Google held its surprise 'Search On' event today and detailed the latest improvements to its Search app and more. It announced that starting today, users can "Hum to Search" using the app to find the song that has been stuck in their head.
To get the feature working, users need to either tap the microphone and say "what's this song?" or hit the "Search a song" button. They need to hum for 10-15 seconds, and then results will be provided from Google Search. Irrespective of users' pitch, machine learning algorithms will determine the most likely matches and provide users with the results, who can then choose the best match. The firm detailed how the feature works:
When you hum a melody into Search, our machine learning models transform the audio into a number-based sequence representing the song’s melody. Our models are trained to identify songs based on a variety of sources, including humans singing, whistling or humming, as well as studio recordings. The algorithms also take away all the other details, like accompanying instruments and the voice's timbre and tone. What we’re left with is the song’s number-based sequence, or the fingerprint.
We compare these sequences to thousands of songs from around the world and identify potential matches in real time. For example, if you listen to Tones and I’s “Dance Monkey,” you’ll recognize the song whether it was sung, whistled, or hummed. Similarly, our machine learning models recognize the melody of the studio-recorded version of the song, which we can use to match it with a person’s hummed audio.
The feature is presently available to both Android and iOS users. While iOS users can only use the service in English, the Android app currently supports more than 20 languages, with support for more languages coming in the future.
Google introduced a similar Now Playing feature with the Pixel 2 in 2017 that identified songs playing around users even when offline. Now, the firm does not need to listen to the original track or even lyrics to recognize them.
16 Comments - Add comment