Salar Shakibhamedan - Electrical Engineering K.N. Toosi University of Technology Tehran, Iran
Seyed Kooshan Hashemifard - Electrical Engineering K.N. Toosi University of Technology Tehran, Iran
Farhad Faradji - Electrical Engineering K.N. Toosi University of Technology Tehran, Iran
Mansour Vali - Electrical Engineering K.N. Toosi University of Technology Tehran, Iran

In this paper, a Persian musical instrument recognition system is proposed. The system listens to the polyphonic music and recognizes the musical instruments being played. It has two stages. The first stage is the blind source separation stage, while the second stage is the feature extraction and classification stage. FastICA is used in the blind source separation stage. The feature vector is a 80-vector based on the Mel-frequency cepstral coefficients, their first and second derivatives, and the audio spectrum flatness coefficient. Classification is a two-step process. In the first step of classification, the family of the musical instrument is recognized. In the second stage of classification, the musical instrument is then recognized. The accuracy rate obtained is 100% for the testing data if the information of the six seconds of the music sound signal is used.

BSS; FastICA; MFCC; ASF; feature extraction; SVM; classification