In this paper, we address the problem of automatically recognising and identifying an instrument from a set of solo recordings. A system using the LSF as features whose statistical properties are learnt using the k-means algorithm is described. During the training phase, models are built by determining an optimised codebook of LSF vectors for each class of instruments. During the identification phase, one codebook is similarly extracted from the unknown audio sample. A distortion measure between two codebooks is then used to retrieve the identity of the presented excerpt. System performances are evaluated using a database of 11 instruments.
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.