Timbre and melody features for the recognition of vocal activity and instrumental solos in polyphonic music

We propose the task of detecting instrumental solos in poly-phonic music recordings, and the usage of a set of four audio features for vocal and instrumental activity detection. Three of the features are based on the prior extraction of the pre-dominant melody line, and have not been used in the context of vocal/instrumental activity detection. Using a support vector machine hidden Markov model we conduct 14 exper-iments to validate several combinations of our proposed fea-tures. Our results clearly demonstrate the benefit of combin-ing the features: the best performance was always achieved by combining all four features. The top accuracy for vocal activity detection is 87.2%. The more difficult task of de-tecting instrumental solos equally benefits from the combi-nation of all features and achieves an accuracy of 89.8 % and a satisfactory precision of 61.1%. With this paper we also release to the public the 102 annotations we used for train-ing and testing. The annotations offer not only vocal/non-vocal labels, but also distinguish between female and male singers, and different solo instruments