Earlier this week, we announced a major update to our Tag API. The Moods 3.0 update introduces a new layer of granularity to our industry-leading AI music tagging product. Musiio Tag now recognises 41 musical moods, giving anyone searching for music a powerful new way to find tracks with the specific moods they seek.
Training an AI to understand more nuanced and specific moods
“Our customers loved our original V1 Mood classifier,” says Hazel Savage, SoundCloud’s VP of Music Intelligence, ”It’s been providing solid data on over 1,000,000 tracks a day for two years, but Mood 3.0 demonstrates our commitment to developing and making available the latest in AI technology for the music industry.”
Both our standard and enhanced mood sets (respectively, V1 and V3 Moods for the devs in the room) are over 90% accurate. And because the enhanced mood set delivers up to four tags per track, that’s a total of six when combined with the standard mood set. Having more tags with greater nuance is good news for musicians, presenting even more chances for their music to be discovered.
To demonstrate how enhanced moods work, we’ve tagged chart hits using both standard and enhanced moods. Try the demo for yourself.
For a top-level mood analysis, Energetic and Romantic are a perfect fit for Beyoncé’s soulful house-inspired track. Meanwhile, Dancy, Playful and Confident give a nuanced picture of Queen B's current sound.
For playlists, you can mix and match between standard and enhanced moods, to produce increasingly refined results, even from large catalogues.
There’s no doubt that Harry Styles’ massive track ‘As It Was’ ticks the box for Happy and Quirky from a sonic perspective. When we drill down with the enhanced mood tags, we get a more detailed picture of the former 1D star’s sound. Dreamy and Warm are especially apt for this indie-influenced pop track, and Calm is bang-on for the verses, which give the chorus the dynamic headroom to pop.
For a Dreamy, Quirky Pop playlist, curators can combine the two mood classifiers with genres to refine thousands of results into a manageable list of candidate tracks.
Bursting into the charts 37 years after its initial release thanks to a memorable Stranger Things placement, ‘Running Up That Hill’ is a timeless 80s cut. It registers as Dark and Dramatic: both perfectly accurate mood descriptors for Bush’s unusual song.
Then, zooming in, Euphoric, Dreamy and Inspiring are equally fitting. The enhanced mood set has an extra granularity that sync supervisors will love. Need a track for a scene that’s both dramatic and dreamy? Simply filter using those tags.
Beautifully crafted, velour-suited pastiche ‘Leave The Door Open’ was one of our favourite records of 2021. It’s unquestionably Romantic and Relaxed in mood. And when we seek more detail on Bruno Mars and Anderson .Paak’s lush performance, the enhanced mood tags deliver. The tags Soulful, Seductive and Heartfelt seem to be exactly what the duo were aiming for.
Curators can also filter catalogues using the same tags to quickly build playlists in the same sonic ballpark, both within a specific genre, or across all genres.
What do enhanced moods mean for catalogue navigation?
The new enhanced mood classifier gives users more routes into music catalogues. By providing more granular musical mood descriptors, video creators, TikTokers, sync supervisors, catalogue managers, and curators can find the tracks they want even faster.
The AI also has the potential to help unearth lesser-known tracks that fit the bill, no matter the use case.
For artists, the beauty of having more tags is that there’s even more chance of being heard by industry decision-makers with the power to propel musical careers.