Managing large-scale audio libraries

Often large-scale audio libraries contain duplicates, tagging errors and missing metadata. An instant-feedback visual tool can speed up the process of analysing the content and sorting through files.

TEST

A random audio shingle from the AWAL library proved to be a piece of spoken word from a children's story. Spoken word provides an interesting starting point as one would assume that it is difficult to match with music files, and possibly return poorly matched results or no results at all.

Three seconds of "Tortoise and the Hare" by the CRS Players, at 10 seconds into the track returned something of an apparent mistake: 10 exact copies of the same. An analysis of the metadata revealed a list of different albums: "Amelia's Great Adventure in Song and Story"; "Amy's Great Adventure in Song and Story"; "Rebecca's Great Adventure in Song and Story"; "Thomas's Great Adventure in Song and Story" etc. Closer inspection of the nature of the albums (by BF) revealed that iTunes provides identical albums of children's stories customised with a different child's name.

There are exactly 100 copies of the same album with a different child's name attached to it, each with 27 tracks on it. In conclusion, the digital library administrator had stored 243 duplicate tracks by copying what appeared to be different album names.

-- MichelaMagas - 31 Oct 2008

• mHashup CRS Players search: