Musical source clustering and identification in polyphonic audio
Citations
356 citations
30 citations
24 citations
Cites background or methods from "Musical source clustering and ident..."
...There is comparatively much less work done in unsupervised musical source clustering [19], [20]....
[...]
...work [19] also employed constrained graph clustering over all...
[...]
...The implementation details are elaborated in our earlier work [19]....
[...]
...For source clustering, the framework for three levels of cognitive clustering, as proposed in our previous work [19], is used: 1) Pitched event decomposition: The decomposition of spectra based on multi-F0 values 2) Group object formation: The clustering of temporally connected F0s across successive frames 3) Source streaming: The clustering of all F0s over all the frames using source-characterizing timbre features and constraints like group objects etc....
[...]
...[20] and our previous work [19], the present work proposes a novel Hidden Markov Random Field (HMRF) model to cluster the F0s into source clusters, while taking into account these constraints....
[...]
20 citations
Cites background from "Musical source clustering and ident..."
...Another main reason behind using the CM-based approaches is that, in many cases, especially the speech and music signals, various signal bands are very narrow and appear around a certain range of frequencies [29]....
[...]
10 citations
References
18,802 citations
9,141 citations
8,923 citations
1,975 citations
1,433 citations