Speech Hypothesis Preservation for Cross-Device Transcription Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Automated assistants often misrecognize spoken utterances, leading to prolonged human-computer dialogs and unnecessary computational resource consumption due to manual corrections by users.
Innovation Solution
Preserve multiple speech hypotheses across computing devices and dialog sessions, allowing automatic correction of misrecognitions and reducing the need for manual user intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If speech recognition is performed to generate multiple term hypotheses, then transcription accuracy can be improved, but computational resources and dialog time are consumed
Solution Approach 1:
The system performs speech recognition and generates multiple term hypotheses in advance, storing them for later use. This preliminary action allows the system to have correction options ready before the user needs them, reducing the need for repeated dialogs and manual corrections.
Solution Approach 2:
The system creates copies of multiple term hypotheses and stores them in memory. When a misrecognition is detected, the system can quickly retrieve and present the copied hypotheses to the user, avoiding the need to re-run speech recognition and reducing dialog time.
2Reliability
If multiple speech hypotheses are stored and preserved across devices, then misrecognition correction capability is improved, but memory usage and system complexity increase
Solution Approach 1:
The system designs the hypothesis storage and management mechanism to serve multiple functions: storing term hypotheses, preserving them across devices, and enabling correction. This multi-functional approach reduces the need for separate systems for each function, thereby managing complexity while improving reliability.
Solution Approach 2:
The system introduces an intermediary component (hypothesis management system) that mediates between speech recognition and the final transcription. This intermediary handles the complexity of storing, preserving, and retrieving multiple hypotheses, isolating the complexity from the core speech recognition and transcription processes.
3Measurement precision
If users manually correct misrecognitions, then transcription accuracy is improved, but user time and computational resources are wasted
Solution Approach 1:
The system implements feedback by presenting multiple term hypotheses to the user when a misrecognition is detected. The user can select from the presented hypotheses, providing feedback that immediately corrects the transcription error without requiring manual re-typing or repeated speech input, thereby improving both accuracy and user efficiency.
Solution Approach 2:
The system enables self-service correction by automatically detecting potential misrecognitions and presenting correction options to the user. This allows the user to correct errors with minimal effort (selecting from presented options) rather than requiring full manual correction, improving productivity while maintaining accuracy.
Data Source
Figure 1
Figure 2
Figure 3A
AI summary
Implementations can receive, at a computing device, audio data corresponding to a spoken utterance of a user, process the audio data to generate, for one or more parts of the spoken utterance, a plurality of speech hypotheses, select a given one of the speech hypotheses, cause the given one of the speech hypotheses to be incorporated as a portion of a transcription associated with the software application, and store the plurality of speech hypotheses. In some implementations, the plurality of speech hypotheses can be loaded at an additional computing device when the transcription is accessed at the additional computing device. In additional or alternative implementations, the plurality of speech hypotheses can be loaded into memory of the computing device when the software application is reactivated and/or when a subsequent dialog session associated with the transcription is initiated.