English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Audio embeddings are crucial tools in understanding large catalogs of music.Typically embeddings are evaluated on the basis of the performance they providein a wide range of downstream tasks, however few studies have investigated thelocal properties of the embedding spaces themselves which are important innearest neighbor algorithms, commonly used in music search and recommendation.In this work we show that when learning audio representations on music datasetsvia contrastive learning, musical properties that are typically homogeneouswithin a track (e.g., key and tempo) are reflected in the locality ofneighborhoods in the resulting embedding space. By applying appropriate dataaugmentation strategies, localisation of such properties can not only bereduced but the localisation of other attributes is increased. For example,locality of features such as pitch and tempo that are less relevant tonon-expert listeners, may be mitigated while improving the locality of moresalient features such as genre and mood, achieving state-of-the-art performancein nearest neighbor retrieval accuracy. Similarly, we show that the optimalselection of data augmentation strategies for contrastive learning of musicaudio embeddings is dependent on the downstream task, highlighting this as animportant embedding design decision.

On the Effect of Data-Augmentation on Local Embedding Properties in the Contrastive Learning of Music Audio Representations