Audio Cache Feedback for Voice Query Error Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Speech recognition engines in television control devices suffer from accuracy issues and incorrect operations due to misidentification of voice queries, leading to frustrating user experiences and increased network traffic.
Innovation Solution
Implement an audio cache on user devices to process common voice queries without speech recognition, disabling incorrect cache entries based on user feedback, and sending ambiguous queries to a speech recognition processor to prevent repeated errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If speech recognition processing is used for all voice queries, then accuracy is improved, but network traffic and processing time increase
Solution Approach 1:
The system segments voice query processing into two paths: common queries are handled by a local audio cache on user devices for immediate processing, while ambiguous or uncommon queries are routed to cloud-based speech recognition processors. This segmentation allows fast local processing for simple cases while maintaining accurate cloud processing for complex cases.
Solution Approach 2:
The audio cache stores pre-processed audio data and corresponding operations for common voice queries in advance. When a user speaks a common query, the system can immediately match it against cached entries without performing full speech recognition, significantly reducing processing time for frequent operations.
2Productivity
If audio cache is used for common queries, then processing speed is improved, but incorrect operations occur due to misidentification
Solution Approach 1:
The system implements feedback mechanisms where user corrections (such as channel changes or repeat queries) are detected and used to disable problematic cache entries for that specific user device. This feedback loop continuously improves accuracy by learning from user corrections while maintaining fast processing for correctly identified queries.
Solution Approach 2:
The system applies different processing qualities to different cache entries based on their reliability for individual users. Cache entries that have been corrected by users are disabled or marked as low-confidence for that specific device, while other entries remain fully functional. This allows the system to maintain high-speed processing for reliable entries while ensuring accuracy for problematic ones.
3Reliability
If cache entries are disabled for all devices to prevent errors, then reliability is improved, but network traffic increases
Solution Approach 1:
Cache entry disabling is applied locally and selectively rather than globally. Each user device has its own cache configuration where only problematic entries are disabled for that specific device. Other devices continue to use the cache entries normally, allowing the system to maintain high reliability for affected users while preserving the energy-saving benefits of caching for the majority of users.
Data Source
AI summary
An audio file associated with a user voice query may be received at a user device. The audio file may be compared to a plurality of references, such as cache entries, corresponding to a plurality of other voice queries. Based on a determination that the voice query corresponds to one of the references, an operation associated with the voice query may be executed. An indication may be received that the operation was not an intended operation associated with the voice query. Based on receiving this indication, the incorrectly identified operation, associated reference, e.g., voice query, may be disabled for the user or the device. However, the cache entry may remain enabled for one or more of a plurality of other devices.