JP2012189908A - Sound recognition device and sound recognition processing method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a sound recognition device and a sound recognition processing method that are capable of performing sound recognition processing according to the priority level when performing sound recognition processing.SOLUTION: A sound recognition device 300 acquires sound data from the outside, stores the acquired sound data into a primary output database 303, compares the sound data stored in the primary output database 303 with a condition stored in a priority level memory 305 that stores a condition for a priority level to be processed, and performs sound recognition processing on sound data that matches the condition according to the priority level of the condition. A sound recognition result obtained through the sound recognition processing is stored into a sound recognition result database 300D.

Description

  Embodiments of the present invention relate to processing by prioritizing voice recognition devices used in, for example, call center systems.

  In a call center system that handles mail order sales, product / service support desks, or requests for materials, etc., a call between a user and an operator is recorded, and the recorded voice is converted into text by voice recognition using a voice recognition device. The contents can be confirmed by text after the end. In a call center system that performs such voice recognition, voice recognition processing is performed in the order of recording. This voice recognition process is not a processing speed that is processed simultaneously with a call because a large number of calls are recorded at the call center at the same time. The voice recognition processing of the recorded data for one day recorded at the call center is completed, for example, on the next day or two days after the recorded date. That is, the voice recognition processing in the call center system takes time as described above, and it takes a considerable time until the contents of the call can be confirmed by text.

  By the way, regarding the voice recognition processing in the call center, the caller's voice until the caller is connected to the operator terminal is voice-recognized, and when the result of the voice recognition matches the NG word, the waiting call order is given priority. There is a known system that changes the head to the top.

JP 2008-219741 A

  However, the speech recognition process itself in the call center cannot be preferentially performed, and it is not possible to respond to the case where it is desired to immediately confirm the contents of the call in text or to perform the speech recognition process first.

  Accordingly, an object of the present invention is to provide a speech recognition apparatus and a speech recognition processing method capable of performing speech recognition processing according to priority when performing speech recognition processing.

A voice recognition apparatus for achieving the above object includes voice acquisition means for acquiring voice data from the outside, a primary output database in which the voice data is recorded by the voice acquisition means,
A priority memory for storing a condition corresponding to a priority to be processed; voice recognition processing means for voice-recognizing voice data; converting the voice data recorded in the primary output database into the priority memory; Compared to the condition stored in the above, the priority control means for outputting the voice data that matches the condition to the voice recognition processing means according to the priority of the condition, and the voice recognition processing by the voice recognition processing means Storage processing means for storing the speech recognition result in the speech recognition result database.

  In addition, a voice recognition processing method for achieving the above object includes a voice acquisition step of acquiring voice data from the outside, a primary recording step of recording the voice data in a primary output database by the voice acquisition step, and voice data A speech recognition processing step for recognizing the text and converting it to text, and comparing the speech data recorded in the primary output database with a condition stored in a priority memory for storing a condition corresponding to a priority to be processed. A priority control step of performing voice recognition processing on the voice data matching the condition according to the priority of the condition, and storing the voice recognition result subjected to the voice recognition processing by the voice recognition processing step in the voice recognition result database And a storage processing step.

The block diagram which shows the structure of the call center system which concerns on one embodiment of this invention. The block diagram which shows the structure of the speech recognition apparatus provided in the call center system of FIG. The figure which shows the example of the management information of the audio | voice data recorded on the primary output database with which the audio | voice recognition apparatus of FIG. 2 is provided. The figure which shows the example of a recording of the speech recognition result recorded on the speech recognition result database of FIG. The sequence diagram which shows the operation | movement of the speech recognition process according to the priority in the speech recognition apparatus of FIG.

  Hereinafter, a voice recognition device and a voice recognition processing method according to the present embodiment will be described with reference to the drawings.

  FIG. 1 is a block diagram showing a configuration of a call center system showing an embodiment. This call center system is, for example, an IP system, and includes an IP exchange device 100, a call recording device 200, a call voice database 200D, a voice recognition device 300, a voice recognition result database 300D, a customer database 400D, and a telephone terminal 11. 12 (hereinafter collectively referred to as telephone terminal 10).

  The IP exchange apparatus 100 is connected to an IP network 400 and a LAN (local area network), and causes the telephone terminal 10 to receive a call from the IP network 400. In addition, the IP exchange device 100 connects a call from the telephone terminal 10 to a call destination. Further, the IP exchange device 100 sends voice data of a call with the telephone terminal 10 to the call recording device 200 via the LAN.

  The call recording device 200 receives the voice data of the call from the IP exchange device 100 via the LAN, and records the received voice data in the call voice database 200D. Further, the call recording device 200 associates management information such as an operator identification number and a calling number with each voice data and records them in the call voice database 200D. This management information is sent from the IP exchange device 100 to the call recording device 200 together with the voice data together with the operator's identification number and transmission number. Alternatively, the operator's identification number, call number, and business number, which will be described later, may be sent to the call recording device 200 based on the operation of the PC 20 by the operator. Note that voice data and management information thereof are recorded in the call voice database 200D in the order of recording.

  In order to acquire the voice data recorded in the call voice database 200D by the call recording device 200, the voice recognition device 300 makes a call voice acquisition request to the call voice recording device 200 and acquires the voice data. The speech recognition apparatus 300 recognizes the acquired speech data as speech and converts it into text, and records it in the speech recognition result database 300D.

  The customer database 400D records customer information received at this call center. Customer information is read out by a PC (personal computer) 20 described later.

  The telephone terminal 10 is connected to the IP exchange apparatus 100 via a LAN, and is used for an external call by an operator. Note that the telephone terminal 10 may be associated with the PC 20 connected to the LAN in advance. When the telephone terminal 10 and the PC are associated with each other, the operator during the telephone operation operates the PC 20 to The customer information recorded in the customer database 400D can be browsed or updated.

  Although only two telephone terminals 10 are shown in FIG. 1, a large number of telephone terminals are connected to the LAN.

  FIG. 2 is a functional block diagram showing the configuration of the speech recognition apparatus 300. The voice recognition device 300 includes a LAN interface (I / F) unit 301, a voice acquisition processing unit 302, a primary output database 303, a priority control unit 304, a priority memory 305, a voice recognition processing unit 306, a storage processing unit 307, and The registration processing unit 308 is provided.

  For example, the voice acquisition processing unit 302 periodically makes a call voice acquisition request to the call recording device 200, and the voice data sent from the call recording device 200 in response to the call voice acquisition request and the voice data The associated management information is received from the LAN interface unit 301 and written to the primary output database 303. When the voice data is acquired next time, the management information of the voice data acquired this time is written next to the management information of the voice data still remaining in the primary output database 303. That is, the audio data is arranged in the acquired order.

The information written in the primary output database 303 is the same as the information recorded in the call voice database 200D. An example of information recorded in the primary output database 303 is shown in FIG. In FIG. 3, a call identification number such as “call 1” for identifying voice data, a call origination number, a work number described later, and an operator identification number are recorded. For example, “call 1” is recorded that the transmission number is 030-1111222, the business number is 21, and the operator is Mr. Sato (the operator identification number is 014552). The same is recorded for “call 2” and thereafter.

  The priority memory 305 stores information related to audio data to be prioritized, and is referred to by the priority control unit 304 for priority information. The information related to the voice data to be prioritized is, for example, a business number indicating the contents of the call center, a call number of an important customer, an identification number of a new operator, or the like. Here, the business number is a number assigned according to the contents of the operator's reception such as a call related to a campaign, a call related to repair, a call related to inquiry, and the like. Giving priority to a specific business number is effective when performing voice recognition on the voice data of a call related to the campaign when the tabulation timing is approaching when tabulating the specific campaign. In order to quickly make use of important customers' needs, voice recognition is given priority to the calling number of important customers, and voices of calls answered by specific operators are given priority to educate new operators. It becomes possible.

  The priority control unit 304 stores management information related to the voice data of the call to be prioritized stored in the priority memory 305, and the transmission number, work number, operator identification number, etc. of the voice data written in the primary output database 303. Whether or not the audio data written in the primary output database 303 is a processing target to be prioritized is determined in the order of management information written in the primary output database 303. Further, the priority control unit 304 reads out the audio data determined to be preferentially processed from the primary output database 303 and sequentially outputs it to the speech recognition processing unit 306. If there is no voice data to be preferentially processed, the audio data that need not be preferentially processed is output to the voice recognition processing unit 306 in the order written in the primary output database 303. Further, when outputting the voice data from the primary output database 303 to the voice recognition processing unit 306, the priority control unit 304 adds the priority information to the management information of the output voice data and outputs it to the storage processing unit 307. .

  Note that the priority control unit 304 may include a timer, and may output to the speech recognition processing unit 306 audio data that does not need to be processed with priority for a certain period of time. In this case, when the timer times out, the priority control unit 304 notifies the voice acquisition processing unit 302 of the timeout, and the voice acquisition processing unit 302 makes a call to the call recording apparatus 200 in response to the notification of the timeout. A voice acquisition request may be made.

  In this way, the process of outputting the voice data that does not need to be prioritized using the timer to the voice recognition processing unit 306 is limited to a certain time, and the next recorded voice data is acquired when a time-out occurs. Thus, the audio data to be processed with priority can be processed with priority. If there is no timer, voice recognition processing of voice data that does not need to be prioritized cannot be performed until there is no voice data to be processed with priority in the primary output database 303. By linking, voice recognition processing for voice data that does not need to be prioritized can be performed little by little (with a lower priority).

  The voice recognition processing unit 306 includes a buffer therein, and voice data is written into the buffer by the priority control unit 304. The speech recognition processing unit 306 performs speech recognition processing in the order in which the speech data written in the buffer is written, and outputs text data to the storage processing unit 307 as a result of speech recognition.

  The storage processing unit 307 collates the text data, which is the voice recognition result received from the voice recognition processing unit 306, with the management information to which the priority information sent from the priority control unit 304 is added. , Classified by customer, business number, or operator, and recorded in the speech recognition result database 300D so that the user can identify what is preferentially processed and what is not. For example, the preferentially processed and the non-prioritized are managed in different folders. An example of recording the speech recognition result in the speech recognition result database 300D is shown in FIG.

  In this way, by classifying or making it possible to identify whether priority is given or not and recording it in the speech recognition result database 300D, it becomes easier to understand what should be confirmed preferentially when speech data is to be confirmed in text. .

  The registration processing unit 308 stores the priority information in the priority memory 305 in response to the priority information registration request input from the LAN interface unit 301. For example, the operator operates the PC 20 to make a priority information registration request to the speech recognition apparatus 300. At this time, when making a registration request of priority information according to the customer, the customer information is browsed by referring to the customer database 400D from the PC 20.

  FIG. 5 is a sequence diagram showing the operation of the speech recognition process according to the priority in the speech recognition apparatus.

  First, the voice acquisition processing unit 302 outputs a call voice request notification to the call recording device 200, receives the voice data and its management information from the LAN interface 301 (S1), and writes it in the primary output database 303 (S2). . The priority control unit 304 compares the audio data management information written in the primary output database 303 with the priority memory 305 to determine whether there is a processing target to be prioritized (S3). When there is a processing target to be prioritized (Yes in step S3), the priority is determined in the order of management information (S4), and if the result of determination is a processing target to be prioritized (priority in step S4), priority should be given The voice data is output to the voice recognition processing unit 306 (S5). At this time, the priority control unit 304 outputs the management information of the audio data to the storage processing unit 307. Thereafter, it is determined whether there is a processing target to be prioritized (S3). As a result of the priority determination, if it is not a processing target to be prioritized (not prioritized in step S4), the audio data is not taken out from the primary output DB, and the priority determination of the next audio data is continued (S6).

  If there is no processing target to be prioritized in step S3 (No in step S3), non-prioritized speech data is extracted from the primary output database 303 and output to the speech recognition processing unit 306. At this time, the priority control unit 304 outputs the management information of the audio data to the storage processing unit 307. Thereafter, it is determined whether or not the timer has timed out (S8), and if it has not timed out (No in step S8), the process of outputting the non-priority audio data in step S7 to the speech recognition processing unit 306 is continued and timed out. If so (Yes in step S8), the process returns to step S1 to acquire voice data from the call recording device 200.

  As described above, the voice recognition apparatus stores in advance the conditions for performing voice recognition processing with priority, and performs voice recognition processing on voice data that matches the conditions, so that the voice recognition processing is performed according to priority. Can be processed. Further, by recording the voice recognition result in the voice recognition result database 300D so that the user can identify whether or not the voice recognition result is prioritized, the prioritized ones can be collectively confirmed when the text is confirmed.

  The present invention is not limited to the above configuration, and various modifications are possible. For example, in the above embodiment, the priority control unit 304 included in the speech recognition apparatus 300 extracts speech data from the primary output database 303 after determining the priority of the speech data recorded in the primary output database 303, and recognizes the speech. Although output to the processing unit 306, the priority control unit 304 has an internal memory, acquires audio data management information (or management information and audio data) from the primary output database, and is determined to be non-prioritized. For voice data, the management information (or management information and voice data) may be returned to the primary output database. Moreover, in the said embodiment, although the priority was made into priority or non-priority, you may make it process from a thing with a high priority divided into three or more steps according to the priority degree.

  In addition, although some embodiment of this invention was described, these embodiment is shown as an example and is not intending limiting the range of invention. These novel embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the scope of the invention. These embodiments and modifications thereof are included in the scope and gist of the invention, and are included in the invention described in the claims and the equivalents thereof.

DESCRIPTION OF SYMBOLS 10, 11, 12 ... Telephone terminal, 20 ... PC, 100 ... IP switching device, 200 ... Call recording device, 200D ... Call voice database, 300 ... Voice recognition device, 300D ... Voice recognition result database, 301 ... LAN interface part, 302 ... Voice acquisition processing unit, 303 ... Primary output database, 304 ... Priority control unit. 305 ... Priority memory, 306 ... Voice recognition processing unit, 307 ... Storage processing unit, 308 ... Registration processing unit, 400 ... IP network.

Claims (6)

Audio acquisition means for acquiring audio data from outside;
The voice data, a primary output database recorded by the voice acquisition means;
A priority memory for storing conditions according to the priority to be processed;
Speech recognition processing means for recognizing speech data and converting it into text;
The voice data recorded in the primary output database is compared with a condition stored in the priority memory, and voice data that matches the condition is output to the voice recognition processing means according to the priority of the condition. Priority control means;
A speech recognition apparatus comprising: storage processing means for storing a speech recognition result subjected to speech recognition processing by the speech recognition processing means in a speech recognition result database. The priority control unit notifies the storage processing unit of the priority information of the voice data as the voice data is output to the voice recognition processing unit according to the priority of the condition.
The voice according to claim 1, wherein the storage processing means stores the voice recognition result in the voice recognition result database for each priority based on priority information notified from the priority control means. Recognition device.   3. The registration processing unit according to claim 1, further comprising: a registration processing unit that registers a priority to be processed and a condition according to the priority in the priority memory in response to an instruction from the outside. Voice recognition device. An audio acquisition step for acquiring audio data from the outside;
A primary recording step of recording the audio data in a primary output database by the audio acquisition step;
A speech recognition processing step for recognizing speech data and converting it into text;
The audio data recorded in the primary output database is compared with a condition stored in a priority memory that stores a condition corresponding to a priority to be processed, and audio data that matches the condition is compared with the priority of the condition. A priority control step for performing voice recognition processing according to
And a storage processing step of storing the speech recognition result subjected to the speech recognition processing in the speech recognition processing step in a speech recognition result database.   5. The speech recognition apparatus according to claim 4, wherein the storage processing step stores the speech recognition result in the speech recognition result database for each priority.   6. The registration process step according to claim 4, further comprising a registration processing step of registering in the priority memory a priority to be processed and a condition corresponding to the priority in accordance with an instruction from the outside. Speech recognition processing method.

Images