In creative media, visual imagery is often combined with music soundtracks. In the resulting artefacts, the consumption of isolated music or imagery will not be the main goal, but rather the combined multimedia experience. Through frequent combination of music with non-musical information resources and the corresponding public exposure, certain types of music will get associated to certain types of non-musical contexts. As a consequence, when dealing with the problem of soundtrack retrieval for non-musical media, it would be appropriate to not only address corresponding music search engines in music-technical terms, but to also exploit typical surrounding contextual and connotative associations. In this work, we make use of this information, and present and validate a search engine framework based on collaborative and social Web resources on mass media and corresponding music usage. Making use of the SRBench dataset, we show that employing social folksonomic descriptions in search indices is effective for multimedia soundtrack retrieval.
|Name||Lecture Notes in Computer Science|
|Publisher||Springer International Publishing AG|
|Conference||20th International Conference on Theory and Practice of Digital Libraries, TPDL 2016|
|Period||5/09/16 → 9/09/16|