Shazam: predicting future hits

On: September 8, 2014
Print Friendly, PDF & Email
About Yvette Ducaneaux


Have you ever listened to a song and almost died trying to find out the name of the song? Then Shazam is going to be your best friend. Shazam is a music recognition app that was launched in 2002. Twelve years later it ‘recognizes’ over 25 million unique tracks. An astonishing amount. How does Shazam actually recognizes music? While searching for this answer on the website of Shazam, I came across a page called ‘Shazam Charts’. (see picture below) Apparently, Shazam has become more than a music recognition app. It now also provides us with lists of music (top 100/ Future hits) by keeping track of how often a sound is Shazamed. What is to say about Shazam and predicting future hits?


Shazam future hits


Shazaming a sound, how does it work?

When clicking on the icon, Shazam starts listening to the sound received by your phone. It then compares the sound to the tracks stored in a massive database which hold over 25 million tracks. Shazam does not store the entire track as this would cost too much space (in terms of memory), instead, Shazam stores tracks as so called ‘digital fingerprints’.

“Shazam does not save nor send the audio it hears; only digital fingerprint summaries of the audio are sent to Shazam’s servers to identify media content in Shazam’s databases. The original audio cannot be reconstructed from Shazam audio fingerprints. The auto feature for Shazam for Mac, as with the Shazam mobile app, creates a digital fingerprint of the audio it hears every several seconds and matches it against Shazam’s global database of recorded music, television shows, and live events, and select TV and movie advertisements.” Shazam Support

Of course Shazam won’t tell us exactly how it works and what algorithm they use for Shazaming and recognizing sound, but if you’re curious how Shazam creates these digital audio fingerprints and why this is so accurate, you can read more about it here.


Testing Shazam: how accurate is it?

While testing Shazam I found that the service was pretty precise and could recognize sounds that only few people knew. When Shazaming a sound on Youtube which (at that time) had only 218 views, Shazam still recognized the track and returned the right track name. 

Not only Shazam decides whether a sound makes it into the database, any artist or person who has the legal rights for that sound can opt for Shazaming (digitally fingerprinting) their sound and adding it to the database: you just have to ask. How do I ask Shazam to include my sound into their database?


Shazam and data: predicting future hits.

Now let us take a look at Shazam’s ‘future hits’ charts. How can Shazam predict future hits?

The user sends a request to Shazam for finding the name of the track that is being listened to. Shazam collects this information: the request of the user and the response.  (Which will probably include the name of the track/time/date/location and maybe more?) A hit(song) can be explained as ‘song that has become very popular’. In that sense, Shazam can predict what songs are about to get very popular by counting how often a request is made for a particular song. The song that is Shazamed the most will most likely become a hit. Shazam already made country specific lists of future hits. People will only Shazam a sound when they like what they hear (or at least be interested in some way).

Over 100 million people use Shazam every month to recognize sound. (Shazam news) This massive amount of request leads to massive amounts of data that in the academic field of Mediastudies we tend to call: Big Data. An exact definition of Big Data is impossible to give in one sentence, but the first part of the term can be easily explained and connected to Shazam: having 100 million users and 25 million request a month is definitely ‘Big’. For more information about ‘Big Data’ you can click here. For now it is important to know that combining data (such as a request and the location the request was made) can lead to new information (counting the request per sound) which can lead to new (or hidden) insights and knowledge (like predicting future hits per country).

Shazam seems to be the perfect example for using data in a great way; users not only create data for Shazam, but they get something in return: a look into the future. Unfortunately, Shazam hasn’t include ‘the Netherlands’ yet in the lists of countries for future hits, so for now we can only guess…

Comments are closed.