Real-Time Speech Analytics for Automated Call Routing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Automated dialing devices struggle in B2B environments and wireless networks due to inaccurate call progress analysis, often mistaking auto attendants and voice mail for answering machines, leading to reduced productivity and efficiency.
Innovation Solution
Implementing real-time speech analytics to navigate calls by detecting keywords and sending appropriate signals or messages to advance the call process, allowing automated dialing devices to differentiate between machines and live parties, and interact with auto attendants or voice mail systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If automated dialing devices use traditional call progress analysis to screen calls, then calls answered by answering machines can be filtered out, but calls to B2B environments are incorrectly screened out because auto attendants and voice mail systems are misclassified as answering machines
Solution Approach 1:
The patent changes the parameters used for call classification by introducing speech analytics that analyze audio characteristics, keyword detection, and call behavior patterns. Instead of relying on traditional CPA parameters that cannot distinguish between answering machines and business automated systems, the system uses new parameters such as speech recognition results, keyword presence, and call navigation patterns to accurately identify call types and route them appropriately.
2Loss of time
If automated dialing devices screen out calls reaching machines and services, then agent time is freed for live parties, but the vast majority of calls to businesses are incorrectly screened out requiring manual dialing
Solution Approach 1:
The patent introduces speech analytics technology as an intermediary between the automated dialing device and the call routing decision. This intermediary analyzes the call audio in real-time, detects keywords and speech patterns, and provides nuanced classification information that enables the system to distinguish between answering machines and business automated systems, thereby making more accurate routing decisions without losing productive calls.
3Ease of operation
If traditional call progress analysis is used on wireless networks, then calls can be monitored, but audio quality issues and network delays cause inaccurate detection of call status
Solution Approach 1:
The patent applies preliminary action by implementing robust audio preprocessing and error tolerance mechanisms before analysis. The speech analytics system compensates for wireless network issues by using multiple detection methods, buffering audio data to account for delays, and applying algorithms that are resilient to audio quality degradation, thereby maintaining accurate call status detection despite network challenges.
Data Source
AI summary
Various embodiments of the invention provide methods, systems, and computer program products for using real-time speech analytics to navigate a call that has reached a machine or service. In various embodiments, a call leg is established from a call handler handling the call to a speech analytics system configured to analyze the call to detect keywords. As the speech analytics system monitors the call, the speech analytics system sends an event to the call handler upon detecting a keyword. In turn, the call handler carries out some action based on the event that is configured to advance the progress of the call. Accordingly, if the call handler determines a live party has been reached on the call as result of the action, the call handler connects the call with a second live party to converse with the live party reached on the call.