Résumé |
We propose a system for accelerating the mixing phase in a recording production, by making use of audio alignment techniques to automatically align multiple takes of excerpts of a music piece against a performance of the whole work. We extend the approach of our previous work, based on sequential Montecarlo inference techniques, that was targeted at real-time alignment for score/audio following. The proposed approach is capable of producing partial alignments as well as identifying relevant regions in the partial results with regards to the reference, for better integration within a studio mix workflow. The approach is evaluated using data obtained from two recording sessions of classical music pieces, and we discuss its effectiveness for reducing manual work in a production chain. |