Expected Next Prompt Audio Caching for Reduced Voice Latency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Mobile computing systems, particularly voice systems, experience latency and inefficiencies during text-to-speech conversion, leading to prolonged response times and reduced productivity in environments like warehouses.
Innovation Solution
Implementing a voice system that generates and employs expected next prompts based on historical data and predefined sequences, using signal and waveform caching to reduce latency by pre-generating audio signals for likely user responses, and comparing these to actual responses to determine prompt delivery.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If text-to-speech conversion is performed in real-time, then the system can respond dynamically to user inputs, but the response time increases due to processing latency
Solution Approach 1:
The system pre-generates audio signals for expected next prompts before they are actually needed. By anticipating likely user responses and preparing corresponding audio signals in advance, the system eliminates real-time text-to-speech conversion latency, thereby reducing response time while maintaining productivity
2Loss of time
If audio signals are pre-generated and cached, then response time is reduced, but system complexity increases
Solution Approach 1:
Audio signals for expected prompts are generated and stored in a cache before being needed. This preliminary action allows the system to retrieve pre-generated audio instead of converting text in real-time, reducing response time while the added complexity is managed through structured caching mechanisms
Solution Approach 2:
Instead of generating unique audio signals for each possible prompt, the system creates copies of audio signals for expected prompts and stores them in a cache. This copying approach reduces the computational burden of real-time generation while providing quick access to frequently needed audio signals
Data Source
AI summary
Various embodiments described herein relate to generating and/or employing an expected next prompt to reduce response time for a voice system. In this regard, a candidate audio signal is generated for a predicted prompt to be presented via a user audio device. Additionally, in response to audio response data provided by the user audio device, the predicted prompt is compared to a prompt associated with the audio response data. In response to a determination that the predicted prompt matches the prompt, the candidate audio signal is presented via the user audio device.