Have you ever had a melody stuck in your head, but couldn’t quite place the song? Frustrating, right? Well, thanks to advancements in technology, those days of musical mystery are over. Now, with a simple “Hey Google, what song is this?” you can hum, whistle, or sing a tune, and let Google work its magic to identify it. But how exactly does this impressive feature work? It’s all thanks to the power of machine learning, turning melodies into digital fingerprints.
Imagine a song’s melody as its unique fingerprint. Just like each person has a distinct fingerprint, every song possesses a unique melodic identity. Google’s engineers have developed sophisticated machine learning models that are trained to recognize these musical fingerprints. When you use the “Hey Google, what song is this?” feature and hum a few bars, your audio is transformed by these models into a number-based sequence. This sequence acts as the digital fingerprint of the melody you’re providing.
These machine learning models are incredibly versatile. They are trained on a vast dataset of musical information, learning to identify songs not just from studio recordings, but also from the nuances of human singing, whistling, and humming. The algorithms are designed to filter out extraneous details like accompanying instruments, vocal timbre, and tone. By stripping away these layers, the models focus purely on the core melody, converting it into that crucial number-based sequence, the melodic fingerprint.
This melodic fingerprint is then compared in real-time against a massive database containing the fingerprints of thousands of songs from across the globe. The system rapidly searches for potential matches. Think about the popular song “Dance Monkey” by Tones and I. You can instantly recognize it whether you hear the original studio version, a live vocal performance, or even someone whistling the tune. Similarly, Google’s machine learning models can recognize the underlying melody of “Dance Monkey” from its studio recording and match it to the number sequence derived from your hummed audio.
This innovative feature is built upon Google Research’s groundbreaking work in music recognition technology. This journey began with “Now Playing” on Pixel 2 in 2017, which brought low-power, on-device music recognition to mobile. In 2018, this technology was further refined and integrated into the SoundSearch feature within the Google app, expanding its reach to a catalog of millions of songs. “Hey Google, what song is this?” represents the next evolution, allowing you to identify songs even without lyrics or a perfect rendition of the original – all you need is to hum a tune. So next time a melody pops into your head, just ask “Hey Google, what song is this?” and let the musical magic unfold!
