Review Analysis of Real World Noise

A human listener can discriminate easily between speech and music signals by listening to a short segment (i.e., few seconds) of an audio signal. In the problem of classification of audio signals, the requirements of low-complexity, high-accuracy and short delay are crucial for some practical scenarios. While many efforts have been made in the audio signal classification field, the noise embedded signal problem is seldom concerned so far, especially in many telecommunication applications, where a real-time and noise robust approach is needed. In this paper we review work on real world noise with less computation time.