Distributed Speech Recognition Task Allocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Mobile devices lack the processing power to perform complex speech recognition tasks efficiently, while offloading these tasks to servers requires network bandwidth and shared resources, leading to suboptimal performance in terms of response time and accuracy.
Innovation Solution
A distributed network system that allocates speech recognition tasks between mobile devices and backend servers using complexity-based, availability-based, and accuracy-based allocation schemes, combining results from multiple speech recognizers to optimize performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If speech recognition is performed on mobile device, then response time is improved and network usage is reduced, but processing capability and recognition accuracy deteriorate due to limited computing resources
Solution Approach 1:
The patent segments speech recognition tasks by complexity level, assigning simple tasks to mobile device recognizers and complex tasks to backend server recognizers. This segmentation allows the system to leverage both local processing speed and remote processing power, resolving the contradiction between fast local response and accurate recognition.
Solution Approach 2:
The patent merges results from multiple speech recognizers (both mobile device and backend server) through a combination mechanism that aggregates and evaluates multiple recognition outcomes. This merging approach improves overall recognition accuracy while maintaining the speed benefits of local processing.
2Reliability
If speech recognition is performed on server, then recognition accuracy is improved and complex tasks are handled, but network bandwidth consumption increases and response time deteriorates
Solution Approach 1:
The patent segments recognition tasks based on complexity, routing only complex tasks to backend servers while handling simple tasks locally on mobile devices. This segmentation reduces unnecessary network traffic and response time while maintaining high accuracy for complex recognitions.
Solution Approach 2:
The system performs partial speech recognition on the mobile device before potentially involving the backend server. This partial action allows simple tasks to be completed quickly locally without full server involvement, reducing network usage while maintaining accuracy when needed.
3Reliability
If speech recognition tasks are allocated to multiple recognizers, then recognition accuracy is improved, but system complexity and resource coordination overhead increase
Solution Approach 1:
The patent implements local quality by assigning different recognition capabilities to different locations: mobile devices have lightweight recognizers for quick local processing, while backend servers have full-capability recognizers for complex tasks. This local differentiation simplifies the allocation decision-making process.
Solution Approach 2:
The task allocation mechanism dynamically determines which recognizer to use based on real-time factors such as task complexity, network availability, and device capabilities. This dynamic approach adapts to changing conditions without requiring complex static configuration.
Data Source
AI summary
A system, method, computer-readable medium, and computer-implemented system for optimizing allocation of speech recognition tasks among multiple speech recognizers and combining recognizer results is described. An allocation determination is performed to allocate speech recognition among multiple speech recognizers using at least one of an accuracy-based allocation mechanism, a complexity-based allocation mechanism, and an availability-based allocation mechanism. The speech recognition is allocated among the speech recognizers based on the determined allocation. Recognizer results received from multiple speech recognizers in accordance with the speech recognition task allocation are combined.