This paper presents a system to align musical notes extracted from an audio signal with the notes of the musical score being played. Building on conventional alignment systems using Dynamic Time Warping (DTW), the proposed method uses fuzzy logic to create the similarity matrix used by DTW. Like a musician following a score, the fuzzy logic system uses high level information as its inputs, such as note identity, note duration and local rhythm. Using high level information instead of frame by frame information reduces substantially the size of the DTW similarity matrix and thus reduces significantly the complexity to find the best path for alignment. Finally, the proposed method can automatically track where a musician starts and stops playing in a musical score.
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.