Morning Coffee: The Great Shazam Mystery


Good morning! I hope the world is being good to you and you’re being good to it.

Grab your coffee or beverage of choice and share a thought with me!

At some point this year, I recognized I need to be more curious and that I’d seemingly lost my natural wonder to discover how things work. I blame reaching a certain number of years being in an adult world where things just need to get done, so you become focused on outcomes rather than the recipe. Kids seem to have that natural inclination to incessantly ask “why” about everything despite not always being able to understand the answer. Some adults have that too, and that curiosity, regardless of age, helps build world knowledge that pays dividends throughout life. The desire to understand why , at least when it does not pertain to something happening to me, is one of my fading qualities.

Fading, however, is not dying, at least not in this case. There is still time and opportunity to reinvigorate my natural curiosity and hopefully add another wrinkle or two to the ol’ gray matter. I don’t know if this is the start, but my crosshairs in this endeavor landed on everyone’s favorite song-identifying app: Shazam.

I saw a promo for Beat Shazam on TV, and it got me thinking how terrible I am at remembering lyrics or naming songs, which gave me a momentary appreciation for the Shazam widget on my phone. It’s not something I use often, but it saves me the endless futility of screaming to the heavens, “WHAT IS THAT SONG?!” I am always amazed that Shazam works almost perfectly, even when it needs to cut through the din of a busy environment. How does that little miracle app do it?

Turns out, Shazam works via a fair bit of auditory and algorithmic complexity. Thanks to the internet, I found this article by Jovan Jovanovic who thoroughly broke down the process and then reconstructed it into three “simple” components for the random curious folk like me. I’ll summarize:

Songs samples, submitted by the user, are broken down into fingerprints, which are then matched against other fingerprints of a library of known songs.

Even though the song sample can come from a different piece of the song, and that piece of song is not consistent or inherently known, the fingerprint’s frequency contains enough data to match the relative timing of a sample to similar samples in known songs. (I’ll be honest, I’m heavily paraphrasing this part since it feels like the most complex piece.)

Shazam can use the fingerprint to compare against the fingerprints of known songs in its database, looking for similar frequency (fingerprint) occurrence and timing to home in on artist and song.

There is plenty more happening behind the scenes, but this simplistic breakdown is how I’m storing the information so my eyes don’t glaze over and my brain doesn’t start emitting smoke. Apologies, Mr. Jovanovic for not doing your article the service it deserves.

Anyway, this is probably where the “The More You Know” star and rainbow fly across the screen and we get back to our regularly-scheduled program. If you didn’t know, now you know.

Have a good day, and stay curious.


Leave a comment