‘Hum to Search’ by Google

You know that Christmas song that goes like “Ba da daaar da, dar da, da da”, but you can’t quite remember the song or lyrics? Well, Google’s new Hum to Search feature can help with just that! At Search On, Google announced how Hum to Search can help you figure it out – with no lyrics, title, artist name or pitch perfect required.


As of now, you can hum, whistle or sing a melody into your Google Assistant, App or Search Widget for them to help solve. An easy way to explain how it all works, is that a song’s melody is somewhat like a fingerprint. They are all unique in their own kind of way. So basically, Google have created machine learning Hum to Search model that can match your hum, whistle or singing to that said fingerprint.


These new models are trained to identify songs based on a variety of sources – humans singing, whistling, humming and studio recordings. The algorithm that accompanies this swanky new tool also takes away all the other details, like accompanying instruments and the voice’s timbre and tone. Cool, right?

Google then compares the sequences you’ve just sampled to thousands of songs from around the world, to identify potential matches in real time.


The team here at Super Digital tried the Hum to Search feature and although a little off-key, it worked! This new tool may not be the most pressing issue of 2020, but it’s definitely a technological advancement in the world of AI.

Especially during the festive season, we all love a good ol’ sing song, are we right? So next time you can’t remember the name of the catchy song you just heard on the radio, just start humming. You’ll have your answer just like that!

Watch the full promo video here to learn more.