JP4466171B2 - Information retrieval device - Google Patents

Description

  The present invention relates to an information search apparatus and method for searching for information by voice.

  The phone number and the voice tag associated with the phone number are stored in the memory, and the voice tag stored in the memory is sequentially presented to the user in the order of registration, so that the user corresponds to the destination. Patent Document 1 discloses a radiotelephone in which a telephone number associated with a selected voice tag is referred to by selecting the voice tag at the time when the voice tag is presented.

Japanese Patent Laid-Open No. 11-103338

  However, in the conventional apparatus that presents the voice tags stored in the memory to the user sequentially in the order of storage, when the amount of data stored in the memory is large, the voice tag corresponding to the destination is located in the rear of the memory. In the case where it is stored in the network, there has been a problem that it takes time until the destination voice tag is presented.

  The present invention stores a plurality of pieces of information and also stores abbreviated audio data (hereinafter referred to as audio tags) of information attached to each information in association with the information. In an information search device and method for searching for information corresponding to a voice tag for which a matching result is obtained from stored information by comparing an abbreviation with a voice tag of each information, a search history of each information is stored. In addition, a list of abbreviations of a plurality of information is presented by voice in an order corresponding to a search history of each information.

  According to the present invention, it is highly possible to shorten the search time until the required information is reached, and convenience can be improved.

  FIG. 1 shows an embodiment of an information search apparatus according to the present invention, and is a block diagram when the information search apparatus is applied to a hands-free telephone system for a vehicle. A hands-free telephone system 100 includes a microphone 101 that inputs a driver's (user) speech, a speaker 102 that outputs sound, a start switch 103 that allows the driver to start the system, and a clock 104 that measures the date and time. And a mobile phone 105 and an information search controller 106. In this vehicular hands-free telephone system, when a driver talks with a calling party, he / she speaks through the microphone 101 and listens to the calling party's voice output from the speaker 102.

  The information retrieval controller 106 includes a telephone directory data storage memory 106a, a telephone directory data update unit 106b, a voice tag output control unit 106c, a voice recognition dictionary memory 106d, a voice recognition unit 106e, and a voice tag-phone number conversion. 106f and a call control unit 106g. The telephone directory data update unit 106b, the voice tag output control unit 106c, the voice recognition unit 106e, the voice tag-phone number conversion unit 106f, and the call control unit 106g are configured in the form of a microcomputer software. Further, a non-volatile memory such as an EEPROM or a backup RAM is used for the telephone directory data storage memory 106a and the voice recognition dictionary memory 106d.

  The telephone book data storage memory 106a stores data in the format shown in FIG. Each record is given a uniquely identifiable identification number 2a in the order of registration of the records, and a telephone number 2b is stored as telephone directory data. The voice tag 2c is abbreviated voice data assigned to each telephone number in order to easily and easily read out the telephone number from the telephone directory, and can be freely set by the user when registering the telephone number. For example, in the example shown in FIG. 2, the voice tag 2 c “home” is set for the telephone number having the identification number 2 a of 1. In FIG. 2, the data (“home”, “company”) stored in the audio tag 2c is represented by characters for convenience of explanation, but is actually stored as audio data represented by a waveform. The

  The last access date and time 2d is updated when the driver refers to the record to call the telephone number, and the last (latest) reference date and time of the record is always stored. The access count counter 2e is incremented by 1 when the driver refers to the record, and stores the record reference count by the driver. That is, when the driver calls the telephone number 2b whose voice tag 2c is “home” as a call destination, the current date and time acquired from the clock 104 is stored in the last access date and time 2d, and the access number counter 2e is set to 1. Is added.

  The phone book data update unit 106b controls registration and update of data in the above-described phone book data storage memory 106a. That is, when registration of phone book data is instructed by the driver, the phone book data is registered in the format shown in FIG. 2 based on the input contents from the driver. At the time of registration, the registration date is stored as the last access date 2d, and 0 is stored as the initial value in the access counter 2e. When the data in the telephone directory data storage memory 106a is referred to by the driver, the last access date 2d and the access count counter 2e are updated as described above. The telephone directory data updating unit 106b also registers the voice tag 2c registered in the telephone directory data storage memory 106a in a voice recognition dictionary memory 106d described later so that the voice data can be collated when the driver speaks. To do.

  The voice tag output control unit 106c presents a list of voice tags 2c stored in the telephone directory data storage memory 106a to the driver via the speaker 102 based on an instruction from the driver. When the driver calls the target destination, the driver utters the voice tag 2c associated with the telephone number 2b. However, the driver may forget what abbreviation of the destination voice tag 2c is registered. In such a case, in order to confirm the contents of the voice tag 2c stored in the telephone directory data storage memory 106a, the driver gives an instruction to output a list of the voice tag 2c.

  At this time, the driver can specify the output order of the list, and in the present embodiment, the driver specifies any one of “registration order”, “access frequency order”, and “access date / time order”. When the output order is designated by the driver, the voice tag output control unit 106c uses the identification number 2a, the access count counter 2e, and the last access date and time 2d as sort information, and the voice tag 2c is voiced in the designated output order. Output.

  When “registration order” is designated as the output order, the voice tag output control unit 106c has a smaller order of registration in the phone book data storage memory 106a, that is, the identification number 2a stored in the phone book data storage memory 106a is smaller. A list of audio tags 2c is output in audio from the record. As a result, it is possible to preferentially output the audio tag 2c which has been registered quite a long time ago and is now likely to be forgotten.

  When “access frequency order” is designated, the list of the audio tags 2c is output as a voice in the order from the lowest access frequency, that is, from the record with the smallest access count counter 2e stored in the telephone directory data storage memory 106a. As a result, it is possible to preferentially output the audio tag 2c that the driver has not used much in the past and is likely to have forgotten.

  When “in order of access date / time” is designated, the list of audio tags 2c is output in audio order from the oldest access time, that is, from the record with the oldest last access date / time 2d. As a result, it is possible to preferentially output the audio tag 2c that has not been used recently by the driver and is likely to be forgotten.

  The voice recognition dictionary memory 106d stores voice data on which the information search controller 106 waits for the driver's speech. As described above, the information search controller 106 according to the present embodiment needs to wait for utterances by the driver of the voice tag 2c and recognize them, so all the voices stored in the telephone directory data storage memory 106a. The tag 2c is stored in the speech recognition dictionary memory 106d. The voice recognition unit 106e performs matching processing between various commands and voice tags 2c uttered by the driver and input via the microphone 101, and various commands and voice tags 2c stored in the voice recognition dictionary memory 106d. Recognize the utterance content of the person.

  The voice tag-phone number conversion unit 106f refers to the phone book data storage memory 106a, acquires the phone number 2b corresponding to the voice tag 2c that has been voice-recognized by the voice recognition unit 106e, and transfers the voice tag 2c to the phone number 2b. Perform the conversion. That is, the telephone number 2b corresponding to the voice tag 2c that has been voice-recognized is searched. Thus, if the driver utters the voice tag 2c and designates the destination, the destination telephone number 2b can be referred to.

  The call control unit 106g controls a call by the mobile phone 105. That is, when making a call, the mobile phone 105 is controlled to make a call to the callee telephone number converted by the voice tag-phone number conversion unit 106f. When receiving a call, the mobile phone 105 is controlled to start a call with the other party in accordance with a call start instruction from the driver.

  FIG. 3 is a flowchart showing a process of newly registering phone book data in the phone book data storage memory 106a. When the activation switch 103 is on in step S10 and the information search controller 106 is in a voice input standby state in step S20, a phone book registration command “phone book registration” is uttered from the driver in step S30, and the microphone If it is determined that voice data has been input via 101, the process proceeds to step S40.

  In step S40, the voice recognition unit 106e performs a matching process on the voice data of “phone book registration” uttered by the driver and the command stored in the voice recognition dictionary memory 106d, and the command with the highest degree of matching is voiced. Determined as a recognition result. When the phone book registration command “phone book registration” uttered by the driver is normally recognized by the voice recognition unit 106e, the information search controller 106 starts processing for phone book registration.

  In step S <b> 50, a guidance voice that prompts the driver to register in the telephone directory is output via the speaker 102. In the present embodiment, first, in order to prompt the driver to speak the voice tag 2c, a guidance voice is output as "Please input the voice tag". Thereafter, in step S60, the information search controller 106 enters a voice input standby state.

  In step S70, when the driver speaks the voice tag 2c, for example, "home", the process proceeds to step S80. In step S80, the telephone directory data update unit 106b assigns the identification number 2a and adds a new registration record in the telephone directory data storage memory 106a. In step S90, the telephone directory data update unit 106b stores the voice tag 2c uttered by the driver in step S70 in the voice tag 2c of the new registration record.

  In step S100, the telephone directory data update unit 106b stores the voice tag 2c of the new registration record in the voice recognition dictionary memory 106d. In step S110, the telephone directory data update unit 106b stores the current date and time acquired from the clock 104 in the last access date and time 2d of the new registration record. In step S120, the telephone directory data update unit 106b stores 0 as an initial value in the access count counter 2e of the new registration record.

  In step S <b> 130, a guidance voice that prompts the driver to input a telephone number, for example, “Please input a telephone number”, is output via the speaker 102. Thereafter, in step S140, the information search controller 106 enters a voice input standby state. In step S150, when the telephone number is uttered by the driver, the process proceeds to step S160.

  In step S160, the speech recognition unit 106e performs a matching process on the number of the telephone number uttered by the driver and the number stored in the speech recognition dictionary memory 106d, and uses the number with the highest degree of matching as the speech recognition result. decide. When the telephone number spoken by the driver is normally recognized by the voice recognition unit 106e, the process proceeds to step S170.

  In step S170, the recognized telephone number is stored in telephone number 2b of the new registration record. Thereafter, the process ends. Through the above processing, a new record is registered in the telephone directory data storage memory 106a in the data format shown in FIG.

  Next, a process for calling an arbitrary destination with reference to the phone book data stored in the phone book data storage memory 106a will be described. FIG. 4 is a flowchart showing a process in which the driver speaks the voice tag 2c of an arbitrary destination and refers to the telephone number of the destination from the telephone directory data storage memory 106a.

  When the activation switch 103 is on in step S210 and the information search controller 106 is in a voice input standby state in step S220, the driver issues a phone book reference command “phone book reference” in step S230, and the microphone If it is determined that the voice data is input via 101, the process proceeds to step S240.

  In step S240, the voice recognition unit 106e performs a matching process on the voice data of “refer to the phone book” uttered by the driver and the command stored in the voice recognition dictionary memory 106d, and the command with the highest degree of matching is voiced. Determined as a recognition result. When the phone book registration command “phone book reference” uttered by the driver is normally recognized by the voice recognition unit 106e, the information search controller 106 starts processing for phone book reference.

  In step S250, in order to prompt the driver to speak the voice tag 2c to be referred to, after outputting a guidance voice such as “Please input voice tag”, the information search controller 106 waits for voice input in step S260. It becomes a state. In step S270, if the driver utters a voice tag 2c, for example, "home" here, the process proceeds to step S280, and the voice recognition unit 106e stores the voice tag spoken by the driver and the voice recognition dictionary memory 106d. The voice tag 2c thus matched is subjected to a matching process, and the voice tag 2c having the highest degree of matching is determined as a voice recognition result.

  In step S290, the voice tag-phone number conversion unit 106f reads the telephone number 2b corresponding to the voice tag 2c spoken by the driver with reference to the telephone directory data storage memory 106a, and from the voice tag 2c to the telephone number 2b. Perform the conversion. That is, the telephone number 2b is acquired by referring to the record in which the voice tag 2c is “home”.

  In step S300, the phone book data update unit 106b updates the last access date 2d of the record whose voice tag 2c in the phone book data storage memory 106a is “home” with the current date and time acquired from the clock 104. In step S310, the telephone directory data update unit 106b adds 1 to the count value of the access count counter 2e of the record whose voice tag 2c is “home”.

  In step S320, the telephone directory data update unit 106b determines whether or not an overflow has occurred in the count value of the record as a result of adding 1 to the count value of the access count counter 2e. When it is determined that an overflow has occurred in the count value, the telephone directory data update unit 106b returns the count value of the access count counter 2e of the record whose voice tag 2c is “home” to the value before adding 1, The count value of the access count counter 2e of the record is halved. Thereafter, 1 is added again to the count value of the access counter e of the record in which the voice tag 2c is “home”. As a result, overflow of digits can be avoided while maintaining the magnitude relationship of the count values between the records. If the count value is not divisible by 2, the decimal part is rounded down.

  If it is determined that the count value has not overflowed, step S330 is skipped. In step S340, in order to confirm to the driver whether or not the voice tag 2c uttered by the speaker, in this case, the telephone number 2b of “home” is to be transmitted as a call destination, for example, “Do you want to call home? Is output via the speaker 102. Thereafter, in step S350, the information search controller 106 enters a voice input standby state.

  In step S360, a response to the voice guidance for confirming the outgoing call is confirmed. If there is a response from the driver, the process proceeds to step S370, where the speech recognition unit 106e performs a matching process between the response content of the driver and “yes”, “no” stored in the speech recognition dictionary memory 106d, The standby word with the highest degree of matching is determined as the speech recognition result.

  In step S380, the driver's response content is confirmed. If the answer is “Yes” to agree to the transmission to the destination, the process proceeds to step S390. In step S390, the communication control unit 106g controls the mobile phone 105 to make a call to the destination. That is, make a phone call. If the response content of the speaker is “No” not agreeing to the destination, the process is terminated without making a call.

  Next, processing for outputting a list of voice tags 2c of all records from the phone book data stored in the phone book data storage memory 106a will be described. FIG. 5 is a flowchart showing a process of outputting a list of voice tags 2c from the phone book data stored in the phone book data storage memory 106a.

  When the activation switch 103 is on in step S410 and the information search controller 106 is in a voice input standby state in step S420, a command “list output” for outputting a list of voice tags 2c from the driver in step S430 is issued. If it is determined that voice data is input through the microphone 101, the process proceeds to step S440.

  In step S440, the voice recognition unit 106e performs a matching process on the voice data of “list output” uttered by the driver and the command stored in the voice recognition dictionary memory 106d, and recognizes the command having the highest degree of coincidence. Determine as a result. When the phone book registration command “list output” uttered by the driver is normally recognized by the voice recognition unit 106e, the information search controller 106 starts processing for referring to the phone book.

  In step S450, in order to prompt the driver to speak the output order of the list of the voice tag 2c, a voice guidance “Please input the list output order” is output. Thereafter, in step S460, the information search controller 106 enters a voice input standby state.

  In step S470, if any one of “registration order”, “access frequency order”, and “access date / time order” is uttered as the list output order by the driver, the process proceeds to step S480. In step S480, the voice recognition unit 106e performs a matching process on the voice data in the list output order uttered by the driver and the voice data in the list output order stored in the voice recognition dictionary memory 106d, and has the highest degree of matching. The list output order is determined as a speech recognition result.

  In step S490, the process branches according to the list output order. If the list output order is "registration order", the process proceeds to step S500, where the voice tag output control unit 106c lists the voice tags 2c in order from the record with the smallest identification number 2a stored in the telephone directory data storage memory 106a. Is output as audio. As a result, it is possible to preferentially output the audio tag 2c which has been registered quite a long time ago and is now likely to be forgotten.

  If the list output order is “access frequency order”, the process advances to step S510, and the voice tag output control unit 106c determines the access frequency counter stored in the phone book data storage memory 106a in the order of low access frequency. A list of audio tags 2c is output in audio from the records with the smallest 2e. As a result, it is possible to preferentially output the audio tag 2c that the driver has not used much in the past and is likely to have forgotten.

  When the list output order is “access date order”, the process proceeds to step S520, and the audio tag output control unit 106c displays the list of the audio tags 2c in order from the oldest access date, that is, the record with the last access date 2d from the oldest. Output audio. As a result, it is possible to preferentially output the audio tag 2c that has not been used recently by the driver and is likely to be forgotten.

As described above, according to the present embodiment, the following operational effects can be obtained.
(1) The driver designates “registration order”, “access frequency order”, or “access date order” as the voice tag list output order, and the voice tag list is voiced based on the designated output order. I decided to output it. As a result, the output order of the list of voice tags can be specified, and the possibility of promptly presenting the voice tag that the driver is looking for increases.
(2) When “access frequency order” is designated as the output order of the audio tags, the audio tag list is output in order from the record with the lowest access frequency. As a result, it is possible to present with priority from voice tags that have not been used much in the past and are likely to be forgotten.

(3) When “order of access date / time” is designated as the output order of audio tags, the audio tag list is output in audio from the records with the oldest access date / time. Thereby, it can present preferentially from the voice tag which has not been used recently by the driver and is likely to be forgotten.
(4) As a result of adding 1 to the count value of the access count counter 2e, if an overflow occurs in the count value of the record, the access count counter 2e of all records stored in the telephone directory data storage memory 106a After the count value was halved, 1 was added again. Thus, overflow can be avoided while maintaining the magnitude relationship of the count values between records, and the processing can be continued.

  When the count value is not divisible by 2, the decimal part is rounded down. However, the decimal part may be rounded up or rounded off. Moreover, although the example in which the count value of the access count counter 2e of all the records is halved to avoid the overflow is shown, the overflow may be avoided by, for example, 1/3, 1/4, and the like.

  In the above-described embodiment, the driver designates one of the output orders from “registration order”, “access frequency order”, and “access date order” as the list output order of the audio tag 2c. However, at least any two of these output orders may be designated.

  Further, in the above-described embodiment, when a new record is registered in the telephone directory data storage memory 106a, the telephone number 2b and the voice other tag 2c are inputted by voice. However, input may be performed by a manual input device (not shown) such as an operation switch. Further, a switch for switching between voice input and manual input may be provided, and the driver may switch to one of the methods as necessary to perform input. When inputting a voice tag by a method other than voice input, when the voice tag is stored in the telephone directory data storage memory 106a, the voice tag is converted into voice data and stored as voice data.

  Furthermore, in the above-described embodiment, the list of the audio tags 2 c is output as audio via the speaker 102. However, the present invention is not limited to this and may be modified as follows. For example, a monitor that displays characters and images may be provided, and a list of voice tags may be output as characters on the monitor screen. In this case, the voice tag 2c may be stored as character data in the telephone directory data storage memory 106a, or the voice tag 2c may be converted into character data and output.

  Furthermore, in the above-described embodiment, an example in which the voice recognition device according to the present invention is applied to a hands-free telephone system for a vehicle is shown. However, the present invention may be applied to, for example, an audio system or a navigation system other than the hands-free telephone system, or may be applied to other information devices or devices other than those for vehicles. In this case, the information associated with the voice tag may be information frequently called by the user in each system and device, for example, destination information in the navigation system, instead of the telephone number.

  The correspondence between the constituent elements of the claims and the constituent elements of the embodiment is as follows. That is, the telephone directory data storage memory 106a and the voice recognition dictionary memory 106d are information storage means, the microphone 101 is a voice input means, the voice tag-phone number conversion unit 106f is an information search means, and the voice tag output control unit 106c is As the abbreviation presenting means, the voice recognition unit 106e constitutes a list order selecting means. In addition, as long as the characteristic function of this invention is not impaired, each component is not limited to the said structure.

It is a block diagram at the time of applying the speech recognition apparatus by this invention to a hands-free telephone system. It is a figure which shows the specific example of the data structure stored in the telephone directory data storage memory 106a. It is the flowchart figure which showed the flow of the telephone directory registration process when a driver | operator registers telephone directory data in the telephone directory data storage memory 106a. It is the flowchart figure which showed the flow of the telephone directory reference process when a driver | operator utters the audio | voice tag 2c and references the telephone number of an applicable transmission destination from the telephone directory data storage memory 106a. It is the flowchart figure which showed the flow of the list output process when outputting the list | wrist of the audio | voice data of the audio | voice tag 2c from the telephone directory data stored in the telephone directory data storage memory 106a.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Hands-free telephone system 101 Microphone 102 Speaker 103 Starting switch 104 Clock 105 Mobile phone 106 Information retrieval controller 106a Telephone book data storage memory 106b Telephone book data update part 106c Voice tag output control part 106d Voice recognition dictionary memory 106e Voice recognition part 106f Voice tag-phone number conversion unit 106g Call control unit

Claims (4)

Information storage means for storing a plurality of pieces of information and storing voice data (hereinafter referred to as voice tags) of information abbreviated to each piece of information in association with each piece of information;
Voice input means for inputting voice;
The abbreviation of the information input by the voice input unit is collated with the voice tag of each information, and information corresponding to the voice tag for which a matching result is obtained is searched from the information stored in the storage unit. In an information search device comprising information search means,
The information storage means further stores a search history of each information,
An information search apparatus comprising: abbreviation presentation means for presenting a list of abbreviations of the plurality of information by voice in an order corresponding to a search history of each information. The information search device according to claim 1,
The search history includes the number of searches of each piece of information, and the abbreviation presentation unit presents a list of abbreviations of the information in order of the information with the smallest number of searches. The information search device according to claim 1,
The search history includes the latest search date and time of each piece of information, and the abbreviation presentation means presents a list of abbreviations of the information in order of the information with the latest search date and time being old. . The information search device according to claim 1,
The search history includes the number of searches for each piece of information and the latest search date and time,
A list order selection means for selecting one of the order of the information stored in the oldest date, the order of the information with a small number of searches, and the order of the information in the latest search date and time;
The abbreviation presentation means presents a list of abbreviations of the information in the order selected by the list order selection means.

Images