Hearing Aid Sound Classification Using MFCC Feature Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Hearing aids face challenges in classifying environmental sounds effectively due to limited power consumption and real-time operation constraints, especially in complex auditory scenarios, with traditional simple features and heuristic classification schemes resulting in unsatisfactory performance.
Innovation Solution
A sound classification system in hearing aids using advanced classifiers like Gaussian Mixture Model (GMM) and Hidden Markov Model (HMM) with a two-stage classification scheme and a feature selection algorithm, employing low-level and Mel-scale Frequency cepstral coefficients (MFCC) features to distinguish between speech, music, and various noises, while optimizing computational cost.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional simple features and heuristic classification schemes are used, then power consumption is reduced and device complexity is lowered, but classification performance deteriorates in complex auditory scenarios
Solution Approach 1:
The patent applies segmentation by implementing a two-stage classification scheme: first classifying sounds into broad categories (speech, music, non-speech), then further classifying non-speech sounds into specific noise types. This hierarchical approach breaks down the complex classification task into manageable segments, improving overall classification performance while controlling computational complexity at each stage.
Solution Approach 2:
The patent transitions from traditional simple feature sets to Mel-scale Frequency cepstral coefficients (MFCCs), adding a new dimensional representation of acoustic features. This dimensional change enables better discrimination of complex sounds by capturing spectral characteristics in a way that aligns with human auditory perception, significantly improving classification accuracy in challenging auditory scenarios.
2Reliability
If advanced classification algorithms are used, then classification performance is improved, but power consumption increases and real-time operation becomes more difficult
Solution Approach 1:
The two-stage classification architecture segments the computational workload, allowing the system to perform only the necessary level of classification for each sound input. Many sounds are classified in the first stage without requiring the more computationally intensive second stage, thereby reducing average power consumption while maintaining high classification performance when needed.
Solution Approach 2:
The patent optimizes classification parameters by selecting a specific number of MFCC features (5 to 7 features identified as optimal) and tuning classifier parameters to achieve the best trade-off between performance and computational cost. This parameter optimization ensures that advanced algorithms run efficiently within the power constraints of hearing aid devices.
3Measurement precision
If more features are used in classification, then classification accuracy improves, but computational cost increases
Solution Approach 1:
The patent extracts and uses only the most discriminative features from the full MFCC set, specifically identifying that 5 to 7 features provide optimal classification performance. This feature extraction approach removes redundant features, maintaining high classification accuracy while significantly reducing the computational burden compared to using all available features.
Solution Approach 2:
The system employs a partial action strategy by using a subset of features (5-7 out of potentially many) rather than all available features. This partial feature set is sufficient to achieve high classification accuracy for hearing aid applications, avoiding the excessive computational cost of processing the complete feature set while maintaining measurement precision.
Data Source
AI summary
A hearing aid includes a sound classification module to classify environmental sound sensed by a microphone. The sound classification module executes an advanced sound classification algorithm. The hearing aid then processes the sound according to the classification.