Speech Processing Apparatus Pause Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing speech recognition technologies misrecognize temporary pauses in utterances as completion, leading to erroneous processing results and inability to differentiate between pauses and complete utterances, resulting in suboptimal speech recognition performance.
Innovation Solution
A speech processing method that recognizes delays during utterances as incomplete, temporarily stores the utterance, processes additional information, and combines it with the previous utterance to recognize the complete statement, deleting duplicate parts to improve recognition accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If the device processes spoken utterances in real-time without delay recognition, then response speed is improved, but recognition accuracy deteriorates due to misidentifying pauses as completion
Solution Approach 1:
The system performs preliminary analysis of the utterance stream to detect pause patterns and delay characteristics before making a final recognition decision. By preliminarily identifying incomplete utterances based on pause duration and contextual cues, the system avoids premature processing while maintaining overall response efficiency
Solution Approach 2:
The system dynamically adjusts its recognition threshold and processing timing based on detected pause patterns. When a pause exceeding a dynamically determined threshold is detected, the system extends its waiting period; otherwise, it proceeds with recognition, creating an adaptive response mechanism that balances speed and accuracy
2Measurement precision
If the device waits for complete utterances before processing, then recognition accuracy is improved, but response time deteriorates due to indefinite waiting
Solution Approach 1:
The system performs partial processing of utterances that appear complete based on probabilistic criteria rather than waiting for absolute certainty. By processing utterances that meet a threshold of completeness confidence, the system achieves near-optimal accuracy without the full time cost of waiting for potential continuations
Solution Approach 2:
The system changes the temporal parameters of utterance completion detection based on contextual analysis. By adjusting the maximum waiting time and pause tolerance parameters dynamically, the system optimizes the balance between waiting for complete utterances and responding in a timely manner
3Loss of information
If the device processes additional spoken utterances during an ongoing utterance, then information completeness is improved, but system complexity deteriorates due to multiple concurrent processing streams
Solution Approach 1:
The system segments the speech stream into distinct utterance units based on pause detection and temporal boundaries. By dividing the continuous audio input into discrete, manageable segments that can be independently processed, the system handles multiple utterances without creating complex interdependencies between processing streams
Data Source
AI summary
Disclosed are a speech processing method and apparatus therefor which execute an installed artificial intelligence algorithm and/or machine learning algorithm to perform speech processing in a 5G communication environment. The speech processing method may include determining a temporary pause of reception of a first spoken utterance, outputting a first spoken response utterance as a result of speech recognition processing of a second spoken utterance received after the temporary pause, determining, as an extension of the first spoken utterance, a third spoken utterance received after outputting the first spoken response utterance, deleting a duplicate utterance part from a fourth spoken utterance that is obtained by combining the first and the third spoken utterance, when performing speech recognition processing on the fourth spoken utterance, and outputting a second spoken response utterance as a result of speech recognition processing of the fourth spoken utterance from which the duplicate utterance part has been deleted.