AI Agent Call Audio Storage via Text-to-Speech Synthesis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Contact centers face significant storage and processing demands due to the need for recording and analyzing voice calls, especially with the increasing capabilities of artificial agents, which can lead to compliance and resource management challenges.
Innovation Solution
The solution involves storing only the customer audio leg and generating the artificial agent audio on demand using synthesis parameters and text, reducing storage requirements and processing loads, while ensuring accurate playback and compliance with regulatory standards.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If full audio recordings of calls involving artificial agents are stored, then compliance and quality analysis requirements are met, but storage space and processing resources are significantly increased
Solution Approach 1:
The patent extracts only the necessary customer audio portions from the full call recording and stores them separately, while storing synthetic agent audio as text-to-speech parameters rather than actual audio waves. This extraction approach retains compliance-essential customer dialogue while eliminating redundant agent audio storage, reducing overall storage requirements by approximately 90%.
Solution Approach 2:
Instead of storing the actual agent audio waveform, the system creates a textual copy of the agent's speech using text-to-speech synthesis parameters. This textual representation serves as a compliant record while consuming minimal storage space compared to preserving the original audio waveform data.
2Reliability
If speech analysis and encryption processing are applied to all recorded audio, then data security and quality insights are improved, but CPU resources and processing time are significantly increased
Solution Approach 1:
The system extracts only customer audio portions for encryption and speech analysis processing, while the synthetic agent audio is processed only as text data. This selective processing approach maintains security for essential customer communication data while dramatically reducing CPU resource consumption compared to processing entire audio recordings.
Solution Approach 2:
The patent replaces complex audio-based processing with simpler text-based processing for agent communications. By converting agent speech to text parameters, the system enables encryption and analysis operations to proceed with minimal computational overhead, substituting heavy audio processing with lighter text processing.
3Quantity of substance
If synthetic agent audio is generated in real-time during calls, then storage requirements are reduced, but processing complexity and system complexity are increased
Solution Approach 1:
The system performs preliminary action by pre-generating and storing text-to-speech parameters for agent responses before the actual call occurs. During the call, these pre-computed parameters are simply retrieved and synthesized, rather than generating audio in real-time from scratch. This preliminary preparation reduces storage requirements while managing processing complexity through efficient parameter retrieval and synthesis.
Data Source
AI summary
Artificial agents utilized for voice interactions continue to improve in their capacity to conduct more sophisticated interactions. Rather than just presenting a limited set of options, artificial agents are continuing to narrow the gap between generated speech and natural human speech. A requirement is often in place that spoken interactions be recorded, however, storing speech, even with data compression, is a resource-demanding task. Generated speech may be provided from content, such as text, and speech data. By recording an identifier of the content and associated speech data, storage processing and space requirements can be greatly reduced. Playback may be provided from a waveform of audio provided by the human participant and by selecting the content associated with the content identifier and generating speech of the content utilizing settings provided by the speech data.