Temporal Segmentation and Pre-analysis for Non-linear Time-scaling of Audio.
We present a method to achieve good segmentation of note events for use with non-linear time scaling algorithms, greatly reducing artefacts due to both rhythmic distortions and soft note transitions being treated as percussive transients. The proposed algorithm isolates percussive transients as a subset of note-onsets, leading to a more meaningful segmentation. A subband based hybrid onset detection algorithm forms the basis of this segmentation scheme. A new frequency content distance measure, automatic threshold setting and subband result validation are all key elements of this scheme. At the subband re-combining stage the algorithm differentiates between note onsets, which may appear in one or more subbands, from percussive transients that appear in multiple subbands.
Click to purchase paper 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 $20 for non-members, $5 for AES members and is free for E-Library subscribers.