New Google Search tool identifies songs based on your humming

A new Google tool identifies songs based on humming.

There’s nothing worse than getting a tune stuck in your head that you can’t remember the words to. Without being able to look it up, there’s a chance you’ll be humming it for the rest of the day. On Thursday, Google introduced several new Search features. One of them should help those who have a pesky song stuck in their head.

It allows users to hum the tune and the Google Assistant then tries to figure out what song it is from. The song-finding tool should help users figure out what song they’re humming and learn more info about it.

Impressive Feat

A tool that can identify a song based on 10 seconds of tone-deaf humming almost sounds too good to be true. I tried Google’s new song-finding tool for myself to see how it works. Impressively, it managed to correctly guess Beartooth’s “In Between,” Fetty Wap’s “Trap Queen,” and the iconic “Imperial March” on the first try. Yes, I wanted to give it some variety.

Manage your supply chain from home with Sourcengine

To be clear, my humming is nowhere close to being considered a musical ability. Yet, the song-finding tool was able to analyze the sound and guess the correct song in a matter of seconds.

To try it firsthand, users can simply open up the Google app or summon their Google Assistant. Then, ask Google “what’s this song?”

The screen will show a blank set of audio bars that serve as a cue to start humming, whistling, or singing. Users can do so for 10 to 15 seconds before the Google Assistant starts processing the audio. From there, the machine learning algorithms take over.

In just a few seconds, Google displays a list of songs that are potential matches as well as the percent probability that they are correct. Users can then tap on each song to listen to it, view the lyrics, or learn more to figure out if it is the one that’s been stuck in their head.

Musical Fingerprints

The song-finding tool itself is neat. However, the technology behind it is even cooler. In a company blog post, Google says, “A song’s melody is like its fingerprint. They [songs] each have their own unique identity. We’ve built machine learning models that can match your hum, whistle, or singing to the right ‘fingerprint.’”

As users hum a song, the Google algorithms translate the audio into a number-based sequence representative of the melody. Since the program strips away things like instruments and background sounds, it is left with a clean fingerprint of the song. That makes it possible to determine matches and make accurate suggestions in real-time.

Currently, the new feature is available on both iOS and Android devices. On the latter, it is accessible in more than 20 languages. However, iOS users are limited to English at this time. More languages will be added in the near future.

Google’s new song-finding tool probably isn’t something that users will need every day. That being said, it will certainly be nice to have when you can’t get a song out of your head.


Please enter your comment!
Please enter your name here