Abstract | ||
---|---|---|
This paper presents an intra-note segmentation method for mono-phonic recordings based on acoustic feature variation; each musical note is separated into onset, steady and offset states. The task of intra-note segmentation from audio signals is detecting change points of acoustic feature. In proposed method, the Markov process is assumed on state transition, and time-varying acoustic feature is represented by three Dirichlet processes (DP) that are emitted by the each state. In order to express the generative process, the sticky hidden Markov model (HMM) with DP emission is employed. This modeling allows us to automatically estimate the state transition while avoiding the model selection problem by assuming countably infinite of possible acoustic feature in musical notes. Experimental result shows that the detection accuracy of onset-to-steady and steady-to-offset were improved 2.3 points and 20.7 points from previous method, respectively. |
Year | DOI | Venue |
---|---|---|
2014 | 10.1109/ICASSP.2014.6853978 | Acoustics, Speech and Signal Processing |
Keywords | Field | DocType |
audio signal processing,hidden Markov models,music,musical instruments,DP emission,Dirichlet processes,Markov process,acoustic feature variation,audio signals,hidden Markov model,intranote segmentation,model selection problem,monophonic recordings,sticky HMM,Dirichlet process,hidden Markov model,intra-note segmentation,music information retrieval | Audio signal,Markov process,Pattern recognition,Segmentation,Computer science,Markov model,Model selection,Artificial intelligence,Dirichlet distribution,Musical note,Hidden Markov model | Conference |
ISSN | Citations | PageRank |
1520-6149 | 0 | 0.34 |
References | Authors | |
8 | 2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Koizumi Yuma | 1 | 41 | 11.75 |
Katunobu Itou | 2 | 319 | 44.36 |