Europeana Sounds was a project focused on accessing digital audio files. The current dataset aims to provide semistructured data for learning audio-representations using machine learning techniques. The motivation for this dataset is to allow experimentation with audio content and metadata. The current dataset is a subset of a dataset used in this research paper containing 24k audios belonging to different musical genres. Find more information about the Europeana Sounds project and dataset in this website, together with raw audio features. The audio files can be downloaded using the original media URL. Additional data about the object may be obtained in JSON or in RDF. In JSON, by querying with the ID on the Europeana Record API. RDF can be obtained by requesting the URI with HTTP Content Negotiation for a well-known RDF serialization format. The URI also provides access to the object's page at Europeana, if HTTP Content Negotiation is not used. The objects were obtained using the Europeana Search API. More information about Europeana APIs can be found here.