Noise-Compensated Audio Command Recognition for Electronic Apparatus
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing electronic apparatuses face challenges in accurately identifying users through speech recognition due to noise variations between different sound collection environments, leading to decreased accuracy and increased system burden when attempting to predict all possible noise environments.
Innovation Solution
The apparatus processes a first audio signal and generates a second audio signal by relating it to noise data, using first and second reference data to determine if the second audio signal matches a command, thereby improving user identification accuracy by accounting for noise in the sound collection environment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a user identification model is generated using audio collected in a specific environment, then user identification can be performed, but the accuracy decreases when noise conditions differ from the training environment
Solution Approach 1:
The system performs preliminary noise measurement and noise data generation before user identification. By measuring the ambient noise in advance and generating corresponding noise data, the system prepares a noise-compensated reference model that adapts to the specific acoustic environment, thereby maintaining high identification accuracy across different noise conditions
Solution Approach 2:
The system changes the parameters of the reference model by incorporating measured noise characteristics. The noise data is integrated into the reference model generation process, transforming it from a static model trained in controlled conditions to a dynamic model that reflects actual environmental noise parameters, thus improving adaptability
2Reliability
If multiple noise environments are predicted and noise data is prepared for each, then identification accuracy under various noise conditions improves, but system burden increases significantly
Solution Approach 1:
The system uses the actual ambient noise measured at the time of operation to generate the corresponding noise data, rather than relying on pre-collected noise data from multiple environments. This self-adaptive approach allows the system to serve itself by using real-time environmental information, eliminating the need for extensive pre-processing of diverse noise datasets
Solution Approach 2:
The system transitions from a static approach with fixed pre-collected noise data to a dynamic approach where noise data is generated based on real-time measurements. The reference model adapts dynamically to the current acoustic environment, allowing the system to handle various noise conditions without requiring comprehensive pre-preparation of all possible noise scenarios
Data Source
AI summary
An electronic apparatus including a processor configured to receive a first audio signal, obtain a second audio signal by relating noise to the received first audio signal, identify whether the second audio signal matches a second command obtained by relating the noise to a first command of first reference data, based on second reference data, obtained by relating the noise to the first reference data, and perform an operation based on identification in response to the identifying that the second audio signal matches the second command.