You hear a song playing in a shop, tap a button on your phone, and seconds later you know the title and artist. But how can your phone recognise a song from just a few seconds of sound?
Apps such as Shazam make this look almost like magic. Your phone listens to a short piece of music, analyses it and compares it with a huge collection of songs.
The secret is not that your phone remembers every song. It creates something called an audio fingerprint.
Your Phone Listens to the Sound
When you ask your phone to identify a song, the microphone captures a short sample of the sound around you.
The recording does not need to contain the entire song. Even a short section can provide enough information for the system to find a match.
Shazam converts the captured sound into a digital representation called an acoustic signature. Apple describes this signature as a compact representation of the sound’s time and frequency characteristics.
Think of it like a musical fingerprint. Just as a fingerprint can help identify a person, a distinctive pattern in a piece of audio can help identify a song.
It Does Not Compare the Entire Song
This is where the process becomes clever.
Instead of sending the entire recording to a system and asking it to compare every second of audio, the technology works with a much smaller signature.
Shazam’s technology creates reference signatures for songs in its catalog. When you try to identify a song, the signature created from your recording is compared with those reference signatures.
If enough of the patterns match, the system can identify the song.
This is one reason the result can appear so quickly.
What If There Is a Lot of Noise?
You might think this would only work in a quiet room.
But real life is much messier. You may be standing in a restaurant, shopping centre or crowded street while trying to identify a song.
Audio fingerprinting is designed to find useful patterns even when the captured sound is not perfect. Apple says Shazam’s matching can work with noisy audio and partial recordings, such as music playing in the background of a restaurant.
The system is looking for characteristics of the music that remain useful even when other sounds are mixed in.
What Happens After the Match?
Once the system finds a strong match, it can return information connected to that recording.
That can include the song title, artist and other details associated with the track in the music catalog.
The app can then show the result on your screen and provide options to listen to the song through supported music services.
The entire process can happen within seconds.
Your Phone Is Not Storing the Song
An interesting part of the technology is that an audio fingerprint is not the same thing as the original recording.
Apple describes the signature as much smaller than the original audio and says it is a one way representation, meaning the original recording cannot be reconstructed from the signature.
Apple also states that Shazam does not record or store the audio it detects.
So the technology is not simply listening to a song and searching the internet for the same recording. It is turning the sound into a compact pattern that can be matched against known music.
The Simple Answer
Your phone can recognise a song so quickly because it does not need to understand the music the way a human does.
It captures a short sample, turns the sound into an audio fingerprint, and compares that fingerprint with reference information in a music catalog.
When the patterns match closely enough, the mystery is solved.
That song playing in the background may sound like just a few seconds of music to you. To your phone, it can be a unique digital signature waiting to be recognised.
