JP2004029804A - Voice recognition conversation device and voice recognition conversation processing method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To improve a recognition rate of voice due to a difference in an age and a gender and to realize conversations adapted for the gender and the various ages by making a part of ROM storing a storage content which is previously set for recognizing voice and outputting a response to recognized voice to be a cartridge type. <P>SOLUTION: In a voice recognition conversation device, inputted voice is analyzed, voice feature data is generated, previously registered standard voice feature data is compared with voice feature data, word detection data is outputted, a meaning of input voice is understood by receiving word detection data and a response content corresponding to the meaning is decided and outputted. A storage means storing the storage content which is previously set for recognizing voice and the storage content which is previously set for outputting the response to recognized voice is disposed on a cartridge 20-side which is loaded on a device main body 1 so that it can freely be loaded/unloaded. When the cartridge 20 is loaded on a device main body 1-side, it is connected to a voice recognition response processing part 10 arranged on the device main body 1-side and the response content for input sound is outputted based on data stored in the cartridge 20. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
[Industrial applications]
The present invention relates to a voice recognition dialogue apparatus and a voice recognition dialogue processing method for recognizing voice and performing a response or a specific operation corresponding to the recognition result.
[0002]
[Prior art]
In this type of speech recognition apparatus, there are a specific speaker speech recognition apparatus capable of recognizing only a specific speaker's speech and an unspecified speaker speech recognition apparatus capable of recognizing an unspecified speaker's speech.
[0003]
The specific-speaker voice recognition device registers a standard voice signal pattern of the specific speaker by inputting words recognizable by a specific speaker one by one according to a predetermined procedure. After the end, when the specific speaker speaks the registered word, the speech recognition is performed by comparing the feature pattern obtained by analyzing the input speech with the registered feature pattern. An example of this type of speech recognition dialogue device is a speech recognition toy. For example, as a plurality of command words to be voice commands, about 10 words such as "good morning", "good night", "hello" are registered in advance by a child who uses the toy, and the speaker uses, for example, "good morning". "" Means that the voice signal is compared with the registered "good morning" voice signal, and when both voice signals match, a predetermined electrical signal for the voice command is output, and based on this, the toy is transmitted to the toy. A specific operation is performed.
[0004]
Such a specific-speaker voice recognition device can recognize only a voice having a voice pattern close to or specific to a specific speaker, and it is necessary to register all words to be recognized one by one as an initial setting. Was extremely troublesome.
[0005]
On the other hand, the unspecified speaker voice recognition device generates in advance the standard voice feature data of the recognition target word as described above using voices uttered by a large number (for example, about 200) of speakers. It is stored (registered), and it is possible to recognize the voice of an unspecified speaker for these pre-registered recognizable words.
[0006]
[Problems to be solved by the invention]
The unspecified speech recognition device capable of recognizing the unspecified speaker's speech certainly secures a relatively high recognition rate for standard speech, but also for almost all speech. A recognition rate is not always obtained. For example, voice characteristics vary greatly depending on age and gender, such as infant voices, adult voices, female voices, and male voices, and even when extremely high recognition rates are obtained for adult questions, There is also a problem that little is recognized when asked by a child.
[0007]
Also, when this type of speech recognition device is applied to a toy such as a stuffed toy, the content of the dialog usually changes depending on the age, sex, etc. of the child playing with the stuffed toy. For example, it is common for infants and elementary school seniors and boys and girls to seek different dialogue content.
[0008]
However, in this type of speech recognition apparatus, the number of recognizable registered words is also limited, and the response to the word is generally limited to a certain extent. Therefore, this kind of toy usually gets tired in a short period of time, and as described above, there is also a problem in the quality of the recognition rate due to the characteristics of the voice according to age and gender. The same applies to not only toys but also all electronic devices that use voice recognition.
[0009]
The present invention has been made in order to solve these problems, and a ROM portion such as a standard voice feature data storage unit for storing standard voice feature data and a response content storage unit for storing response content is a cartridge type. By selecting cartridges as appropriate and enabling them to be installed, recognition corresponding to voice characteristics due to differences in age, gender, etc. becomes possible, while improving recognition rate and enabling conversations corresponding to various ages and genders It is intended to be.
[0010]
[Means for Solving the Problems]
The speech recognition dialogue apparatus of the present invention analyzes input speech to generate speech feature data, compares the speech feature data with standard speech feature data of a pre-registered recognizable word, and detects word detection data. In response to the word detection data, to understand the meaning of the input voice, in the voice recognition dialogue device that determines and outputs the corresponding response content, in the storage content preset for performing voice recognition, A storage unit for storing storage contents set in advance for outputting a response to the recognized voice is provided on a cartridge side detachably mountable to the apparatus main body, and the cartridge is mounted on the apparatus main body. Thereby, the voice recognition response processing unit is connected to the voice recognition response processing unit provided on the apparatus main body side, and based on the data stored in the cartridge, the voice recognition response processing unit It is characterized by generating a response content to.
[0011]
The storage means provided on the cartridge side is standard voice feature data storage means for storing standard voice feature data for pre-registered recognizable words. Input means; voice analysis means for analyzing voice input by the voice input means to generate voice feature data; and inputting voice feature data from the voice analysis means to provide a standard voice provided on the cartridge side. Word detection means for outputting word detection data for the input voice based on the storage content of the characteristic data storage means; receiving word detection data from the word detection means, understanding the meaning of the input voice, and responding accordingly. Speech comprehension conversation control means for determining the content, and speech synthesis output based on the response content determined by the speech comprehension conversation control means Response data instruction content storage means for causing the speech data to be output, and voice synthesis means for generating a speech synthesis output based on the instruction of the response data instruction content storage means in response to the response content determined by the voice understanding conversation control means; And a voice output unit for outputting a voice synthesis output from the voice synthesis unit to the outside.
[0012]
The storage means provided on the cartridge side includes a conversation content storage means for storing a response content corresponding to a recognizable word registered in advance, and an instruction for instructing what kind of speech synthesis output is to be generated. Response data instruction content storage means for storing the content, wherein the apparatus main body includes a voice input means for inputting voice, and a voice for generating voice feature data by analyzing the voice input by the voice input means. Analysis means; standard voice feature data storage means for storing standard voice feature data of pre-registered recognizable words; and word detection data for the input voice based on the stored contents of the standard voice feature data storage means. Receiving the word detection data from the word detecting means, understand the meaning of the input voice, and send the corresponding response content to the cartridge. And a response data instruction content storage means provided on the cartridge side for the response content determined by the voice understanding conversation control means. Speech synthesis means for generating a speech synthesis output based on the instruction, and speech output means for outputting the speech synthesis output from the speech synthesis means to the outside.
[0013]
The storage means provided on the cartridge side is response data instruction content storage means for storing instruction contents for instructing what kind of speech synthesis output is to be generated. Voice input means, a voice analysis means for analyzing voice input by the voice input means to generate voice feature data, and a standard voice for storing standard voice feature data of recognizable words registered in advance. Feature data storage means, word detection means for outputting word detection data for the input speech based on the storage contents of the standard speech feature data storage means, and receiving the word detection data from the word detection means, And a voice understanding conversation control means for determining a response content corresponding thereto, and the response content determined by the voice understanding conversation control means, A configuration comprising: voice synthesis means for generating a voice synthesis output based on the instruction of the response data instruction content recording means provided on the cartridge side; and voice output means for outputting the voice synthesis output from the voice synthesis means to the outside. And
[0014]
The storage means provided on the cartridge side stores standard voice feature data storage means for storing standard voice feature data for pre-registered recognizable words, and stores response contents corresponding to the registered recognizable words. A conversation content storage means, a response data instruction content storage means for instructing what kind of speech synthesis output is to be generated, and a voice input means for inputting a voice on the apparatus main body side; Voice analysis means for analyzing input voice and generating voice feature data; and inputting voice feature data from the voice analysis means, based on storage contents of standard voice feature data storage means provided on the cartridge side. A word detecting means for outputting word detection data for the input voice, and receiving the word detection data from the word detecting means to determine the meaning of the input voice. Comprehending and determining the corresponding response content with reference to the conversation content storage unit provided on the cartridge side, and the response content determined by the voice understanding conversation control means, Speech synthesis means for generating a speech synthesis output based on the instruction of the response data instruction content storage means provided on the cartridge side, and speech output means for outputting the speech synthesis output from the speech synthesis means to the outside And
[0015]
Further, the speech recognition dialogue processing method of the present invention analyzes input speech to generate speech feature data, and compares the speech feature data with standard speech feature data of a pre-registered recognizable word. In the voice recognition dialogue processing method of outputting word detection data, receiving the word detection data, understanding the meaning of the input voice, and determining and outputting the corresponding response content, a preset voice recognition dialogue processing method is used. The stored contents, which are set in advance to output a response to the recognized voice, and the like, are written in storage means on the cartridge side which can be detachably mounted to the apparatus main body, and this cartridge is stored in the apparatus main body. It is characterized in that, by being mounted, a response to an input voice is generated based on data stored in the cartridge.
[0016]
The storage content stored on the cartridge side is standard voice feature data for a pre-registered recognizable word, and the apparatus main body analyzes voice input by voice input means and outputs voice data. A voice analysis step of generating feature data, and a word detection step of inputting voice feature data from the voice analysis step and outputting word detection data for the input voice based on the standard voice feature data stored on the cartridge side Receiving the word detection data from the word detection step, comprehending the meaning of the input voice, and determining a response content corresponding thereto; and a response content determined by the voice understanding conversation control step. On the other hand, a speech synthesis step of generating a speech synthesis output based on the response data instruction content indicating what kind of speech synthesis output is to be performed, A structure having an audio output step of outputting a speech synthesis output from the voice synthesizing process to the outside.
[0017]
The storage contents stored on the cartridge side are response contents corresponding to a pre-registered recognizable word and response data instruction contents for instructing what kind of speech synthesis output is to be generated. The apparatus main body side includes a voice analysis step of analyzing voice input by the voice input means to generate voice feature data, and a word corresponding to the input voice based on standard voice feature data of a pre-registered recognizable word. A word detection step of outputting detection data, receiving the word detection data from the word detection step, understanding the meaning of the input voice, referring to the corresponding response content with reference to the conversation content stored in the cartridge side. A voice understanding conversation control step to be determined, and a response data instruction content stored on the cartridge side with respect to the response content determined by the voice understanding conversation control step. Zui a speech synthesis step of generating a speech synthesis output was, a structure having an audio output step of outputting a speech synthesis output from the speech synthesis step to the outside.
[0018]
Further, the storage content stored in the cartridge side is an instruction content for instructing what kind of voice synthesis output is to be generated, and the apparatus main body side analyzes the voice input by voice input means. A speech analysis step of generating speech feature data, a word detection step of outputting word detection data for an input speech based on standard speech feature data of a pre-registered recognizable word, and a word from the word detection step. Upon receiving the detection data, the meaning of the input voice is understood, and a voice understanding conversation control step of determining a response content corresponding thereto, and the response content determined by the voice understanding conversation control step are stored in the cartridge side. A voice synthesis step for generating a voice synthesis output based on the response data instruction content, and a voice output step for outputting the voice synthesis output from the voice synthesis step to the outside And a configuration.
[0019]
The stored contents stored on the cartridge side include standard voice feature data for pre-registered recognizable words, response contents corresponding to the registered recognizable words, and what kind of voice synthesis output. A voice analysis step of analyzing the voice input by the voice input means to generate voice characteristic data; and a voice characteristic from the voice analysis step. A word detection step of inputting data and outputting word detection data for the input voice based on the standard voice characteristic data stored in the cartridge side; and receiving the word detection data from the word detection step, And a voice understanding conversation control step of determining a response content corresponding thereto with reference to the conversation content stored in the cartridge. A voice synthesis step for generating a voice synthesis output based on the response data instruction content stored on the cartridge side in response to the response content determined by the disassembly / conversation control step; and outputting the voice synthesis output from the voice synthesis step to the outside. And an audio output step of outputting.
[0020]
[Action]
The present invention is directed to a cartridge which can be removably attached to a main body of a device, such as storage contents set in advance for performing voice recognition, and storage contents set in advance for outputting a response to the recognized voice. When the cartridge is mounted on the main body of the apparatus, the contents of the response to the input voice are output based on the data stored in the cartridge. By changing the cartridge, a dialog according to various ages or genders becomes possible. Therefore, a cartridge according to the user can be selected, and a recognizable word and a response content can also be selected in a cartridge unit, so that a wide variety of conversation contents and the like can be provided. It is possible to make it work
[0021]
【Example】
Hereinafter, embodiments of the present invention will be described with reference to the drawings. In this embodiment, a case where the present invention is applied to a toy will be described as an example, and a case where the present invention is applied to a stuffed animal such as a dog will be described. An example in which the present invention is applied to an unspecified speaker speech recognition device capable of recognizing an unspecified speaker's speech will be described.
[0022]
(First embodiment)
FIG. 1 is a view for explaining the overall schematic configuration of the present invention. Generally, a voice recognition response processing section 10 (details will be described later) housed in a stuffed dog (apparatus main body) 1 The stuffed dog 1 includes a cartridge section 20 (which will be described later in detail) which can be detachably attached to a predetermined portion of the stuffed dog 1.
[0023]
FIG. 2 is a block diagram illustrating a configuration of the voice recognition response processing unit 10 and the cartridge unit 20 according to the first embodiment. In the first embodiment, an example will be described in which three ROMs of a standard voice feature data storage unit 21, a conversation content storage unit 22, and a response data instruction content storage unit 23 are provided on the cartridge side.
[0024]
The voice recognition response processing means 10 includes a voice input unit 11, a voice analysis unit 21, a word detection unit 13, a voice understanding conversation control unit 14, a voice synthesis unit 15, a voice output unit 16, and the like. Note that, among these components, the voice analysis unit 12, the word detection unit 13, the voice understanding conversation control unit 14, the voice synthesis unit 15, and the like are housed near the abdomen of the stuffed toy 1, for example, and the voice input unit (microphone) 11 The stuffed toy 1 is provided, for example, at the ear, and the audio output unit (speaker) 16 is provided, for example, at the mouth.
[0025]
On the other hand, the cartridge section 20 includes a standard voice feature data storage section 21, a conversation content storage section 22, and a response data instruction content storage section 23, and a cartridge mounting section (not shown) provided near the abdomen of the stuffed toy 1, for example. It is easily removable from the outside. When the cartridge unit 20 is correctly mounted on the cartridge mounting unit, the cartridge unit 20 is connected to each unit of the voice recognition response processing unit 10 and can transmit and receive signals. Specifically, the standard voice feature data storage unit 21 is connected to the word detection unit 13, the conversation content storage unit 22 is connected to the speech understanding conversation control unit 14, and the response data instruction content storage unit 23 is connected to the speech understanding It is connected to a conversation control unit 14 and a speech synthesis unit 15.
[0026]
The standard voice feature data storage unit 21 stores a standard pattern of a recognizable word (referred to as a registered word) created in advance using speech uttered by a large number (for example, about 200) of speakers for each word. Is stored (registered) in the ROM. In this example, a stuffed animal is used as an example, so the number of registered words is about 10 words. However, the present invention is not limited to this, and various words can be registered, and the number of registered words is not limited to 10 words. Further, the conversation content storage unit 22 stores contents such as what words are registered words and how to respond to each registered word. The conversation content storage unit 22 is originally provided in the voice understanding conversation control unit 14, but is provided on the cartridge unit 20 side because the registered words and the like may change depending on the cartridge. In addition, the response data instruction content storage unit 23 stores the content instructing what kind of speech synthesis output is to be made corresponding to each registered word. The content mainly indicating the voice quality is set in advance, such as a speech synthesis output or a speech synthesis output based on how a girl speaks.
[0027]
By the way, as described above, the cartridge section 20 is provided with various types of standard voice feature data, registered words, and further, response contents to each registered word, voice quality, and the like. The user to use can be selected arbitrarily. For example, in the cartridge for an infant, a plurality of registered words based on the standard audio feature data so that the infant's voice can be easily recognized are stored in the standard audio feature data storage unit 21, and what words are the registered words. Or what kind of response to each registered word is stored in the conversation content storage unit 22, and further, for each registered word, an instruction such as what kind of speech synthesis output should be made It is stored in the data instruction content storage unit 23.
[0028]
Next, the functions of the above-described units and the overall processing will be sequentially described below.
[0029]
The voice input unit 11 of the voice recognition / response processing unit 10 includes a microphone, an amplifier, a low-pass filter, an A / D converter, etc., although not shown, and converts the voice input from the microphone into an appropriate voice through an amplifier and a low-pass filter. After being converted into a waveform, the signal is converted into a digital signal (for example, 12 KHz.16 bits) by an A / D converter and output, and the output is sent to the voice analysis unit 12. The voice analysis unit 12 analyzes the frequency of the voice waveform signal sent from the voice input unit 11 at short intervals using a computing unit (CPU), and extracts a several-dimensional feature vector representing a frequency feature ( The LPC-CEPSTRUM coefficient is generally used), and a time series of this feature vector (hereinafter, referred to as an audio feature vector sequence) is output.
[0030]
Although not shown, the word detection unit 13 mainly includes an arithmetic unit (CPU) and a ROM that stores a processing program, and the word registered in the standard voice feature data storage unit 21 provided in the cartridge unit 20. Is detected in which part of the input voice and with certainty. As the word detection unit 13, a hidden Markov model (HMM) method, a DP matching method, or the like can be used. In this case, a keyword spotting processing technique using a DRNN (dynamic recurrent neural network) method (for this technique, The present applicant has already applied for a patent in Japanese Patent Application Laid-Open Nos. Hei 6-4097 and Hei 6-119476) to detect words for enabling speech recognition close to continuous speech recognition by an unspecified speaker. It is assumed that data is output.
[0031]
The specific processing of the word detection unit 13 will be briefly described with reference to FIG. The word detection unit 13 detects a word registered in the standard voice feature data storage unit 21 and at which part in the input voice and with certainty. Now, it is assumed that a voice such as "Tomorrow's weather is ..." is input from the speaker, and a voice signal as shown in FIG. 3A is output. In the phrase "Tomorrow's weather is ...", "tomorrow" and "weather" are keywords in this case, and these are one of about 10 registered words in advance, which are standard words. The pattern is stored in the voice feature data storage unit 21. If the registered words are, for example, 10 words, a signal for detecting each word is output corresponding to these 10 words (hereinafter, referred to as word 1, word 2, word 3,...). Then, from the information such as the value of the detection signal, the degree of certainty that the corresponding word exists in the input voice is detected. That is, when the word “weather” (word 1) is present in the input voice, the detection signal waiting for the signal “weather” is, as shown in FIG. Stand up in parts. Similarly, when the word "tomorrow" (word 2) is present in the input voice, the detection signal waiting for the signal "tomorrow" is changed to the "tomorrow" of the input voice as shown in FIG. Stand up at the part. In FIGS. 9B and 9C, a numerical value such as 0.9 or 0.8 is a numerical value indicating certainty (degree of approximation), and a higher numerical value such as 0.9 or 0.8 is higher. It can be said that the registered word having certainty is a recognition candidate for the input voice. That is, the registered word "tomorrow" exists with a certainty of 0.8 in the w1 portion on the time axis of the input voice signal as shown in FIG. As shown in FIG. 6B, it can be seen that the input audio signal exists at a certainty of w2 on the time axis with a certainty of 0.9.
[0032]
Further, in the example of FIG. 3, a signal for waiting for a word 3 (this word 3 is a registered word of “what time”) is also received as shown in FIG. , With a certain degree of certainty (the numerical value is about 0.6) in the portion of w2 on the time axis. As described above, when two or more registered words exist as recognition candidates at the same time with respect to the input speech signal, the word having the highest degree of approximation (a numerical value indicating the likelihood) is selected as the recognition word. A method and a correlation table indicating the correlation rules between the words are prepared in advance, and a certain recognition candidate word is determined by using a method such as selecting one of the words as a recognition word. decide. For example, if the recognition method is determined by the former method, the approximation degree corresponding to w2 on the time axis is the highest in the detection signal for detecting “weather”. It is determined that the recognition candidate for the input voice of the part is “weather”. The recognition of the input voice is performed by the voice understanding conversation control unit 14 based on these degrees of approximation.
[0033]
The voice understanding conversation control unit 14 is mainly composed of an arithmetic unit (CPU) and a ROM storing a processing program, inputs word detection data from the word detection unit 13, and based on the word detection data, Recognizing the voice (understanding the meaning of the entire input voice), referring to the conversation content storage unit 22 provided in the cartridge unit 20, determining the response content corresponding to the meaning of the input voice, Referring to the storage unit 23, a signal indicating what kind of speech synthesis output is to be sent to the speech synthesis unit (mainly composed of a CPU and a ROM) 15.
[0034]
For example, detection data (referred to as word lattices) as shown in FIGS. 3B to 3E from the word detection unit 13. The word lattices include a registered word name, an approximation degree, and a start point s and an end point e of a word. When a word or a word is included in the input speech, one or more words are determined based on the word lattice. In this example, since the input voice is "Tomorrow's weather is ...", "Tomorrow" and "Weather" will be detected, and "Tomorrow's weather" will be detected from the keywords "Tomorrow" and "Weather". .. ", And the response content corresponding to the continuous input voice is selected and output. In this case, the response content is such as "Tomorrow's weather is fine", but this is not shown here, but is performed by state detection means (temperature detection unit, air pressure detection unit, calendar Section, timekeeping section, etc.), for example, if it is information about weather, it judges the change in weather based on the situation of the change in atmospheric pressure from the atmospheric pressure detecting section, and if the atmospheric pressure is rising, it responds to it The read response content is read from the response data instruction content storage unit 23. Similarly, responses regarding temperature, time, date, and the like are possible.
[0035]
Further, the above-described speech recognition processing similar to continuous speech recognition by the keyword spotting processing can be applied not only to Japanese but also to other languages. For example, if the language to be used is English, the registered recognizable words include, for example, “good morning”, “time”, “tomorrow”, “good night”, and the like. Recognizable registered word feature data is stored in the standard voice feature data storage unit 21. And now, when the speaker asks “what time is it now”, the word “time” becomes the keyword in this case in the phrase “what time is it now”, and the word “time” is the input voice. , The detection signal waiting for the audio signal of “time” rises at the “time” portion of the input audio. When the detection data (word lattice) is input from the word detection unit 13, first, one or more words as keywords in the input voice are determined based on the word lattice. In this example, since the input voice is “what time is it now”, “time” is detected as a keyword, and based on this keyword, a continuous input of “what time is it now” is performed. Understand audio content.
[0036]
A CPU for controlling the above-described speech analysis, word detection, speech understanding conversation control, speech synthesis, and the like may be provided respectively. However, one main CPU for performing all these processes is provided. May perform the entire processing of the present invention.
[0037]
In such a configuration, for example, if the mounted cartridge is intended for an infant, and if the infant asks “good morning”, the input voice is analyzed by the voice analysis unit 12. , Sent to the word detection unit 13. Then, the word detection unit 13 performs the above-described processing by the word detection unit 13 based on the feature data stored in the standard speech feature data storage unit 21 to obtain word detection data (word lattice) for the input speech. Is output. At this time, since the cartridge section 20 is for infants, the contents of the standard voice feature data storage section 21 are standard voice feature data obtained based on the voice features of infants. Recognition becomes possible.
[0038]
Then, the speech understanding conversation control unit 14 receiving the word lattice from the word detection unit 13 refers to the conversation content storage unit 22 of the cartridge unit 20 based on the word lattice, and the input speech is “good morning”. After understanding the facts and obtaining the response contents, the response data instruction contents storage unit 23 is referred to. As a result, the speech synthesis unit 15 outputs a speech synthesis output based on the information obtained from the response data instruction content storage unit 23 and outputs the speech synthesis unit 16 as the response content. In this case, since the response content is a response to the infant, for example, "good morning" is answered in a manner of speaking for the infant.
[0039]
In this way, by attaching the cartridge section 20 arbitrarily selected by the user to the stuffed animal 1, a dialog according to the contents of the cartridge section 20 becomes possible. For example, if a cartridge for an infant as described above is attached, conversation for an infant can be performed, and if a cartridge for an elementary school student is attached, conversation corresponding to the conversation can be performed. In addition, this cartridge can be prepared variously according to various ages or genders. More specifically, detailed classification such as for boys and boys, for girls and boys, and for boys and girls in elementary school lower grades is also possible.
[0040]
Thus, even if there is only one apparatus main body (in this case, a stuffed toy), by changing the cartridge, it is possible to carry out a dialog according to various ages or genders. In this case, since the standard voice feature data storage unit 21, the conversation content storage unit 22, and the response data instruction content storage unit 23 are incorporated in the cartridge unit 20, the recognizable words and their standard voice feature data are stored in the cartridge unit. For each cartridge, the content of the response to the recognizable word and the quality of the voice can be different for each cartridge. Therefore, by increasing the variations of the types of cartridges, it is possible to perform a dialog according to various ages or genders.
[0041]
(Second embodiment)
In the first embodiment described above, the example in which the standard voice feature data storage unit 21, the conversation content storage unit 22, and the response data instruction content storage unit 23 are provided in the cartridge unit 20 has been described. Instead of having all of these three elements, for example, only the standard voice feature data storage unit 21 may be provided in the cartridge unit 20, and only the response data instruction content storage unit 23 may be provided in the cartridge unit 20. Various combinations may be considered.
[0042]
Here, when only the standard voice feature data storage section 21 is provided in the cartridge section 20 as the cartridge section 20, when the conversation content storage section 22 and the response data instruction content storage section 23 are provided in the cartridge section 20, the response The case where only the data instruction content storage unit 23 is provided in the cartridge unit 20 will be described.
[0043]
First, a case where only the standard voice feature data storage unit 21 is provided in the cartridge unit 20 will be described. FIG. 4 is a block diagram showing the configuration, and the same parts as those in FIG. 2 are denoted by the same reference numerals. That is, in this case, the conversation content storage unit 22 and the response data instruction content storage unit 23 are provided on the voice recognition response processing unit 10 side. The response content storage unit 22 may be provided in the voice understanding conversation control unit 14, but here, a case where it is provided separately is shown.
[0044]
In this way, when only the standard voice feature data storage unit 21 is provided on the cartridge unit 20 side, and the conversation content storage unit 22 and the response data instruction content storage unit 23 are on the device (stuffed toy 1) side, each registered word is The type of response is determined in advance by the device. Therefore, in this case, the registration word, word predetermined by apparatus (e.g., as described above, "good morning", "Hello", 10 about words like that as "Good night") but only, respectively Various standard voice feature data corresponding to the age and gender of the registered word can be prepared for each cartridge. For example, a cartridge in which standard voice feature data of an infant is stored for each registered word, a cartridge in which standard voice feature data of an elementary school student is stored for each registered word, and an adult female for each registered word The standard voice for each registered word for each age and gender, such as a cartridge that stores standard voice feature data for each cartridge and a cartridge that stores standard voice feature data for an adult male for each registered word A cartridge in which the characteristic data is stored is prepared.
[0045]
In this way, for example, when used by an infant, by selecting the cartridge storing the voice characteristic data for the infant and mounting it on the main body of the apparatus, the cartridge can be obtained based on the audio characteristics of the infant. Since it can be compared with the voice feature data, it can be recognized with a high recognition rate. Then, for the recognized word, a predetermined response content is voice-synthesized and output. As described above, by selecting the cartridge of the standard voice feature data according to the age and gender, the recognition rate can be significantly improved.
[0046]
Next, a case where the conversation content storage unit 22 and the response data instruction content storage unit 23 are provided in the cartridge unit 20 will be described. FIG. 5 is a block diagram showing the configuration, and the same parts as those in FIG. 2 are denoted by the same reference numerals. That is, in this case, the standard voice feature data storage unit 21 is provided on the voice recognition response processing unit 10 side, and the conversation content storage unit 22 and the response data storage unit 23 are provided on the cartridge unit 20 side.
[0047]
As described above, when only the conversation content storage unit 22 and the response data instruction content storage unit 23 are provided on the cartridge unit 20 side and the standard voice feature data storage unit 21 is provided on the voice recognition response processing unit 10 side, the registered words are predetermined words on the side (e.g., the above-described manner, "Good morning", "Hello", "Good night" 10 about words like that, etc.) only, response content and each response contents for each registered word Various voice synthesis outputs (such as voice quality) can be prepared for each cartridge. For example, what kind of response content is to be given to the word "good morning", and what kind of speech synthesis output is to be used for the response content, are determined in advance, and these are determined for each cartridge. In the conversation content storage unit 22 and the response data instruction content storage unit 23. Specifically, cartridges for infants respond to registered words such as "good morning" and "good night" with responses for infants and sound quality for infants. As a response to the registered word, for example, the response content and voice of each age and gender are matched, such as imitating the way of speaking of TV anime characters and responding with response content for elementary school students. A variety of cartridges that respond by quality are prepared.
[0048]
In this way, for example, when an elementary school student uses the above-mentioned cartridge for the elementary school student and attaches it to the device, when the elementary school student speaks some registered words, the television as described above is used. This makes it possible to return the response set in the cartridge with the quality of the voice imitating the manner of speaking the animation character and the cartridge.
[0049]
Next, a case where only the response data instruction content storage unit 23 is provided on the cartridge unit 20 side will be described. FIG. 6 is a block diagram showing the configuration, and the same parts as those in FIG. 2 are denoted by the same reference numerals. That is, in this case, the standard voice feature data storage unit 21 and the conversation content storage unit 22 are provided on the voice recognition response processing unit 10 side of the apparatus main body, and only the response data instruction content storage unit 23 is provided on the cartridge unit 20 side. ing.
[0050]
When only the response data instruction content storage unit 23 is provided on the cartridge unit 20 side and the conversation content storage unit 22 and the standard voice feature data storage unit 21 are provided on the voice recognition response processing unit 10 side, the registered word is a predetermined word on the side (for example, as the "good morning", "Hello", 10 about words like that, such as "Good night") is only, also, the response content is basically for these registration word The content is preset on the device side, but it is possible to set various types of speech synthesis output for the response content for each cartridge. For example, the response content to the word "good morning" is determined in advance as to what kind of voice quality is to be output, and these are stored in the response data instruction content storage unit 23 for each cartridge. Specifically, in the case of a cartridge for an infant, for a variety of registered words such as "good morning" and "good night", a response such as a mother's voice is made to the registered words. Instructions or instructions that give a response that resembles the voice of a TV anime character for toddlers are stored.As a cartridge for elementary school students, the response to registered words is similar to how a TV anime character for elementary school students speaks. For example, various cartridges in which the content for instructing what kind of speech synthesis output is to be output for each registered word according to each age and gender are stored. Have it ready. At this time, the contents of each response to each registered word are basically the contents set in advance.
[0051]
In this way, for example, when used by elementary school children, by selecting a cartridge for elementary school children as described above and attaching it to the device, if the elementary school student speaks any registered word, the response content is basically Although the content is set in advance, it is possible to return a response in a voice imitating the manner of speaking a TV anime character for elementary school children as described above.
[0052]
As described above, according to the present invention, the ROM parts such as the standard voice feature data storage unit 21, the conversation content storage unit 22, and the response data instruction content storage unit 23 are of a cartridge type, and standard ROMs corresponding to the age and gender of the user are used. Various cartridges having voice characteristic data, response contents, and the like are prepared, so that the user can arbitrarily select the cartridge. Therefore, even if only one recognition response device is used, by replacing the cartridge, it can be used by a wide range of people, and voice recognition and conversation can be performed according to the user.
[0053]
In each of the above embodiments, an example in which the present invention is applied to a stuffed toy as a toy has been described, but the present invention is not limited to a stuffed toy. Of course, the present invention can be applied not only to toys, but also to game machines, various electronic devices used daily, and the like, and the applicable range is considered to be extremely wide.
[0054]
【The invention's effect】
As described above, according to the first aspect of the present invention, the voice recognition interactive device according to the first aspect includes a storage unit configured to store storage contents set in advance for performing voice recognition and responding to the recognized voice. The cartridge is detachably mounted to the cartridge, and the cartridge is mounted on the main body of the apparatus, so that the contents of the response to the input voice are output based on the data stored in the cartridge. Therefore, even if only one apparatus main body is used, by changing the cartridge, it is possible to perform a dialog according to various ages or genders. Therefore, for example, when the present invention is applied to a toy or the like, it is possible to select a cartridge according to the child's growth, and also to select various words and recognizable words by selecting the cartridge. Since it is possible, even a single toy is less likely to get tired on the way, can be used for a long period of time, and can be used widely without regard to age and gender. In addition to the application of not only toys but also electronic devices, by selecting a cartridge suitable for the user, a wide variety of conversations can be provided, and various operations can be performed accordingly. The effect becomes extremely large.
[0055]
According to the second aspect, the standard voice feature data storage means for storing the standard voice feature data of the pre-registered recognizable words is provided on the cartridge side. Various standard voice feature data corresponding to each cartridge can be prepared, and by selecting and using a cartridge adapted to the user, the recognition rate can be significantly improved.
[0056]
According to the third aspect, a conversation content storage means for storing a response content corresponding to a pre-registered recognizable word, and a response for storing an instruction content for instructing what kind of speech synthesis output is to be generated. Since the data instruction content storage means is provided on the cartridge side, it is possible to prepare various cartridges having response contents according to their respective ages and genders. When a word is spoken, it is possible to respond with children's response contents and children's voice.
[0057]
According to the fourth aspect, the response data instruction content storage means for storing the instruction content for instructing what kind of speech synthesis output is to be provided is provided on the cartridge side. In accordance with the above, various cartridges in which contents indicating what kind of speech synthesis output is to be output for each registered word can be prepared. For example, if a cartridge for elementary school students is selected, some registered words can be prepared. When speaking, the response content is basically the content set in advance, but it is possible to return a response with a voice that imitates the way of speaking the TV anime character for elementary school children as described above. Become.
[0058]
According to the fifth aspect, a standard voice feature data storage unit that stores standard voice feature data of a pre-registered recognizable word, and a conversation that stores a response content corresponding to the registered recognizable word. Content storage means and response data indicating the type of speech synthesis output to be generated are provided on the cartridge side, so that standard voice feature data corresponding to the age and gender of each registered word is provided. Various cartridges can be prepared for each cartridge, and by selecting and using a cartridge that is suitable for the user, the recognition rate can be significantly improved. Also, recognizable registered words can be set for each cartridge. The number of conversation variations can be greatly increased. Further, it is possible to prepare various cartridges having response contents and voice synthesis output corresponding to each age and gender. As a result, even with a single apparatus main body, by changing the cartridge, a wide variety of dialogues according to various ages or genders can be performed.
[0059]
According to the speech recognition dialogue processing method of the present invention, it is possible to removably attach storage contents set in advance for performing speech recognition and responding to the recognized speech to the apparatus main body. When the cartridge is attached to the main body of the apparatus, the contents of the response to the input voice are generated based on the data stored in the cartridge. Even so, by changing the cartridge, a dialog according to various ages or genders becomes possible. Therefore, for example, when the present invention is applied to a toy or the like, it is possible to select a cartridge according to the child's growth, and also to select various words and recognizable words by selecting the cartridge. As a result, even a single toy is less likely to get tired on the way, can be used for a long period of time, and can be used widely regardless of age and gender. In addition to the application of not only toys but also electronic devices, by selecting a cartridge suitable for the user, a wide variety of conversations can be provided, and various operations can be performed accordingly. And the scope of application will be extremely wide.
[0060]
According to the present invention, the standard voice feature data of the recognizable word registered in advance is stored in the cartridge, so that the standard voice feature data corresponding to the age and gender of each registered word is stored in the cartridge. A variety of cartridges can be prepared for each case, and by selecting and using a cartridge suitable for the user, the recognition rate can be significantly improved.
[0061]
According to the eighth aspect, the response content corresponding to the recognizable word registered in advance and the instruction content for instructing what kind of speech synthesis output is generated are stored in the cartridge side. Various cartridges with response contents according to their age and gender can be prepared. For example, if a cartridge for children is selected, if a registered word is spoken, response contents for children and children It is possible to respond with a voice for the user.
[0062]
According to the ninth aspect, the instruction content for instructing what kind of speech synthesis output is to be generated is stored in the cartridge side, so that registration is performed according to each age and gender. For each word, various cartridges can be prepared in which contents indicating what kind of speech synthesis output is to be stored. For example, if a cartridge for elementary school students is selected, if a registered word is spoken, the response content Is basically a preset content, but it is possible to return a response with a voice that imitates the way of speaking a TV anime character for elementary school children as described above.
[0063]
According to the tenth aspect, standard voice feature data of a pre-registered recognizable word, response contents corresponding to the registered recognizable word, and what kind of voice synthesis output is to be generated are instructed. Since the response data instruction contents are stored in the cartridge side, standard voice characteristic data corresponding to the age and gender of each registered word can be prepared variously for each cartridge, and a cartridge suitable for the user can be selected. By using this, the recognition rate can be greatly improved, and recognizable registered words can be set for each cartridge, so that conversation variations can be greatly increased. Further, it is possible to prepare various cartridges having response contents and voice synthesis output corresponding to each age and gender. As a result, even with a single apparatus main body, by changing the cartridge, a wide variety of dialogues according to various ages or genders can be performed.
[Brief description of the drawings]
FIG. 1 is a diagram illustrating an outline of the present invention.
FIG. 2 is a block diagram illustrating a first embodiment of the present invention.
FIG. 3 is a view for explaining word detection processing by a word detection unit and speech recognition processing by a speech understanding conversation control unit;
FIG. 4 is a block diagram illustrating a second embodiment (part 1) of the present invention.
FIG. 5 is a block diagram for explaining a second embodiment (No. 2) of the present invention.
FIG. 6 is a block diagram illustrating a second embodiment (3) of the present invention.
[Explanation of symbols]
1 ... stuffed animal (device main body) 10 ... voice recognition response processing unit
11: voice input unit 12: voice analysis unit
13: Word detection unit 14: Voice understanding conversation control unit
15: Voice synthesis unit 16: Voice output unit
20: cartridge section 21: standard voice feature data storage section
22: conversation content storage unit 23: response data instruction content storage unit

Claims (10)

The input voice is analyzed to generate voice feature data, and the voice feature data is compared with standard voice feature data of a pre-registered recognizable word to output word detection data. In the voice recognition dialogue device that receives and understands the meaning of the input voice, determines and outputs the corresponding response content,
A storage means for storing data set in advance for performing voice recognition, data set in advance for outputting a response to the recognized voice, and the like is provided on a cartridge side detachably mountable to the apparatus main body. When the cartridge is mounted on the apparatus main body, it is connected to a voice recognition response processing unit provided on the apparatus main body side, and outputs a response content to an input voice based on data stored in the cartridge. A speech recognition dialogue device characterized by the following. The storage means provided on the cartridge side is a standard voice feature data storage means for storing standard voice feature data for pre-registered recognizable words,
On the device main body side, voice input means for inputting voice, voice analysis means for analyzing voice input by the voice input means to generate voice feature data, and voice feature data from the voice analysis means. A word detecting means for inputting and outputting word detection data for the input voice based on the storage content of the standard voice feature data storage means provided on the cartridge side; and receiving word detection data from the word detecting means. Speech understanding conversation control means for understanding the meaning of the input speech and determining a response content corresponding thereto, and response data instruction content for generating a speech synthesis output based on the response content determined by the speech understanding conversation control means Storage means, and voice synthesis based on the content of the response data instruction content storage means for the response content determined by the voice understanding conversation control means. And speech synthesis means for generating a force, the voice recognition interaction device according to claim 1, characterized in that and an audio output means for outputting the speech synthesized output to the outside from the speech synthesis means. The storage means provided on the cartridge side includes a conversation content storage means for storing a response content corresponding to a pre-registered recognizable word, and an instruction content for instructing what kind of speech synthesis output is to be generated. Response data instruction content storage means for storing,
The apparatus main body includes a voice input unit for inputting voice, a voice analysis unit for analyzing voice input by the voice input unit to generate voice feature data, and a standardized recognizable word. Standard voice feature data storage means for storing voice feature data; word detection means for outputting word detection data for an input voice based on the storage contents of the standard voice feature data storage means; and word detection from the word detection means By receiving the data, the meaning of the input voice is understood, and the corresponding response content is determined with reference to the conversation content storage means provided on the cartridge side. For the determined response content, a voice for generating a voice synthesis output based on the content of the response data instruction content storage means provided on the cartridge side And forming means, the speech recognition interaction device according to claim 1, characterized in that and an audio output means for outputting the speech synthesized output to the outside from the speech synthesis means. The storage unit provided on the cartridge side is a response data instruction content storage unit that stores instruction content for instructing what kind of speech synthesis output is to be generated,
The apparatus main body includes a voice input unit for inputting voice, a voice analysis unit for analyzing voice input by the voice input unit to generate voice feature data, and a standardized recognizable word. Standard voice feature data storage means for storing voice feature data; word detection means for outputting word detection data for an input voice based on the storage contents of the standard voice feature data storage means; and word detection from the word detection means Speech comprehension / conversation control means for receiving the data to understand the meaning of the input speech and determining the response content corresponding thereto, and a response provided on the cartridge side for the response content determined by the speech understanding / conversation control means. Speech synthesis means for generating a speech synthesis output based on the contents of the data instruction content storage means, and outputting the speech synthesis output from the speech synthesis means to the outside Speech recognition interaction device according to claim 1, characterized in that and an audio output unit that. The storage means provided on the cartridge side includes a standard voice feature data storage means for storing standard voice feature data for pre-registered recognizable words, and a conversation for storing response contents corresponding to the registered recognizable words. Content storage means, response data instruction content storage means for instructing what kind of speech synthesis output is to be generated,
On the device main body side, voice input means for inputting voice, voice analysis means for analyzing voice input by the voice input means to generate voice feature data, and voice feature data from the voice analysis means. A word detecting means for inputting and outputting word detection data for the input voice based on the stored contents of the standard voice feature data storage means provided on the cartridge side; and receiving and inputting the word detection data from the word detecting means. Voice understanding conversation control means for understanding the meaning of voice and determining the corresponding response content with reference to the conversation content storage unit provided on the cartridge side; and response content determined by the voice understanding conversation control means A voice synthesizing means for generating a voice synthesizing output based on the contents of the response data instruction content storing means provided on the cartridge side; Speech recognition interaction device according to claim 1, characterized in that and an audio output means for outputting the speech synthesized output to the outside from the synthesis means. The input voice is analyzed to generate voice feature data, and the voice feature data is compared with standard voice feature data of a pre-registered recognizable word to output word detection data. Receiving, understanding the meaning of the input voice, and determining and outputting a response corresponding to the input voice.
Data set in advance to perform voice recognition, data set in advance to output a response to the recognized voice, and the like are written in storage means on the cartridge side that can be detachably attached to the apparatus main body, A speech recognition dialogue processing method characterized by outputting a response to an input speech based on data stored in the cartridge when the cartridge is mounted on the apparatus main body. The storage content stored on the cartridge side is standard voice feature data for a pre-registered recognizable word,
On the side of the device main body, a voice analysis step of analyzing voice input by voice input means to generate voice characteristic data, and voice characteristic data from this voice analysis step are input and stored in the cartridge side. A word detection step of outputting word detection data for the input voice based on the standard voice feature data; receiving the word detection data from the word detection step to understand the meaning of the input voice and determine the corresponding response content A voice comprehension conversation control step; a speech synthesis step for performing speech synthesis based on response data instructing what kind of speech synthesis output is to be given to the response content determined in the speech comprehension conversation control step; 7. A speech recognition dialogue processing method according to claim 6, further comprising a speech output step of outputting a speech synthesis output from the speech synthesis step to the outside. The storage contents stored on the cartridge side are response contents corresponding to a pre-registered recognizable word and response data instruction contents for instructing what kind of speech synthesis output is to be generated,
On the device body side, a voice analysis step of analyzing voice input by voice input means to generate voice feature data, and a standard voice feature data of a pre-registered recognizable word, A word detection step of outputting word detection data, receiving the word detection data from the word detection step, understanding the meaning of the input voice, and referring to the corresponding response content to the conversation content stored in the cartridge side. A voice comprehension conversation control step, which is determined by the voice comprehension conversation control step, and a speech synthesis step for generating a speech synthesis output based on the response data instruction content stored on the cartridge side for the response content determined by the voice understanding conversation control step 7. The speech recognition dialogue processing method according to claim 6, further comprising a speech output step of outputting a speech synthesis output from the speech synthesis step to the outside. The storage content stored on the cartridge side is an instruction content for instructing what kind of speech synthesis output is to be generated,
On the device body side, a voice analysis step of analyzing voice input by voice input means to generate voice feature data, and a standard voice feature data of a pre-registered recognizable word, A word detection step of outputting word detection data, a speech understanding conversation control step of receiving the word detection data from the word detection step, understanding the meaning of the input speech, and determining a response content corresponding thereto; In response to the response content determined by the control step, a voice synthesis step for generating a voice synthesis output based on the response data instruction content stored in the cartridge side, and outputting the voice synthesis output from the voice synthesis step to the outside. 7. The speech recognition dialogue processing method according to claim 6, further comprising a speech output step. The storage contents stored on the cartridge side include standard voice feature data for pre-registered recognizable words, response contents corresponding to the registered recognizable words, and what kind of voice synthesis output is to be generated. The response data to be instructed,
On the side of the device main body, a voice analysis step of analyzing voice input by voice input means to generate voice characteristic data, and voice characteristic data from this voice analysis step are input and stored in the cartridge side. Based on the standard voice feature data, a word detection step of outputting word detection data for the input voice, receiving the word detection data from the word detection step to understand the meaning of the input voice, A voice understanding conversation control step which is determined by referring to the conversation content stored in the cartridge side, and a response content determined in the voice understanding conversation control step is based on the response data instruction content stored in the cartridge side. A voice synthesis step of generating a voice synthesis output, and a voice output step of outputting the voice synthesis output from the voice synthesis step to the outside. 6. speech recognition dialogue processing method according to symptoms.

Images