Implementation of Audio Comparison Based on Image Recognition Technology

In this paper an image features based method for audio comparison has been proposed. During audio comparison, converting audio to the graphic is the principal step for this method. After that, the modulus maxima acquired by transform wavelet and feature points generated by using the Harris operator can both be regarded as feature parameters of audio. It is necessary to remove the noise generating by wavelet transform modulus maxima or feature points. The rest still keep the most important feature of audio. Lastly, the audio comparison is accomplished according to the relationship between feature parameters and the threshold value by formatting the distance function or computing the correlation degree of feature points. Compared with the traditional program, it is more intuitive but also simpler on computation.