Telephone call voice recognition system and telephone call voice recognition method

Abstract

To optimize voice recognition of calls in a call center to enable proper operation of the call center and effective use of a server resource.SOLUTION: A call voice recognition system includes: a call recording apparatus; a call record management apparatus; a voice recognition apparatus; and a voice recognition control apparatus which receives call information and voice data from the call record management apparatus, transmits the voice data to the voice recognition apparatus, and issues an instruction to transcribe the voice data to text. The voice recognition control apparatus holds a recognition order management table that stores recognition priority for each call. The voice recognition control apparatus includes a recognition order control unit for controlling voice recognition order of calls for each call, and a recognition execution instruction unit that instructs the voice recognition apparatus to transcribe the voice data of the calls to text. The recognition execution instruction unit instructs a call to be subjected to voice recognition to the voice recognition apparatus, based on the recognition priority stored in the recognition order management table.SELECTED DRAWING: Figure 16

Description

[Technical field]

[0001] The present invention relates to a telephone call voice recognition system and a telephone call voice recognition method, and more particularly to a telephone call voice recognition system and a telephone call voice recognition method that are suitable for appropriate operation of a call center and effective use of server resources by performing voice recognition in accordance with an appropriate voice recognition processing order depending on the type of site handling the call, the need for voice recognition, and the resources required for voice recognition of the call. [Background technology]

[0002] In telephone response work at call centers and the like, systems that support customer interactions by performing voice recognition on conversations between customers and operators, converting the recognition results into text and displaying them on the operator's terminal or by displaying the conversation between the operator and the customer on a supervisor's terminal are becoming common.

[0003] A technology for improving the efficiency of supervisors' monitoring work and the quality of operators' customer service by using voice recognition in call centers is described, for example, in Patent Document 1. The call center system described in Patent Document 1 recognizes the contents of a call between a customer and an operator using voice recognition, monitors specific words in the resulting text, and issues an alert to the supervisor when a warning value becomes large. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2020-150409 A Summary of the Invention [Problem to be solved by the invention]

[0005] Generally, speech recognition of a call requires server resources (hardware resources, software resources) of a speech recognition device, and if the number of recognition requests exceeds the server resources, a waiting period occurs in the recognition process.

[0006] For example, in the case of a system that operates a voice recognition service for multiple companies or locations using a so-called cloud-based service format, it is common to treat service requests from all companies or locations on an equal footing. Therefore, in order to smoothly recognize voice calls, it is necessary to reserve server resources for the voice recognition device required for voice recognition of calls, assuming the busy periods of the business formats of all companies or locations. This incurs costs for introducing the system, but reducing this cost will result in a decrease in the service level. Therefore, in order to maintain the required service level and reduce operating costs, it is necessary to introduce control into the voice recognition system that efficiently utilizes limited server resources.

[0007] In particular, in call center operations where calls must be monitored in real time, operators must answer calls while looking at the text, and supervisors must monitor calls that require special attention, effective allocation of server resources is an important issue.

[0008] Also, generally, depending on the type of call to the call center, particularly depending on the company, there are calls that require attention, such as calls related to complaints, and calls that are relatively easy for the operator to handle, such as inquiries about fees and products. In this way, when there are simultaneous calls from customers for a certain period of time depending on the type of call to the call center, the order in which the calls are converted into text by voice recognition and displayed to the operator or supervisor is important for improving the quality of the operator's response to customers.

[0009] In the telephone call voice recognition described in Patent Document 1, no consideration is given to how to assign the order of voice recognition for telephone calls.

[0010] An object of the present invention is to provide a call voice recognition system and method suitable for proper operation of a call center and effective use of server resources by performing voice recognition in a call center according to an appropriate voice recognition processing order depending on the type of site handling calls, the need for voice recognition, and the resources required for voice recognition of calls. [Means for solving the problem]

[0011] The call voice recognition system of the present invention is preferably a call voice recognition system that recognizes call voice, and includes a call recording device that records the voice of a call, a call recording management device that receives recorded voice data from the call recording device and transmits the voice data to a voice recognition control device in association with the call information, a voice recognition device that inputs the call voice and converts it to text, and a voice recognition control device that receives the call information and voice data from the call recording management device and transmits the voice data to the voice recognition device and instructs it to convert the voice data to text, the voice recognition control device holds a recognition order management table that stores the recognition priority for each call, the voice recognition control device has a recognition order control unit that controls the voice recognition order of each call, and a recognition execution instruction unit that instructs the voice recognition device to convert the voice data of the call to text, and the recognition execution instruction unit instructs the voice recognition device to perform voice recognition on the call based on the recognition priority stored in the recognition order management table. Effect of the Invention

[0012] According to the present invention, a call center can provide a call voice recognition system and a call voice recognition method that are suitable for appropriate operation of call centers and effective use of server resources by performing voice recognition in a call center according to an appropriate voice recognition processing order depending on the type of site handling calls, the need for voice recognition, and the resources required for voice recognition of calls. [Brief description of the drawings]

[0013] [Figure 1]FIG. 1 is a diagram illustrating an overview of the configuration and processing of a telephone call voice recognition system according to a first embodiment. [Diagram 2] 1 is a block diagram showing a configuration of a call recording device; [Diagram 3] 1 is a block diagram showing a configuration of a call recording management device. [Figure 4] FIG. 2 is a block diagram showing a configuration of a voice recognition control device. [Diagram 5] FIG. 1 is a block diagram showing a configuration of a voice recognition device. [Figure 6] FIG. 2 is a block diagram showing a configuration of a recognition result management device. [Figure 7] FIG. 11 is a diagram illustrating an example of a call information table. [Figure 8] FIG. 11 illustrates an example of an operator management information table. [Figure 9] FIG. 11 is a diagram illustrating an example of an IP extension information table. [Figure 10] FIG. 11 is a diagram showing an example of a recording information table. [Figure 11] FIG. 11 is a diagram illustrating an example of a recognition result table. [Figure 12] FIG. 13 illustrates an example of a priority management table. [Figure 13] FIG. 11 illustrates an example of a site information management table. [Figure 14] FIG. 13 illustrates an example of a recognition order management table. [Figure 15A] This is a diagram (part 1) that explains a series of processes from when an external call terminal makes a call to when the call is voice-recognized and the text is displayed on a PC terminal. [Figure 15B] This is a diagram (part 2) that explains the series of processes from when an external calling terminal makes a call to when the call is voice-recognized and the text is displayed on a PC terminal. [Figure 16] FIG. 2 is a diagram illustrating a series of processes for voice recognition. [Figure 17] FIG. 11 is a diagram for explaining an overview of the configuration and processing of a telephone call voice recognition system according to a second embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0014] Hereinafter, each embodiment of the present invention will be described with reference to FIGS.

[0015] [Embodiment 1] Hereinafter, a first embodiment of the present invention will be described with reference to FIGS.

[0016] First, the configuration and processing of a telephone call voice recognition system according to the first embodiment will be outlined with reference to FIG. As shown in Fig. 1, the speech recognition system of this embodiment has a configuration in which an IP-PBX (Internet Protocol-Private Branch eXchange, IP line compatible private branch exchange) 30 connected to a call terminal 1 of a caller (a person making a call) such as a customer via a public network 5 is connected to an operator's call terminal 20 (indicated as 20a, 20b, and 20c in the figure) via a network 7 to make a call. The operator can operate the call terminal 20 from an operator's PC terminal 25 (indicated as 25a, 25b, and 25c in the figure), and when an incoming call from a caller is displayed on the operator's PC terminal 25, the operator operates the answer from the PC terminal 25, the call terminal 20 performs an answering operation, and the caller and the operator are in a conversation state.

[0017] In this embodiment, the call center is divided into departments such as site S1, site S2, and site S3. For example, site S1 is a department for handling customer complaints, site S2 is a department for handling questions from customers about products, services, etc., and site S3 is a department for accepting new customers' requests for products, services, etc., and operators who handle calls from these departments respond to each call.

[0018] The telephone speech recognition system has a supervisor who manages and supervises the operators, and monitors the situation using the telephone terminal 20s and the supervisor's PC terminal 25s, and performs operations to assist the operators when necessary. In addition to the telephone control unit that the PC terminal 25 has, the supervisor's PC terminal 25s also operates a management tool that displays call center operation information for the administrator.

[0019] The telephone call voice recognition system 100 is configured by connecting an IP-PBX 30, a CTI (Computer Telephony Integration) device 40, a telephone call recording device 50, a telephone call recording management device 60, a voice recognition control device 110, a voice recognition device 120, and a recognition result management device 130 via a network 7.

[0020] The IP-PBX 30 performs protocol conversion between the IP network and the public network 5, controls incoming and outgoing calls, etc. The CTI device 40 is the core device for call control in the call center, and has functions such as managing the status of the call center and transmitting call status information received from the IP-PBX 30 to the operator PC terminal 25 and the supervisor PC terminal 25s.

[0021] The call recording device 50 is a device that creates recorded data by a call recording process. The call recording management device 60 is a device that manages the recorded data and the call information added thereto, and transmits the voice to the voice recognition control device 110. When a voice recognition request is received from the call recording management device 60, the voice recognition control device 110 is a device that selects from a plurality of voice recognition devices 120 in consideration of the usage status and hardware load of the voice recognition device 120, and the priority of the recognition of the call, and transmits the call voice to be recognized to the voice recognition device. The voice recognition device 120 is a device that is equipped with a voice recognition engine and converts the transmitted voice into a call text. The recognition result management device 130 is a device that manages the data that has been recognized and converted into text by the voice recognition device 120.

[0022] Next, the configuration of each part of the telephone call voice recognition system will be described in detail with reference to FIG. 2 to FIG.

[0023] First, the configuration of the call recording device will be described with reference to FIG. As shown in FIG. 2, the call recording device 50 includes a network I / F (Interface) unit 501, a recording unit 502, a call recording management device I / F unit 503, a terminal I / F unit 504, and a storage unit 505. The network I / F (Interface) unit 501 is a functional unit that manages an interface with an IP network and receives RTP packets. The recording unit 502 is a functional unit that records calls using RTP packets. The call recording management device I / F unit 503 is a functional unit that transmits recorded voice data to the call recording management device 60. The terminal I / F unit 504 is a functional unit that manages an interface with the operator's PC terminal 25 and the supervisor's PC terminal 25s. The storage unit 505 is a functional unit that stores tables and data. The storage unit 505 of the call recording device 50 stores an IP extension information table 72, a recording information table 73, and a voice data file 90. Details of each table will be described later.

[0024] Next, the configuration of the call recording management device 60 will be described with reference to FIG. As shown in Fig. 3, the call recording management device 60 has a call recording judgment unit 601, a call recording data acquisition unit 602, a CTI device I / F unit 603, a call recording device I / F unit 604, a voice recognition control device I / F unit 605, and a memory unit 606. The call recording judgment unit 601 is a functional unit that judges whether or not a call should be recorded from call information. The call recording data acquisition unit 602 is a functional unit that receives recorded voice data from the call recording device 50. The CTI device I / F unit 603 is a functional unit that manages the interface with the CTI device 40. The call recording device I / F unit 604 is a functional unit that manages the interface with the call recording device 50. The voice recognition control device I / F unit 605 is a functional unit that manages the interface with the voice recognition control device 110. The memory unit 606 is a functional unit that stores tables and data. The storage unit 606 of the call recording management device 60 stores a call information table 70 and a voice data file 90. The details of the tables will be described later.

[0025] Next, the configuration of the voice recognition control device 110 will be described with reference to FIG. As shown in FIG. 4, the voice recognition control device 110 includes a voice data file management unit 1101, a recognition execution instruction unit 1102, a recognition order control unit 1103, a speech time calculation unit 1104, a sentiment analysis unit 1105, a call recording management device I / F unit 1106, a voice recognition device I / F unit 1107, and a storage unit 1108. The voice data file management unit 1101 is a functional unit that manages the voice data file 90. The recognition execution instruction unit 1102 is a functional unit that transmits the voice data file 90 to the voice recognition device 120 and instructs voice recognition. The recognition order control unit 1103 is a functional unit that controls the recognition order of the voice data of the call. The speech time calculation unit 1104 is a functional unit that calculates the speech time of the speaker in the call. The sentiment analysis unit 1105 is a functional unit that analyzes the sentiment of the speaker in the call and quantifies it. The storage unit 1108 is a functional unit that stores tables and data. The call recording management device I / F unit 1106 is a functional unit that controls the interface with the call recording management device 60. The voice recognition device I / F unit 1107 is a functional unit that controls the interface with the voice recognition device 120. The storage unit 1108 of the voice recognition control device 110 stores a call information table 70, a priority management table 76, a site information management table 77, a recognition order management table 79, and a voice data file 90. The tables will be described in detail later.

[0026] Next, the configuration of the voice recognition device 120 will be described with reference to FIG. As shown in FIG. 5, the voice recognition device 120 has a recognized voice receiving unit 1201, a recognition target designation unit 1202, a recognition engine 1203, a voice recognition control device I / F unit 1204, a recognition result management device I / F unit 1205, and a storage unit 1206. The recognized voice receiving unit 1201 is a functional unit that receives voice data from the call recording device 50. The recognition target designation unit 1202 is a functional unit that receives a notification of a call to be voice recognized from the voice recognition control device 110 and designates a recognition target to the recognition engine 1203. The recognition engine 1203 is a functional unit that recognizes voice data and converts it into text. The voice recognition control device I / F unit 1204 is a functional unit that controls an interface with the voice recognition control device 110. The recognition result management device I / F unit 1205 is a functional unit that controls an interface with the recognition result management device 130. The storage unit 1206 is a functional unit that stores tables and data. The storage unit 1206 of the voice recognition device 120 stores a recognition result table 74 and a recognition result text 93. The details of the table will be described later.

[0027] Next, the configuration of the recognition result management device 130 will be described with reference to FIG. As shown in FIG. 6, the recognition result management device 130 includes a recognition result management unit 1301, a speech evaluation unit 1302, a voice recognition device I / F unit 1303, a terminal I / F unit 1304, and a storage unit 1305. The recognition result management unit 1301 is a functional unit that manages the text of the recognition result received from the voice recognition device 120 and related information. The speech evaluation unit 1302 is a functional unit that performs evaluation, for example, whether a warning word is included from the text of the recognition result, and generates speech evaluation data to be stored in the speech evaluation table 80. The voice recognition device I / F unit 1303 is a functional unit that manages an interface with the voice recognition device 120. The terminal I / F unit 1304 is a functional unit that manages an interface with the operator PC terminal 25 and the supervisor PC terminal 25s. The storage unit 1305 is a functional unit that stores tables and data. The recognition result table 74, the recognition result text 93, and the speech evaluation table 80 (details not shown) are stored in the storage unit 1305 of the recognition result management device 130. The details of the tables will be described later.

[0028] Next, the data structure used in the telephone speech recognition system will be described with reference to FIGS.

[0029] The call information table 70 is a table in which the call recording management device 60 holds information related to calls sent from the CTI device 40. As shown in Fig. 7, the call information table 70 has fields for a calling number 70a, an extension number 70b, a call start time 70c, and a call end time 70d.

[0030] The calling number 70a stores the calling number of the call terminal 1 that made the call. The extension number 70b stores the extension number in the call voice recognition system of the call that received this call. The call start time 70c stores the time when the incoming call started. The call end time 70d stores the time when the incoming call ended.

[0031] The operator management information table 71 is a table that holds setting information of an operator, and although not shown, is stored in a memory unit of the CTI device 40. As shown in FIG. 8, the operator management information table 71 has fields for an extension number 71a, an operator ID 71b, an operator name 71c, and an affiliated site ID 71d. The extension number 71a stores the extension number assigned to this operator. The operator ID 71b stores an identifier assigned to the operator. The operator name 71c stores the name of this operator. The affiliated site ID 71d stores the ID of the department (site) to which this operator belongs.

[0032] The IP extension information table 72 is a table showing the correspondence between extension numbers and IP address correspondence tables, and is held in the call recording device 50. As shown in Fig. 9, the IP extension information table 72 has fields for extension number 72a, IP address 72b, and site ID 72c. The IP extension information table 72 means that the extension number in the extension number 72a field has the IP address in the IP address 72b field. The site ID 72c stores a name or identifier indicating the department (site) to which the extension number belongs.

[0033] The recording information table 73 is a table for holding information related to the recording of a call, and is held in the call recording device 50. As shown in FIG. 10, the recording information table 73 has fields for a call recording device ID 73a, a recording ID 73b, a recording start time 73c, an IP address 73d, and a voice data file 73e. The call recording device ID 73a stores an ID for identifying the call recording device 50. The recording ID 73b stores an ID for identifying an entry of the call recording. The recording start time 73c stores the start time of the call recording. The IP address 73d stores the IP address of the call terminal 20 related to the call. The voice data file 73e stores the file path of the voice data file of the call.

[0034] The recognition result table 74 is a table for holding information on the voice recognition result, and is held in the voice recognition device 120, copied to the recognition result management device 130, and held as the recognition result table 74. As shown in FIG. 11, the recognition result table 74 has fields for a call recording device ID 74a, a recording ID 74b, an extension number 74c, a call ID 74d, and a recognition result text 74e. The call recording device ID 74a stores an ID for identifying the call recording device 50. The recording ID 74b stores an ID for identifying an entry of a call recording. The extension number 74c stores an extension number in the center of the call terminal 20. The call ID 74d stores an ID for identifying a call. The recognition result text 74e stores a file path name of a text file when an utterance in a call is voice-recognized and converted into text.

[0035] The priority management table 76 is a table that stores information regarding the priority of each priority level, and is held in the voice recognition control device 110. As shown in FIG. 12, the priority management table 76 has fields for priority consideration items 76a and priority 76b. The priority consideration items 76a store character strings that identify items for considering each priority level. In this embodiment, the priority consideration items include "site priority" indicating the priority of each site, "utterance time" which is the utterance time for each call, and "emotion level" which is the result of emotion analysis of the call. The priority 76b stores the priority of the corresponding priority consideration item when determining the recognition order of the calls. Here, it is assumed that the smaller the value of the priority 76b, the higher the priority for voice recognition.

[0036] The voice recognition control device 110 instructs to recognize the voice corresponding to the call according to this priority. For example, as shown in Fig. 12, since the priority consideration item 76a "site priority" is the first priority, the recognition order of the call is determined according to the site priority, and if the "site priority" is the same, the recognition order of the call is determined according to the next priority "emotion level", and if the "emotion level" is the same, the recognition order of the call is determined according to the next priority "utterance time", and if the "emotion level" is the same, the recognition order of the call is determined according to the next priority "utterance time".

[0037] The site priority is set so that a high priority is given to a site whose management side needs to know the details of the call contents, such as a complaint handling department.

[0038] In addition, the emotion level increases the priority of a call when the customer's emotion is undesirable, such as anger.

[0039] Regarding the duration of a call, if the duration of a call is long, it will consume a lot of hardware and software resources for voice recognition, so the priority of that call will be lowered.

[0040] In this embodiment, the recognition priority is determined based on three types of priority consideration items, but one or two of these may be used, or other priority consideration items may be added.

[0041] The site information management table 77 is a table that stores information for each site, and is a table held by the voice recognition control device 110. As shown in FIG. 13, the site information management table 77 consists of a site ID 77a, a site name 77b, and a site priority 77c. The site ID 77a stores an ID for uniquely identifying a site. The site name 77b stores the name of the site. The site priority 77c stores the priority of the site when determining the recognition priority of the site. Here, the smaller the site priority value, the higher the priority.

[0042] The recognition order management table 79 includes a call ID 79a, a recognition request time 79b, a site ID 79c, a call duration 79d, a voice data file 79e, a speech time 79f, an emotion level 79g, and a recognition priority 79h. The call ID 79a stores an ID that uniquely identifies the call. The recognition request time 79b stores the time when the voice recognition control device 110 requests the voice recognition device 120 to perform voice recognition. The site ID 79c stores the site ID of the site to which the extension number of the call belongs. The call duration 79d stores the call duration of the call. The voice data file 79e stores the file path of the voice data file. The speech time 79f stores the time when the speaker of the call spoke. The emotion level 79g stores the level of emotion analysis of the call. For example, the higher the emotion level value, the more angry the customer's emotion is, and the lower the value, the calmer the customer's emotion is. The recognition priority 79h stores the priority of the request for voice recognition of the call. The smaller the value of the recognition priority 79h, the earlier the recognition order is assigned.

[0043] Next, the process of the telephone call voice recognition system will be described in detail with reference to FIGS. 15A to 16. FIG.

[0044] First, a series of processes from when an external call terminal makes a call to when the recognition results are displayed on the operator's PC terminal 25 and the supervisor's PC terminal 25s will be described with reference to Figs. 15A and 15B.

[0045] First, a call is made from the customer's telephone terminal 1 (A01), is switched by the IP-PBX 30, and is then received by the telephone terminal 20 of the call center operator (A02).

[0046] The IP-PBX 30 transmits the call information to the CTI device 40 (A03), and the CTI device 40 transmits the call information to the call recording management device 60 (A04).

[0047] The IP-PBX 30 transmits an RTP packet to the call terminal 20, and at the same time, the IP-PBX 30 transfers the copied RTP packet to the call recording device 50 (A05).

[0048] The call recording device 50 reports the start of the call to the call recording management device 60 using the extension number as a parameter (A06), starts split recording based on the RTP packets (S02a, S02b, ...), and writes the necessary information into the recording information table 73 shown in Figure 10 by referring to the IP extension information table 72 shown in Figure 9.

[0049] In addition, the CTI device 40 transmits the call information to the call recording management device 60 .

[0050] The call recording management device 60 uses the received extension number as a key to refer to a call recognition information table (not shown) and determines whether or not to perform voice recognition on the call to that extension number (S03). The call recognition information table contains information for each extension number as to whether or not to perform real-time voice recognition on the call to that extension number. When the call recording management device 60 determines that the call should be subjected to real-time voice recognition, it instructs the call recording device 50 to transfer the voice data (A07).

[0051] The call recording device 50 receives the voice data, associates it with the call, and transmits the voice data to the call recording management device 60 (A08).

[0052] Next, the call recording management device 60 transmits the call information and the voice data to the voice recognition control device 110 (A09).

[0053] The voice recognition control device 110 refers to information such as free resources of the voice recognition device 120 and selects the voice recognition device 120 that will perform voice recognition (S04).

[0054] Next, the voice recognition control device 110 refers to the recognition priority 79h in the recognition order management table 79 shown in FIG. 14, and transmits the voice file of the corresponding call to the voice recognition device 120 according to the value of the recognition priority 79h, and instructs the voice recognition device 120 to recognize the voice (A10).

[0055] The voice recognition device 120 receives the voice recognition and performs voice recognition on the voice file of the corresponding call (S06).Then, the voice recognition device 120 transmits the recognition result including the recognized text to the recognition result management device 130 (A11).

[0056] The recognition result management device 130 performs speech evaluation (S07), and transmits speech evaluation information and the recognition result text to the operator's PC terminal 25 and the supervisor's PC terminal 25s (A12, A13).

[0057] Then, the operator's PC terminal 25 and the supervisor's PC terminal 25s display the recognized text (S08, S09).

[0058] Next, a series of processes for voice recognition will be described with reference to FIG. First, the call recording management device 60 transmits call information and a voice data file to the voice recognition control device 110 (A100, A101).

[0059] The recognition order control unit 1103 reads the transmitted call information (A102).

[0060] Moreover, the recognition order control unit 1103 reads the site information management table 77 shown in FIG. 13 (A103).

[0061] On the other hand, the voice file management unit 1101 reads the voice data file 90, and transmits it to the speech time calculation unit 1104 and the emotion analysis unit 1105 (A105, A107).

[0062] The speech time calculation unit 1104 analyzes the voice data associated with the call, calculates the speech time of the call, and writes the calculated speech time into the speech time 79f of the recognition order management table 79 (A106).

[0063] The emotion analysis unit 1105 analyzes the voice data associated with the call, analyzes the emotion of the speaker's call, quantifies it, classifies it into a level, and writes it in the emotion level 79g of the recognition order management table 79 (A108). Various techniques for analyzing the emotion of a speaker from the voice have already been researched as AI-related techniques, and are publicly known techniques.

[0064] The recognition order control unit 1103 reads (A104) data of the priority management table 76 shown in Fig. 12, and reads (A109) data of the recognition order management table 79 shown in Fig. 14. Then, the recognition order control unit 1103 obtains the site priority 77c of the site corresponding to the call in the site information management table 77, the speech time 79f, and the emotion level 79g in the recognition order management table 79, calculates the recognition priority for each call according to the priority 76b in the priority management table 76, and writes the calculated recognition priority to the recognition priority 79h (A110).

[0065] The recognition priority is calculated in the order of priority 76b, and the higher the site priority (the smaller the value), the higher the priority in terms of emotion level (the larger the value), and the shorter the speech time, the higher the recognition priority (the smaller the value) is set.

[0066] The recognition execution instruction unit 1102 acquires the recognition priority 79h from the recognition order management table 79 (A111), and notifies the recognition target instruction unit 1202 of the voice recognition device 120 of the calls to be subjected to voice recognition in accordance with the acquired recognition priority 79h (A112).

[0067] In response to this, the recognition target designation unit 1202 of the voice recognition device 120 instructs the recognition engine 1203 on the call to be voice recognized (A114). The recognition engine 1203 inputs voice data corresponding to the instructed call (A113), performs voice recognition, writes necessary information in the recognition result table 74, and outputs the recognition result text 93 (A115).

[0068] As described above, according to this embodiment, the voice recognition control device instructs the voice recognition device to perform voice recognition in accordance with an appropriate voice recognition processing order depending on the type of site handling the call, the need for voice recognition, and the resources required to voice recognize the call.

[0069] This optimizes the processing order of voice recognition, enabling proper operation of call centers and the like and effective use of server resources.

[0070] [Embodiment 2] Hereinafter, the second embodiment will be described with reference to FIG. In the first embodiment, an example has been described in which a department in a call center is regarded as a "site" and site priority is determined based on the necessity for voice recognition of calls made to the site.

[0071] The second embodiment is an example in which a voice recognition service is provided to each company as a cloud, and a client that receives one voice recognition service is regarded as a "site."

[0072] The site priority may be set higher for those with higher contract fees that depend on a fixed term (such as a one-year contract) or on a quantitative amount such as the number of speech recognition words, etc. Also, the priority of a client that the voice recognition service considers to be important may be set higher.

[0073] In the system configuration of the second embodiment, a cloud speech recognition system 1000 is connected to each of sites S1, S2, and S3 via a router 300a and a router 300b, respectively, through a global network 9 such as the Internet.

[0074] Each of sites S1, S2, and S3 has an IP-PBX 30, a CTI device 40, a call recording device 50, and a call recording management device 60 similar to those in embodiment 1, and calls made and received from customers and recording of calls are performed at each site.

[0075] The cloud speech recognition system 1000 has a speech recognition control device 110, a speech recognition device 120, and a recognition result management device 130 similar to those in the first embodiment, and transmits recognition result text of a speech recognition result to sites S1, S2, and S3. The cloud speech recognition system 1000 receives speech from each site, and determines the recognition priority taking into account the site priority, similar to the first embodiment.

[0076] As described above, according to this embodiment, each site is regarded as a single corporate entity, and when a voice recognition service is provided in the form of a cloud, a voice recognition service for telephone calls can be provided using voice recognition in accordance with appropriate grounds such as a contract fee. [Explanation of symbols]

[0077] 1...Call terminal (customer) 5. Public Network 7. Network 9. Global Network 30…IP-PBX 40...CTI device 50...Call recording device 60...Call recording management device 110...Voice recognition control device 120...Voice recognition device 130...Recognition result management device 20…Operator call terminal 25…Operator PC terminal 20s…Telephone terminal 25s…Supervisor's PC terminal 1000…Cloud voice recognition system

Claims

1. A telephone call voice recognition system for performing voice recognition on a telephone call, comprising: A call recording device for recording the audio of a call; a call recording management device that receives the recorded voice data from the call recording device, and transmits the voice data to a voice recognition control device in association with call information; A voice recognition device that inputs voice messages and converts them into text; a voice recognition control device that receives call information and voice data from the call recording management device, transmits the voice data to the voice recognition device, and instructs the voice recognition device to convert the voice data into text; the voice recognition control device holds a recognition order management table that stores a recognition priority for each call; The voice recognition control device includes a recognition order control unit that controls a voice recognition order for each call; a recognition execution instruction unit that instructs the voice recognition device to convert voice data of a call into text; the recognition execution instruction unit instructs the speech recognition device to perform speech recognition on a call based on the recognition priority stored in the recognition order management table; Furthermore, the voice recognition control device holds a site information management table that stores site priorities of sites determined for each telephone number of a call destination, The speech recognition system according to claim 1, wherein the recognition order is determined based on site priorities in the site information management table.

2. A telephone call speech recognition system for performing speech recognition on a telephone call, comprising: A call recording device for recording the audio of a call; a call recording management device that receives the recorded voice data from the call recording device, and transmits the voice data to a voice recognition control device in association with call information; A voice recognition device that inputs voice messages and converts them into text; a voice recognition control device that receives call information and voice data from the call recording management device, transmits the voice data to the voice recognition device, and instructs the voice recognition device to convert the voice data into text; the voice recognition control device holds a recognition order management table that stores a recognition priority for each call; The voice recognition control device includes a recognition order control unit that controls a voice recognition order for each call; a recognition execution instruction unit that instructs the voice recognition device to convert voice data of a call into text; the recognition execution instruction unit instructs the speech recognition device to perform speech recognition on a call based on the recognition priority stored in the recognition order management table; The voice recognition control device further includes a speech time calculation unit that calculates a speech time of a speaker in a call, A telephone call voice recognition system, wherein the recognition order is determined based on the speaking time of each speaker in the telephone call.

3. Furthermore, the voice recognition control device has an emotion analysis unit that analyzes and quantifies the emotion of a speaker in a call, 2. The speech recognition system according to claim 1, wherein the recognition order is determined based on emotion levels quantified by the emotion analysis unit.

4. A method for recognizing a telephone call voice, the method being carried out in a telephone call voice recognition system for recognizing a telephone call voice, comprising: The call recorder records the audio of the call, A call recording management device receives the recorded voice data from the call recording device, and transmits the voice data to the voice recognition control device in association with the call information and the call information; The voice recognition device inputs the voice of the call, converts it into text, a voice recognition control device that receives the call information and the voice data from the call recording management device, transmits the voice data to the voice recognition device, and instructs the voice recognition device to convert the voice data into text; a recognition order control unit of the voice recognition control device, which holds a recognition order management table that stores the recognition priority of each call, controls the voice recognition order of each call; a recognition execution instruction unit instructs the voice recognition device to convert the voice data of the call into text; When the recognition execution instruction unit instructs the voice recognition device to perform voice recognition on a call based on the recognition priority stored in the recognition order management table, Furthermore, the telephone call voice recognition method is characterized in that the voice recognition control device holds a site information management table that stores the site priority of sites defined for each telephone number of the destination of the call, and determines the recognition order based on the site priority of the site information management table.

5. A method for recognizing a telephone call voice, the method being carried out in a telephone call voice recognition system that recognizes a telephone call voice, The call recorder records the audio of the call, A call recording management device receives the recorded voice data from the call recording device, and transmits the voice data to the voice recognition control device in association with the call information and the call information; The voice recognition device inputs the voice of the call, converts it into text, a voice recognition control device that receives the call information and the voice data from the call recording management device, transmits the voice data to the voice recognition device, and instructs the voice recognition device to convert the voice data into text; a recognition order control unit of the voice recognition control device, which holds a recognition order management table that stores the recognition priority of each call, controls the voice recognition order of each call; a recognition execution instruction unit instructs the voice recognition device to convert the voice data of the call into text; When the recognition execution instruction unit instructs the voice recognition device to perform voice recognition on a call based on the recognition priority stored in the recognition order management table, The telephone call voice recognition method is further characterized in that the voice recognition control device has a speech time calculation unit that calculates a speech time of a speaker in a telephone call, and determines the recognition order based on the speech time of the speaker in the telephone call.

Images