1. Valero-Mas, J.J.; Iñesta, J.M.

, vol. Late Breaking/Demo extended abstract, 16th International Society for Music Information Retrieval Conference (ISMIR), pp. . Málaga, Spain (2015)

Onset detection still has room for improvement. State-of-the-art onset detection algorithms achieve good results for a range of applications, but for some situations in which the accuracy is a must, human intervention is required to correct the mistakes committed. In such scheme, accuracy in the result is guaranteed at the expense of the manual correction of all errors. Hence, the issue now lies on finding schemes for efficiently exploiting and reducing that user effort. In this work we present an Interactive Pattern Recognition approach for tackling this issue: using a pre-trained classification-based onset detection algorithm, every time the user corrects an error in the estimation, the system modifies its performance accordingly and recalculates the output. Initial results show that user effort is effectively reduced under our proposal.