Google launches hum to search feature

Users can also whistle or sing a melody to Google to find the song name. The tool is accessible on mobile devices via the latest version of the Google app or the Google Search widget.

The company says it has built machine learning models that can match a hum, whistle or melody to a song’s unique fingerprint. Only a few thousand songs can match after just a few notes. The tool will display the most likely options based on the tune, and users don’t need perfect pitch to use the feature.

“When you hum a melody into Search, our machine learning models transform the audio into a number-based sequence representing the song’s melody,” a Google blog post reads. “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.”

To use the service, tap the mic icon and say ‘what's this song?’ or click the ‘Search a song’ button, then start humming for 10 to 15 seconds. On Google Assistant, say ‘Hey Google, what’s this song?’ and then hum the tune. The tool is available in English on iOS, and in more than 20 languages on Android. The company says it hopes to expand the service in more languages in the future.

Before the tech giant launched the hum-to-search service, users needed to ask Google to identify the song based on a machine recording. In 2017, Google introduced its Now Playing feature, which uses neural networks to bring low-power music recognition to its mobile devices. In the following year, it expanded the feature to the Google app and SoundSearch.

Meanwhile, Spotify recently launched a new tool that lets users search for a song by its lyrics.