Deezer researchers have developed artificial intelligence that can identify emotions in songs.

The work is described in a new article published on Arxiv.org, “Music Emotion Detection Based on Audio Lyrics and Deep Neural Networks.”

To determine the song’s musical mood, the team considered audio signals and lyrics.

Then, to teach it how to determine the mood of a song, they used the Million Song Data Set (MSD), a set of metadata for more than a million contemporary songs.

The Million Songs database only contains metadata for the songs, not the songs themselves, so the team uses identifiers like song titles/artist names and album titles to pair all that information with Deezer’s catalog. About 60% of the resulting data set is used to train the AI, with the rest used to validate and further test the system.

Hello, if you are a future advocate of AI technology, or an industry explorer dedicated to technology convergence, then congratulations, AI Insight can help you. Our wechat official account (Rebuild_ai) will bring you an in-depth understanding of the AI industry and insight into the future trend.