The Realization and Comparison of Audio Alignment Algorithm based on Feature Extraction

摘要

The purpose of audio monitoring can be achieved through audio comparison technology. This article mainly studies the audio comparison algorithm based on feature extraction. Aiming at the problem of audio similarity calculation, a weighted comprehensive similarity algorithm was proposed, considering the structural similarity of audio. The short time average magnitude and Mel frequency cepstrum coefficient(MFCC) of physical characteristics and pitch of the perceptual features were extracted. Analytic hierarchy process was used to perform weight distribution, and the final audio similarity was calculated. Experimental results show that compared with the dynamic time warping (DTW) algorithm and the matrix similarity algorithm based only on MFCC parameters, the weighted comprehensive similarity calculation method has better accuracy and stronger anti-interference ability.