KR20010086258A - Controlling navigation paths of a speech-recognition process - Google Patents

Abstract

The present invention relates to a method and a computer program for controlling a navigation path of a speech recognition process in which a user is prompted and an action is taken based on a comparison of a spoken response of a user with a stored speech element. The method includes constructing prompts, actions, and speech elements in a hierarchy of nodes, including group nodes and terminal nodes, and displaying the nodes in a hierarchical list indicating navigation paths of the speech recognition process. The user changes the navigation path in the speech recognition process by editing the node position in the hierarchical list. The speech recognition process includes a calling routine process.

Description

Controlling navigation paths of a speech-recognition process

In business, a major telephone number is often provided to the telephone network as an entry point. Customers who frequently call a major phone number need an assistive device to contact a specific person or department. Voice-based autoattendant uses speech recognition technology to quickly transfer calls from a primary telephone number to a specific local telephone. For example, the automated guidance software asks "What is the name of the employee you want to contact?" And then parse the caller's verbal answer to determine the call destination. Automating these tasks reduces the burden of providing key business phone numbers.

Unfortunately, information management tasks can partially offset the benefits of installing automatic guidance based on voice. For example, scripting automatic guided questions and responses (“call flows”) on the caller's voice is time consuming. Moreover, integrating other speech generation, speech recognition, and phonebook data requires significant database management efforts.

The present invention relates to a method and a computer program for controlling a navigation path of a speech recognition process.

1A to 1D are diagrams describing the automatic guidance function.

2 illustrates a computer platform including an autonomous guidance component.

3 shows an automatic guidance component.

4 shows table correlations in an automated guidance relationship database.

5 shows a hierarchical record.

6 is a flow chart describing the interrelationships between hierarchically organized nodes and call flow navigation paths.

7 shows a screen display of a hierarchical user interface (GUI) for managing automatic guidance.

8A-8D illustrate screen displays of an automatic guidance GUI dialog.

In general, the present invention facilitates the navigation path of the speech recognition process in which an action is taken based on a comparison of the spoken response of the user with a prompted and stored speech element. The present invention recognizes prompts, actions, and speech elements in a hierarchy of nodes, including group nodes and terminal nodes, and displays the nodes in a hierarchical list that indicates the navigation path of the speech recognition process. The user can change the navigation path of the speech recognition process by editing the location of the node in the hierarchical list.

One embodiment of the invention may include one or more of the following features. The navigation path can be changed by interacting with the graphical user interface. The user can edit the position of the node in the hierarchical list by dragging and dropping operations that can edit the position of the node included hierarchically. The user can add, delete or edit. Prompt, action, and voice elements can be recognized as records in the relational database.

The user may display a hierarchically included node of a group node selected from hierarchically included nodes. The user can reduce and expand the group node to change the display of the hierarchical list.

The speech recognition process may be a calling routine process. The call routine process may include sending a call to a telephone premises phone or operating a voice file.

The present invention may have one or more advantages as follows.

Controlling the flow of calls using an intuitive user interface reduces the burden of call flow management. In addition, the ability to control the flow of calls based on hierarchical data not only helps the user to reach the appropriate local telephone, but can also speed up the automated prompt response by narrowing the acceptable prompts to any particular prompt. .

The invention can be performed in hardware, software or a combination thereof. Preferably, the present invention is programmable including a processor, a storage medium readable by the processor (including volatile and nonvolatile memory and / or storage elements), at least one input device, and at least one output device. It is realized as a computer program running on a computer. The program code is applied to the data input by the input device to execute the above functions to generate output information.

Each program is preferably realized in a high level or general purpose programming language to communicate with a computer. However, the program can be realized in assembly or machine language if necessary. In any case, the language can be a compiled or interpreted language.

Each such program is preferably stored on a storage medium or storage device (eg, a ROM or magnetic diskette), the storage medium or storage device being a computer for the storage medium or storage device to execute the procedures described herein. It can be read by a general purpose or special purpose programmable computer to configure and operate the computer when it is read. The system is considered to be realized as a computer readable storage medium composed of computer programs. The storage medium thus configured allows the computer to be executed in a specific and predetermined manner.

Other features and advantages of the invention will be apparent from the following detailed description and claims, including the drawings.

Referring to FIGS. 1A-1D, the autonomous guidance structure 10 transmits an input call 12 received by a switch 20, such as a PBX (premises switched telephone system), to the autonomous guidance unit 24. Automated guidance 24 may request caller questions to determine the premises phone 14, 16, 1 or 22 that the caller would like to contact. The automatic guide 24 controls the switch 20 to connect the input call 12 to the determined office telephone shown in FIG. 1B.

Different extension phones can belong to different groups. For example, the premises telephones 16 and 18 may be the telephones of employees working in the sales department 26. If no description of the hierarchical department / employee interrelationship is provided, the automatic guidance unit 24 may ask the caller which department the caller wishes to contact. After determining the caller's response, the automatic guide 24 sends the call to the premises telephone in the sales department 26 or whether the premises telephone in the sales department 26 that the caller wants to contact is the premises telephone 16 or the premises telephone. (18) The caller may be asked to confirm.

As shown in FIG. 1C, the guidance unit 24 handles calls originating from the on-premises telephones connected to the switch 20, for example when an employee 14 needs to communicate with another employee within a particular department 26. can do. After analyzing the caller's response to the question, automatic guide 24 may connect the caller as shown in FIG. 1D.

Instead of routinely making an input call 12, the automatic guide 24 may perform other call processing. For example, the automatic guide 24 may execute the technical support information or the voice file in the load direction.

Referring to FIG. 2, the automatic guide 24 may include a computer system 32 having a processor 36, a memory 34, and other components such as a bus interface circuit (not shown). Computer platform 24 includes a standard PC-type keyboard 28, a pointing device such as a mouse 30, and a monitor 27. Computer system 32 includes mass storage elements 38 such as CDs, floppy disks, hard disks, and the like. Computer system 32 receives an input call via line card 37. Portions of mass storage element 38 are transmitted to memory 34 and processor 36 during operation.

Mass storage element 32 includes automatic announcement management software 40, data 44, and voice user interface (VUI) software 42. The management software 40 displays automatic guidance information on the monitor 27 and provides a graphical user interface (GUI) that allows the administrator to quickly edit and organize the data 44 to provide the proper call flow. VUI 42 processes input calls based on data 44 arranged by management software 40. The locking technique allows the management software 40 to change the data 44 without interfering with the VUI 42 service.

Referring to FIG. 3, data 44 includes other relational databases 50, 52, but other implementations use a single relational database. Each database 50, 52 corresponds to a different call flow and generates different prompts 54, 58 and grammar files 56, 60.

Prompt files 54 and 58 contain index signal information used by VUI 42 to generate an automatic announcement voice. For example, after accessing relational database 50 or 52 to determine which prompts to act on, VUI 42 may prompt files 54 or 58 to generate specific prompts (eg, thank you for calling our department "). Information may be retrieved Prompt files 54 and 58 may include specific pre-recorded prompts and specific prompts with determined locations.

The grammar files 56 and 60 contain indexed signal information representing other spoken elements that the automatic guide can recognize in response to a particular prompt (eg, a sales department). After analyzing the caller's voice, the VUI 42 accesses the relational database 50 or 52 to determine how the auto attendant should respond (eg, send a call to a local telephone or issue another prompt). can do.

Compiling software 64 generates grammar files 56 and 60 from relational database 50 and 52 records. Compilation can occur either during good run-time intervals or when an administrator initiates an incremental compilation and automatically schedules a dormant compilation.

The management software 40 service manager 66 allows the manager to assign different call flows contained in other databases to different input channels 62. For example, you may have a telephone line set for customer inquiries and another telephone line set for technical support. VUI 42 may examine call transfer file 48 to determine if relational database 50 or 52 provides an assigned call flow. Many channels 62 use the same relational database 50, 52 at the same time.

Referring to FIG. 4, each relational database, such as database 50, includes a call flow configuration record 64. The configuration record 64 stores data describing general parameters such as the type of switch connected to the automatic guide and the language used (eg English or Spanish). The configuration record 64 may store information indicating an office telephone (eg, voice mail or attendant) that processes messages received after business hours, holidays and after business hours.

Configuration record 64 stores a configuration identifier that controls the prompt that VUI 42 uses to ask the caller. For example, a "basic" configuration generates a prompt for a flat configuration that does not look for a local telephone or another group within the group. For example, the basic configuration prompts "tell the person of the department you want to contact" and sends the call based on the caller's response.

The "Department-Name" configuration raises the prompt for a multilevel configuration to look up the local telephone, other groups within the group, and so on. The department-name configuration prompts the caller, "To contact the department, say the name of the department or wait for the list of departments." If the caller responds to the department name, VUI 42 may list the person names belonging to the selected department.

Each relationship database 50 includes a table of measurement records 66. 5, each hierarchical record 66 describes a node within hierarchy 68. As shown in FIG. The node may be a group node 26, 130, 132, 134 or a terminal node. The terminal node may represent a file that includes an extension telephone 14, 16, 18, 24, 140, voice files 136, 138, or other call processing instructions (not shown). Groups 26, 130, 132, and 134 may include some of the other node types hierarchically. Connections between nodes and callers from other navigation paths can navigate the nodes. Referring again to FIG. 4, the hierarchical table 66 record contains a unique identification number used as an index 5 as another database 50 table.

The records in the name item 70 and group item 72 tables describe each hierarchical table 66 record. For example, the name entry table 70 record includes an employee's name and a local telephone, and the VUI 42 can be used to send an incoming call. The group entry table 72 record contains the group name but does not contain the premises telephone data. This is because a group in a department-named configuration does not send a call until another caller asks (eg who do you want to contact?).

The pronunciation table 74 describes the words of the name entry 70 (eg, person's name) and group entry 72 (eg, name of the group) table. For example, the name "Jone Doe" contains two words and is represented by two corresponding pronunciation table 74 records. The pronunciation table 74 record includes a word and a phoneme constituting the word. For example, the phonemes "j", "ah" and "n" describe the word "Jone". The compilation process (64 in FIG. 3) stores the phoneme collected as an entry in the grammar file with the unique identification number of the hierarchy table 66 record corresponding to the pronunciation record 74. When the caller speaks, the VUI 42 detects the phoneme in the caller's voice, examines the phoneme in contrast to the phoneme in the grammar file, and records the hierarchical table 66 record corresponding to the grammar file phoneme that matches the caller's voice. Search. For example, when the caller says "Jone", the VUI 42 finds j-jh-n in the grammar file and searches the relational database 50 hierarchy table 66 for the corresponding hierarchy 66 record. do. If the layer 66 record stores a name, the VUI 42 may connect the caller to the extension telephone stored in the corresponding name entry 7 record. If the layer 66 record stores voice file criteria, the VUI 42 may issue a group-level prompt to ask the caller another question.

The management system 40 stores the database (from Microsoft Excel or Access) from various sources, including comma separated value (CSV) files, and any ODBC data source (e.g., Microsoft Excel or Access) when the administrator supplies the appropriate ODBC (Open Database Connectivity). 50) can be input data. After the administrator links the fields within the data source entered using the Auto Guidance Database fields, the Auto Guidance can load each data source record into the Hierarchy 66, Name Entry 70, and Group Entry 72 records. have. The automatic guide generates a corresponding pronunciation table 74 record. The administrator can specify whether an input record (eg, information of a particular person) overlaps or is ignored with an existing record. By entering an existing human resource data file, the administrator can quickly get started using automatic guidance without difficult data entry.

The database 50 contains data that controls the prompts that the VUI 42 acts on the caller. The automatic announcement data 44 includes a pre-recorded prompt that responds to certain events (i.e., the caller misses the response or the caller speaks with a pronunciation that is not in the Grammar file). 5, the prompt corresponds to caller navigation to another layer node (record). For example, navigation to configuration node 64 may trigger a message that tells the caller that the call occurred after business hours. Navigation to group nodes 26, 130, and 132 can trigger a group prompt that tells the caller to select a particular employee or other node in the group. Each node can have several associated prompts. VUI 42 may select a prompt based on the caller's action. For example, VUI 42 may track how many times a caller fails to respond and executes a series of other prompts before ending the call.

Referring to FIG. 4, the template prompt table 73 stores reference statements in prompts stored in advance in a prompt file. Administrators can write on pre-stored prompts that generate invalid Prompt Table 75 records that contain specific department information and refer to various prompts in the prompt file. During the call, VUI 42 checks the prompt table 75 for the prompt record before checking the template prompt table 73. The VUI 42 may then retrieve the corresponding prompt file information to generate an automatic announcement voice.

Referring to Figure 6, the call flow follows a hierarchy defined in the database. After receiving the input call, the VUI 42 sets the caller's position at the root of the hierarchy, ie, the configuration node 64 (110). VUI 42 then reserves the caller with a "conversation" that controls navigation through layers 114-124.

As described, each node has a set of associated prompts. VUI 42 executes a prompt for the current caller node location 112 based on the caller's behavior (how many times the caller visits the same node). VUI 42 analyzes the caller's response to the prompt by examining the node's associated grammar file 114. VUI 42 identifies the hierarchy table 66 record corresponding to the caller's voice. If the caller is designated, name record (i.e., a record about the premises phone) 118, the automatic guide can transmit the call 120. If the caller specifies a group record, VUI 42 connects the caller to the group node corresponding to the caller's voice 124 and initiates prompt / response exchange Again 112.

Referring to FIG. 7, the description of the automatic announcement data structure and call flow seems complicated. Fortunately, management software 40 significantly reduces the complexity of managing the system.

The management software 40 includes a graphical user interface (GUI) 84 constructed from other Microsoft creation class (MFC) controls (eg, buttons, list controls, and dialogs). In addition to providing an intuitive relational database management system, the GUI 84 provides administrators with the ability to quickly define and change the database node hierarchy (FIG. 6). GUI 78 provides a menu bar 86, a toolbar button 88, a parallel display including a hierarchical list display 90 and a hierarchically included node 92 of a group selected from the hierarchical list display 90. It includes a display.

Hierarchical list display 90 shows the outline of the call flow included in the node hierarchy. Hierarchical list display 90 lists the names of configurations and hierarchy nodes. Each listed name represents a folder icon 94 and a sign 93 (eg, "+" or "-"). The sign 93 indicates whether the hierarchical list display 90 shows nodes included in a particular node.

Expanding the hierarchical node in the hierarchical list display 90 (clicking the list " + " sign 93) expands the hierarchical list display 90 to show the nodes hierarchically included in the expanded node. For example, expanding group node 96 generates a hierarchical list display 90 that includes a list of group nodes 96 included in association with expanded group node 96. Closing a hierarchy node (e.g., clicking a list "-" sign) hides the node in the hierarchy hierarchy display from hierarchy list display 90. For example, closing group node 96 hides group node 95 from the display on hierarchical list display 90.

The administrator manipulates the group from the hierarchical list display 90. For example, an administrator can add and delete group nodes from the configuration. Hierarchical list display 90 provides "drag and drop" capabilities. For example, the administrator can drag the selected group to another group. While providing the caller with more levels of navigation, this technique helps to window the caller with a large amount of information, premises telephone and technical support information. The response that VUI 42 needs to consider to determine the Grammer file is matched with the response of the caller response rate VUI 42.

The hierarchically included node display 92 shows the contents of the selected hierarchical list display 90 element. For example, selecting a group node 97 in the hierarchical list display 90 causes the selected group element icon for the open folder, the list of hierarchically included groups, and the premises phone of the hierarchically contained node display 92 to be selected. Is changed. Display 92 may include node information (eg, name, local telephone, or notice). Display 92 includes management information for each node. For example, if an employee does not record the phonetic notation of his name, display 92 may indicate the above by marking the node with a point of admiration (not shown). The administrator can sort the display 92 by various criteria such as alphabetical order when nodes are added to the hierarchy and the like.

Administrators can add, delete, and edit display 92 elements. The administrator moves the element (ie, group or name) to another location in the call hierarchy by dragging and dropping the element into a hierarchically included node display 90 or another group in the hierarchical list display 92. The management system 40 changes the database contents based on these operations. This allows administrators to quickly recognize data and change the call flow.

8A and 8B, the GUI dialog facilitates management of database information. For example, selecting the configuration node (FIG. 7) for editing generates the tab dialog shown in FIGS. 8A and 8B. In FIG. 8A, an administrator can edit information in a dialog field describing a configuration record. In FIG. 8B, an administrator may change the configuration level prompt message sent by VUI 42 in response to an event caused by navigating to a configuration node that includes initial call processing.

When the dialog "OK" button 100 is selected, the edited information is stored in the relational database, in which case the configuration record is potentially updated and a new prompt record is added. Management software is additionally recorded when database changes occur to coordinate record locking. "Close" button 102 discards the editor. The GUI provides the administrator with the "Keep Changes Made" dialog when the administrator exits the administration section. Database concepts similar to the database concept will recognize that the "OK" button is similar to SQL INSERT but the "Keep Made Changes" dialog causes the dialog to accept or roll back.

However, this similarity to the database concept is unnecessary for system administration because the GUI provides the database concept in a dialog button similar to an accidental word processing user.

Referring to FIG. 8C, another dialog edits group node information and optionally records pronunciation of the group name. Changing the group node information may change the node description of the hierarchical table that potentially adds, deletes, or modifies prompts and pronunciation records. Again, the management system 40 hides the database change from the manager to the stable automatic guidance management.

In Fig. 8D, when a name node is selected, a name specific dialog is generated. In this dialog, the administrator can change the phone of the employee's premises or change the phonemes that make up the employee's name. Similar to the group node dialog, the administrator records the pronunciation of the employee name or the management software generates the pronunciation based on the spelling. The management software 40 allows an individual employee to record the pronunciation of his or her name remotely (ie from any phone). Any change generates or changes the pronunciation record and generates an entry in the prompt file.

Other embodiments are implemented within the scope of the appended claims. The above description is not limited to the automatic guidance function and can be applied in various ways.

Claims (22)

A method of controlling a navigation path of a speech recognition process in which a prompt is provided to a user and an action is taken based on a comparison of a user's spoken response and a stored speech element, the method comprising: Organizing prompts, actions, and voice elements into a hierarchy of nodes, including group nodes and terminal nodes, Displaying the node in a hierarchical list indicating a navigation path of the speech recognition process; Changing a navigation path of the speech recognition process in response to a user edit of a node location in the hierarchical list. The method of claim 1, Changing the navigation path comprises interacting with a graphical user interface. The method of claim 1, And editing the position of the node in the hierarchical list comprises drag and drop operations that comprise editing the position of the hierarchically included node. The method of claim 1, Organizing prompts, actions, and voice elements into the hierarchy of nodes comprises organizing records in a relational database. The method of claim 1, Organizing prompts, actions, and voice elements into the hierarchy of nodes includes adding, deleting, or editing nodes. The method of claim 1, And displaying the hierarchically included nodes of the group nodes selected from the individual list of hierarchically included nodes. The method of claim 1, The group node in the hierarchical list can be expanded to display hierarchically included nodes or shrunk to hide the hierarchically included nodes. The method of claim 1, And the speech recognition process comprises a calling routine process. The method of claim 8, And said terminal node comprises a local telephone. The method of claim 9, The operation includes sending a call to a local telephone when the user navigates to the local telephone terminal node. The method of claim 8, And said terminal node comprises a voice file. A computer program resident on a computer readable medium that prompts a user and controls a navigation path of a speech recognition process in which an action is taken based on a comparison of a spoken response of a user with a stored speech element. The computer program, Commands for organizing the prompt, action, and voice elements into a hierarchy of nodes including group nodes and terminal nodes, Displaying the node in a hierarchical list indicating a navigation path of the speech recognition process; Instructions for changing a navigation path of the speech recognition process in response to user editing of node locations in the hierarchical list. The method of claim 12, And the command to change the navigation path includes a command to interact with a graphical user interface. The method of claim 12, And the instructions for editing the position of the node in the hierarchical list comprise drag and drop operations for constructing instructions for editing the position of the hierarchically included nodes. The method of claim 12, And instructions for configuring the prompt, action, and voice elements in a hierarchy of nodes includes instructions for organizing records in a relational database. The method of claim 12, And instructions for configuring the prompt, action, and voice elements in a hierarchy of nodes includes instructions for adding, deleting, or editing the node. The method of claim 12, Further comprising instructions for displaying the hierarchically included nodes of the group nodes selected from the respective list of hierarchically included nodes. The method of claim 12, The group nodes in the hierarchical list may be expanded to display hierarchically included nodes or contracted to hide the hierarchically included nodes. The method of claim 12, And the speech recognition process comprises a calling routine process. The method of claim 19, The terminal node comprises a local telephone. The method of claim 20, The operation includes instructions for sending a call to an extension telephone when the user navigates to an extension telephone terminal node. The method of claim 19, And the terminal node comprises a voice file.