r1 - 17 Sep 2008 - 15:04:58 - TimCrawfordYou are here: OMRAS2 >  Main Web  >  TWikiUsers > MichaelCasey > AudioDB > AudioDBDeveloperDiscussionPage > ADBSimilarPassage

Use Case Scenario: Find similar/same passage(s)

Short Description

Given a user-selection in a track, identify the same passage, or those which are very similar, elsewhere in the track and in other tracks in the database. A variant is to select a passage containing a prominent 'motif' and search for other occurrences of this motif.

Various well-recognised musical entities are based on this, such as the rondo, conventional 'sonata form' and the fugue; the latter two definitely depend on transposition-invariant matching (though usually with very specific constraints). The motif variant is exemplified, of course, by the Wagnerian leitmotif, for which there is an extensive (though not very systematically compiled) literature.

Beneficiaries

Musicologists, but also others (?)

Task Description using fftExtract and AudioDB

Related to detection of sampled passages in remixes/mashups, but will require feature such as chroma rather than mfcc.

Data Set and data set preparation

Ground Truth

Sets of 'ground truth' (dangerous term) can be compiled by reference to scores, though for automatic evaluation we would need some mechanism for audio score-tracking or manual pre-labelling of audio.

Experimental Method

Results Tables and Figures

Conclusions

-- TimCrawford - 17 Sep 2008

Edit | Attach | Printable | Raw View | Backlinks: Web, All Webs | History: r1 | More topic actions
 
EPSRC OMRAS2
This site is powered by the TWiki collaboration platformCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding OMRAS2? Send feedback