IVR Speech Recognition Visual Intent Display
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Live agents at call centers lack sufficient information to quickly understand a caller's intent, as Activity IDs from IVR systems provide limited context, leading to increased inquiry times.
Innovation Solution
An IVR system equipped with a speech recognition mechanism that processes the caller's voice response to a request signal, such as 'how can I help you?', and displays the recognized intent as a visual object on the agent's terminal, using CTI units for data transfer and screen popping functions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If Activity IDs are used to represent caller's actions in IVR system, then data transfer from IVR to call center is enabled, but the information provided is limited and does not allow live agent to understand caller's actual intent
Solution Approach 1:
A speech recognition mechanism is introduced as an intermediary component between the IVR system and the live agent. This mechanism captures the caller's voice response to the request signal and converts it into a data signal that accurately represents the caller's intent. The speech recognition mechanism bridges the gap between the limited Activity IDs and the need for comprehensive intent understanding, providing the live agent with meaningful information without requiring complex changes to the existing IVR architecture.
Solution Approach 2:
The patent replaces the traditional mechanical/manual information capture method (caller navigating through IVR menus and generating Activity IDs) with an automated speech recognition system. Instead of relying on the caller's interaction with predefined menu options, the system uses speech recognition to directly capture and interpret the caller's stated intent, substituting the mechanical navigation process with an automated acoustic and computational process.
2Loss of time
If more information about caller's intent is provided to live agent, then inquiry time is reduced, but system complexity increases
Solution Approach 1:
The system performs preliminary action by capturing and processing the caller's intent information before the call is transferred to the live agent. The speech recognition mechanism processes the voice response and generates the intent data signal in advance, so that when the live agent receives the call, the information is already prepared and ready for immediate display. This preliminary processing reduces inquiry time without requiring the agent's system to handle complex real-time processing.
Solution Approach 2:
The CTI unit serves as an intermediary that facilitates the transfer of processed intent information from the IVR system to the live agent's terminal. It receives the data signal representing the caller's intent and forwards it through the existing CTI infrastructure, leveraging the intermediary's capability to bridge different systems without requiring direct integration or complex point-to-point connections between all components.
3Measurement precision
If speech recognition mechanism is added to process caller's voice response, then accurate caller intent is captured, but device complexity and processing requirements increase
Solution Approach 1:
The speech recognition mechanism is designed to perform multiple functions within a single integrated component. It not only captures the caller's voice response but also processes the speech data, recognizes the intent, and generates the appropriate data signal for transmission. This multi-functional approach consolidates what could be multiple separate systems into one unified mechanism, reducing overall system architecture complexity while maintaining high recognition accuracy.
Data Source
AI summary
A system for providing a live agent with information on a telephone call has an interactive voice response (IVR) mechanism responding to a telephone call placed by a caller by providing a request signal transferred to the caller. The caller may be requested to produce a voice response indicating a purpose of the telephone call. A speech recognition mechanism processes the caller's voice response so as to produce a data signal representing a recognized voice response. A computer telephony integration (CTI) unit forwards to the live agent the telephone call from the caller, concurrently with forwarding the data signal representing the recognized voice response to a terminal for display to the live agent. The CTI unit may perform a screen popping function to display the data signal as a visual object at the terminal of the live agent. The visual object may include a text corresponding to the recognized voice response.