The access to audio data available in huge volume on public networks like Internet requires an efficient indexing and annotation mechanism. Non-stationary nature and discontinuities in audio signal had made the segmentation and classification of audio signal difficult. Also the difficulty in extracting and selecting optimal features in audio signal, automatic music classification and annotation is a challenging task. Audio classification and retrieval systems are used in application areas like speaker recognition, gender classification, music genre classification, etc. One of the major challenge in developing audio retrieval systems is identifying appropriate content-based features for representation of the audio-signals. Hence, we have proposed a solution which segments, extracts features and classify the audio signals.