Audio-Based Indoor Health Monitoring System Using FLAC Features

In this paper, we present a novel methodology for indoor health monitoring based on acoustic information. The system detects the patient´s symptom sounds which are tightly related to health condition. For extracting the features of symptom sounds, we employ the time-frequency feature extraction method by computing local auto-correlations on complex Fourier values (FLAC). At detection stage, we utilize (complex) subspace method to the extracted FLAC features to detect the symptom sounds. To evaluate the proposed method, we conduct experiments under different acoustic environments. The experimental results show the effectiveness of the proposed approach including robust to variation of acoustic environments, high efficiency and promising performance.