Instrumental sound recognition and classification

Musical Audio Content Analysis has a wide application area such as structral encoding, sounddatabase search systems, automatic note transcription, musical education etc. The mostimportant part of these applications is the automatic instrument recognition task.In this work an active learning PNN was used to recognise instrumental sounds. LPC andMFCC coefficients with different orders were used as features. Existing active learningalgorithms are not suitable for PNN. Hence, a new active learning algorithm for PNN wasdeveloped.The best feature sets were found by using passive PNNs and these sets were used with activelearning PNNs. By realizing some experiments, it was shown that the entire performance wasimproved by using the active learning algorithmKeywords: Computational auditory scene analysis, Probabilistic Neural Networks, activelearning, instrument recognition.xi