This Manuscript probe delinquent of classification of uninterrupted of broad-spectrum aural data for content based recovery. This paper is dealing with scheme for classifying aural data & segmentation is also done on same data so that processing rate is faster. Aural data is able to classify into eight categories Simple speech, noise, silence, music single speech with music, double speech with music, speech without music, instrument sound There are so many features are there, among linear prediction coefficient, Mel-frequency Cepstral coefficients etc.
We studied all possible features. Depending upon Cepstral based features which provide accurate classification. To reduce errors aural segmentation is done. So that processing rate is faster & to get more accuracy.