Voice Utterance Processing System Domain Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current voice recognition systems in electronic devices face challenges in accurately interpreting user voice utterances without explicit service domain names, leading to potential misinterpretation and incorrect execution of user intents.
Innovation Solution
The system employs a voice utterance processing system that includes a user terminal, an intelligent server, and a service server, utilizing natural language understanding (NLU) models to determine user intents and select appropriate service domains, either based on explicit domain names or by predicting domains from context, ensuring accurate execution of user commands.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the system uses traditional voice recognition without service domain selection, then the operation is simpler, but the accuracy of interpreting user intents deteriorates
Solution Approach 1:
The system segments the voice processing task into distinct stages: service domain detection, NLU model selection, and intent recognition. By dividing the processing pipeline, each component can be optimized independently, improving overall accuracy without proportionally increasing complexity.
Solution Approach 2:
The system performs service domain detection before intent recognition. By pre-selecting the appropriate service domain based on the voice input, the system narrows down the context beforehand, which significantly improves the accuracy of subsequent intent interpretation without requiring the full complexity to be active simultaneously.
2Adaptability or versatility
If the system pre-selects multiple service domains, then the adaptability to different user needs improves, but the difficulty of detecting and measuring the correct domain increases
Solution Approach 1:
The system uses confidence scores as feedback to evaluate the quality of domain detection. When the confidence score exceeds a threshold, the detected domain is confirmed; otherwise, the system falls back to default behavior or requests clarification. This feedback mechanism improves adaptability while managing detection difficulty through quantitative evaluation.
Solution Approach 2:
The system changes the parameter of domain detection by using confidence score thresholds to dynamically adjust the decision-making process. By introducing this quantitative parameter, the system can adaptively select among multiple pre-configured service domains based on the clarity of the voice input, balancing versatility with detection accuracy.
3Measurement precision
If the system requires explicit service domain names in voice commands, then the measurement precision of domain identification improves, but the ease of operation deteriorates
Solution Approach 1:
The system performs self-service by automatically detecting and selecting the service domain without requiring the user to explicitly state it. The voice processing system analyzes the input and autonomously determines the appropriate domain, eliminating the need for users to remember or use specific domain names while maintaining high identification precision.
Solution Approach 2:
The service domain detection module acts as an intermediary between the user's natural language input and the NLU model selection. This intermediary layer translates implicit user intent into explicit domain identification, allowing users to speak naturally without knowing technical domain names while ensuring precise domain recognition.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A system is provided. The system includes at least one memory, and at least one processor operatively connected to the memory. The memory may store instructions that, when executed, cause the processor to receive an input of selecting at least one domain from a user and store the input in the memory, recognize, at least partially based on data regarding a user utterance received after the input is stored, the utterance, determine, when the utterance does not comprise a domain name, whether or not the utterance corresponds to the selected domain, and generate a response by processing the utterance by using the selected domain when the utterance corresponds to the selected domain.