Multi-Microphone Hearing DOA Estimation Using Neural Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing hearing devices struggle to accurately determine the direction of arrival (DOA) of sound sources, which affects their ability to enhance speech intelligibility and provide effective sound processing in challenging listening environments.
Innovation Solution
The implementation of a hearing device with multiple microphones and a neural network-based DOA estimation system that transforms time domain signals into frequency streams to extract directional features, classifying these features to estimate the angles of arrival of sound sources and process sound accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional DOA estimation methods are used in hearing devices, then the device complexity is reduced, but the measurement precision of sound source direction is insufficient
Solution Approach 1:
The patent replaces traditional mechanical signal processing methods with a neural network-based system. The neural network classifier processes frequency transform streams from multiple microphones to determine direction of arrival, substituting conventional algorithmic approaches with machine learning-based processing to achieve superior DOA estimation accuracy.
Solution Approach 2:
The patent transforms time domain signals into frequency domain representations through frequency transform streams, adding a frequency dimension to the analysis. This dimensional transformation enables the neural network to extract directional features more effectively, improving DOA estimation precision by analyzing signals in both time and frequency domains.
2Measurement precision
If multiple microphones are used to improve DOA estimation, then the measurement precision increases, but the device complexity and processing requirements increase
Solution Approach 1:
The patent combines signals from multiple microphones into frequency transform streams that are processed together by the neural network. The microphones are integrated into a unified processing pipeline where their individual time domain signals are transformed and analyzed collectively, enabling accurate DOA estimation while managing system complexity through coordinated processing.
Solution Approach 2:
The patent extracts directional features from the frequency transform streams obtained from multiple microphones. The neural network classifier isolates and processes only the relevant directional information from the multi-microphone signals, separating the essential DOA data from the raw input to reduce processing complexity while maintaining precision.
3Measurement precision
If frequency transform streams are used to characterize phase information, then the DOA estimation accuracy is improved, but the processing time and computational load increase
Solution Approach 1:
The patent performs frequency transformation of the time domain signals into frequency transform streams before neural network processing. This preliminary transformation prepares the data in an optimized format that characterizes phase information effectively, enabling the neural network to quickly extract directional features without excessive processing time during the actual DOA estimation phase.
Data Source
Figure 1
Figure 2~4
Figure 5~6
AI summary
A hearing device has two or more different microphones and a processor operably coupled to the two or more microphones. The processor is operable via instructions to obtain time domain signals from the microphones and transform the time domain signals into respective frequency transform streams. The frequency transform streams characterize phase information in the time domain signals. The processor further determines, from the frequency transform streams, features indicative of directional information of one or more sound sources. The processor inputs the features to a classifier neural network. The classifier network is trained to classify the features into probabilities of angles of arrival of the sound sources. Based on the classification of the features, the processor estimates one or more directions of arrival (DOA) of the sound sources.