An Approach to Sound Feature Extraction Method Based on Gammatone Filter

摘要

A feature extraction method for specified sound based on Gammatone filter is researched in this paper, which can be applied to the audio security system. Firstly, audio data was carried out fourier transform, and then its spectral energy was calculated. Secondly, the inverse discrete cosine transform was computed, after the spectral energy was filtered by use of Gammatone filter, so the feature vector set was constructed. Finally, the audio data are trained and tested by making use of SVM. The experimental results show that the extracted audio features are effective and reasonable, and lead to satisfactory classification and detection performance.