Audio Intent Classification via Semantic Analysis Model
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In human-computer dialogue, existing technologies face challenges in distinguishing between intended audio requests and unintended audio interactions, leading to false awakenings and responses, particularly in scenarios with high volumes of audio requests where not all are genuinely intended by the user.
Innovation Solution
A method and device utilizing a semantic analysis model with independent speech and text feature extraction portions to classify audio interactions, determining whether to reject recognition based on classification results, thereby reducing false responses and improving data processing speed and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If semantic analysis is performed on all audio requests, then response accuracy is improved, but processing time and computational resources increase
Solution Approach 1:
The system performs preliminary classification of audio requests into intended and unintended categories before executing full semantic analysis. By using a classification model to pre-sort requests, the system can skip comprehensive analysis for unintended requests (reducing processing time) while ensuring thorough analysis for intended requests (maintaining response accuracy).
Solution Approach 2:
The processing pipeline is segmented into two stages: a classification stage that separates audio requests into intended and unintended categories, and an analysis stage that performs full semantic analysis only on intended requests. This segmentation allows the system to handle high volumes of audio requests efficiently while maintaining high response accuracy for relevant requests.
2Productivity
If the system processes high volumes of audio requests, then productivity is improved, but false awakenings and responses increase
Solution Approach 1:
The classification model performs preliminary filtering of audio requests to distinguish between intended and unintended requests before full processing. This preliminary action enables the system to maintain high processing throughput by quickly identifying and discarding unintended requests, while ensuring high reliability by thoroughly analyzing only intended requests.
Solution Approach 2:
The system incorporates feedback mechanisms where classification results are continuously refined based on user behavior patterns and contextual information. This feedback loop enables the system to improve its ability to distinguish intended from unintended requests over time, reducing false responses while maintaining high processing capacity.
Data Source
AI summary
A method for processing information includes that: a current audio is acquired, and a current text corresponding to the current audio is acquired; feature extraction is performed on the current audio through a speech feature extraction portion in a semantic analysis model, to obtain a speech feature of the current audio; feature extraction is performed on the current text through a text feature extraction portion in the semantic analysis model, to obtain a text feature of the current text; semantic classification is performed on the speech feature and the text feature through a classification portion in the semantic analysis model, to obtain a classification result; and recognition of the current audio is rejected in response to the classification result indicating that the current audio is to be rejected for recognition.