Multimodal Metric Learning for Tag-based Music Retrieval
This work addresses efficient browsing of large-scale music libraries, but it is incremental as it adapts existing metric learning techniques to the music domain.
The paper tackled the problem of tag-based music retrieval by introducing multimodal metric learning with triplet sampling, acoustic/cultural features, and domain-specific word embeddings, achieving quantitative and qualitative improvements in retrieval performance.
Tag-based music retrieval is crucial to browse large-scale music libraries efficiently. Hence, automatic music tagging has been actively explored, mostly as a classification task, which has an inherent limitation: a fixed vocabulary. On the other hand, metric learning enables flexible vocabularies by using pretrained word embeddings as side information. Also, metric learning has already proven its suitability for cross-modal retrieval tasks in other domains (e.g., text-to-image) by jointly learning a multimodal embedding space. In this paper, we investigate three ideas to successfully introduce multimodal metric learning for tag-based music retrieval: elaborate triplet sampling, acoustic and cultural music information, and domain-specific word embeddings. Our experimental results show that the proposed ideas enhance the retrieval system quantitatively, and qualitatively. Furthermore, we release the MSD500, a subset of the Million Song Dataset (MSD) containing 500 cleaned tags, 7 manually annotated tag categories, and user taste profiles.