Google just launched its “hum to search” feature, which allows users to hum, whistle, or sing for 10-15 seconds in order to identify a song.
The feature currently works in 20 languages, and Google hopes to add more.
Google has finally launched the perfect feature for when a song is stuck in your head but you don’t know any of the words.
“Hum to search” launched today on both the Google app for iOS and Android, according to Google’s blog, The Keyword. Users can also whistle or sing directly into the mic to identify a song.
The technology works like this: the user can hum (whistle, or sing) for 10-15 seconds, and then Google’s technology takes the song’s melody and turns it into a numbers-based sequence.
From there, the sequence can be used to “identify songs based on a variety of sources, including humans singing, whistling or humming, as well as studio recordings,” according to Google’s announcement. The sequence also strips away any other outside noise, like accompanying instruments.
Google launched the music-identifying AI in 2017 with its Pixel 2, which was a small-scale attempt by the company to allow phones to be able to identify music, even without an internet connection. The following year, the company expanded the AI to the app, where users could ask the app what song was playing.
The technology doesn’t expect the user to be a perfect singer — the results are based on the tune. Once the results pop up, the user can pick the song they think best matches their submission.
Also included in the results are accompanying music videos, song analyses, and even any available covers for the song. There is also a link to listen to the song on major streaming services.
The technology also works in the Google Assistant app. The feature is currently available in 20 languages and Google hopes to add languages in the future.