding level estimation parameter s3 are obtained. The understanding level R is obtained from the combination, and in the third embodiment, the understanding level R is obtained from the combination of the understanding level estimation parameter s1 and the understanding level estimation parameter s4. In each example, the understanding level R is set to R = 0, The dialogue control is executed by calculating as one of the three states of R = 1 and R = 2, but the combination of the understanding level estimation parameters for obtaining the understanding level R is not limited to this, and the understanding level estimation parameter s2 and Combination of understanding level estimation parameter s3, or combination of understanding level estimation parameter s2 and understanding level estimation parameter s4, or understanding level estimation parameter s3 and understanding level estimation parameter From the combination of s4, the understanding level R may be calculated as any one of the three states of R = 0, R = 1, and R = 2. Even if the understanding level estimation parameter is such a combination, the first to third The same effect as the embodiment can be obtained.

  Furthermore, the combination of the understanding level estimation parameters for obtaining the understanding level R is not limited to two, and one understanding level estimation parameter is associated with the understanding level R, and the understanding level R is 2 (R = 0, R = 1). One of the two states may be calculated and interactive control may be performed, or the understanding level R may be set to correspond to 3 or 4 understanding level estimation parameters, and the understanding level R may be set to 4 states of R = 0 to 3; Or you may calculate as either of the five states of R = 0-4, and may perform dialog control. In this case, in the dialogue progress flow based on the understanding level R described with reference to FIG. 6 in the above-described embodiment, the processing flow is changed according to the number of states of the understanding level R.

It is a block diagram which shows the whole structure of the voice interactive apparatus of 1st Example of this invention. It is a block diagram which shows the structure of the signal processing part of the voice interactive apparatus of the Example. It is a figure which shows the relationship between the utterance example and the element for calculating an understanding level estimation parameter. It is a flowchart which shows the calculation procedure of the understanding level estimation parameter s1 in an understanding level calculation part. It is a flowchart which shows the calculation procedure of the understanding level estimation parameter s2 in an understanding level calculation part. It is a flowchart which shows the dialogue progress flow based on the understanding degree R of the speech dialogue apparatus of the Example. It is a flowchart which shows the calculation procedure of the understanding level estimation parameter s3 of the understanding level calculation part in 2nd Example of this invention. It is a flowchart which shows the calculation procedure of the understanding level estimation parameter s4 of the understanding level calculation part in the 3rd Example of this invention.

Explanation of symbols

1 Microphone (voice input means)
22 Voice recognition unit (voice recognition means)
23 Keyword determination unit (keyword determination means)
25 Keyword Database (Keyword-based utterance time storage means)
26 Recognition word count unit (input sound number measuring means)
27 Time keyword combining unit (keyword appearance time measuring means, total utterance time measuring means, keyword utterance duration measuring means)
28 Understanding level calculator (keyword appearance time measuring means, total utterance time measuring means, keyword utterance duration measuring means, understanding level judging means)
29 Dialogue control unit (dialogue control means)
S1 to S3 Keyword appearance time measuring means S11 to S12 Input sound number measuring means S31 Total utterance time measuring means S32 and S42 Keyword utterance duration measuring means S4 to S6, S13 to S15, S33 to S35, S43 to S45 Understanding degree judging means

Claims (6)

Voice input means for inputting voice spoken by the user;
Speech recognition means for performing recognition processing of input speech;
Keyword determination means for extracting a predetermined keyword from the recognized voice of the user;
Input sound number measuring means for measuring the sound number of the recognized user's voice;
Understanding level determination means for determining the level of understanding of the user in the dialogue based on the ratio of the number of sounds of the keyword to the total number of sounds input by the user's utterance;
A spoken dialogue apparatus comprising dialogue control means for controlling dialogue response according to the degree of understanding of the user. Voice input means for inputting voice spoken by the user;
Speech recognition means for performing recognition processing of input speech;
Keyword determination means for extracting a predetermined keyword from the recognized voice of the user;
Keyword appearance time measuring means for measuring a keyword appearance time from when the user requests utterance until the user utters the keyword;
Understanding level determination means for determining the level of understanding of the user in the dialogue based on the length of the keyword appearance time;
A spoken dialogue apparatus comprising dialogue control means for controlling dialogue response according to the degree of understanding of the user. Voice input means for inputting voice spoken by the user;
Speech recognition means for performing recognition processing of input speech;
Keyword determination means for extracting a predetermined keyword from the recognized voice of the user;
A total utterance time measuring means for measuring a total utterance time from the start of the user to the end of the utterance;
Keyword utterance duration measuring means for measuring a keyword utterance duration required for the user to utter the keyword itself;
Understanding level determination means for determining the level of understanding of the user in the dialogue based on the ratio of the keyword utterance duration to the total utterance time;
A spoken dialogue apparatus comprising dialogue control means for controlling dialogue response according to the degree of understanding of the user. Voice input means for inputting voice spoken by the user;
Speech recognition means for performing recognition processing of input speech;
Keyword determination means for extracting a predetermined keyword from the recognized voice of the user;
A keyword-based utterance time storage means for storing each keyword as a keyword-based utterance time as a standard time required to utter the keyword;
Keyword utterance duration measuring means for measuring a keyword utterance duration required for the user to utter the keyword itself;
Understanding level determination means for determining the level of understanding of the user in dialogue based on the ratio of the keyword utterance duration to the keyword reference utterance time;
A spoken dialogue apparatus comprising dialogue control means for controlling dialogue response according to the degree of understanding of the user. Voice input means for inputting voice spoken by the user;
Speech recognition means for performing recognition processing of input speech;
In addition to keyword determination means for extracting a predetermined keyword from the recognized voice of the user,
Furthermore, the input sound number measuring means for measuring the sound number of the recognized user's voice,
Alternatively, keyword appearance time measuring means for measuring a keyword appearance time from when the user requests utterance until the user utters the keyword,
Alternatively, the total utterance time measuring means for measuring the total utterance time from the start of the user to the end of the utterance and the keyword utterance for measuring the keyword utterance duration required for the user to utter the keyword itself Duration measurement means,
Alternatively, keyword reference utterance time storage means for storing each keyword as a keyword reference utterance time as a standard time required for uttering the keyword, and keyword utterance duration required for the user to utter the keyword itself At least two sets of keyword utterance duration measuring means for measuring
The ratio of the number of sounds of the keyword to the total number of sounds input by the user's utterance, the length of the keyword appearance time, the ratio of the keyword utterance duration to the total utterance time, and the keyword reference utterance time Understanding level determination means for determining the level of understanding of the user in a dialogue based on a combination of at least two of the ratios of the keyword utterance durations to
A spoken dialogue apparatus comprising dialogue control means for controlling dialogue response according to the degree of understanding of the user. A travel environment determination unit for determining the travel environment of the mounted vehicle;
The understanding level determination means changes a threshold value for determining the level of understanding of the user in the dialogue according to the driving environment of the vehicle determined by the driving environment determination unit. The voice interactive apparatus according to any one of claims 1 to 5.

Images