Abstract:

Mel based cepstral coefficients have been calculated for 30 short (1--2 s) segments of oboe sounds and 30 short saxophone sounds. These were used as features in a pattern analysis to determine for each of these sounds comprising the test set whether it belongs to the oboe or to the sax class. The training set consisted of longer sound segments of 1 min or more for each of the instruments. A k means algorithm was used to calculate clusters for the training data, and Gaussian probability density functions were formed from the mean and variance of each of the clusters. Each member of the test set was then analyzed to determine the probability that it belonged to each of the two classes, and a Bayes decision rule was invoked to assign it to one of the classes. Initial results classifying the oboe sounds have been very good, but are less impressive for the saxophone set. Results of a human perception experiment identifying these same sound segments will be reported.