There are different ways to search for a song stuck in your head. The most common way is to throw the lyrics at Google, and it will bring you the most relevant song titles and the artist names.
But this method is only possible if you remember partial lyrics of the song. However, what happens if you know the flow (e.g., da daaaa da da daaaa na naa naa ooohh yeah) but can’t remember any lyrics? Yes, it’s possible to search for a song without knowing the lyrics and even the artist’s names. Google search now lets you find a song by humming.
Just hum and Google will help you discover the song. This method doesn’t need you to know the name of a song or partial lyrics… a 10-15 seconds hum, whistle, or melody is all you need. So how can you hum to search for a song on Google? This guide will help you with that.
How to find a song by humming
Step 1. On your smartphone, open the Google Search app or tap the Google Search widget and hit the mic icon.
Step 2. Say “what’s this song?” or click the “Search a song” button at the bottom
Step 3. Start humming the tune stuck in your head for 10-15 seconds.
Google will do the heavy-lifting by identifying potential song matches. The options will be arranged in order of their matching percentages. Songs with a high percentage match will be shown first.
You can view accompanying song information like lyrics, music videos and even listen to the song on your favorite music streaming app.
Google Assistant simplifies this process. Just say the wake keyword “Hey Google” followed by “what’s this song?” then start humming.
Google’s hum to search for song feature is available on both Android and iOS. On Android you can search for songs in more than 20 languages and in iOS, only English is available with more to come soon.
Behind the scenes: How Google Search music discovery works
Google uses a machine learning algorithm to identify songs. In its simplest form, the algorithm knows every song by a unique fingerprint. Thus every song has its own unique identity. When you hum, sing or whistle, the algorithm looks for a combination of fingerprints that seem to match the search.
The algorithm does this by first transforming the hum into a number-based sequence that represent the song’s melody. This profile is built by factoring in other tiny details as well like accompanying instruments, your voice’s timbre and tone to come up with a fingerprint. The fingerprint is then run against thousands of existing song sequences looking for potential matches.
Google has tuned its machine learning algorithm to identify songs based on different sources including humans’ singing, whistling or humming, and studio recordings.