A song identified from its sound alone, 1999.

Shazam began as a question asked in a classroom. A post on the Berkeley MBA blog of the Haas School of Business, written by Rahul Sampat and published on 30 July 2019, says Chris Barton conceived the idea in 1999 while he was an MBA student at UC Berkeley. He arrived planning radio-monitoring software and changed direction after a Strategic Innovation course, asking what would happen if someone could identify a song using the actual sound of the music captured over a mobile phone. English-language Wikipedia dates the founding of Shazam Entertainment Limited to 1999 and places the company in London.

The founding team came from both sides of the Atlantic. Wikipedia names Barton and Philip Inghelbrecht, both Berkeley MBA students, and Dhiraj Mukherjee, who worked at the London internet consulting firm Viant. The Haas post describes Mukherjee as a long-time friend of Barton’s and a Stanford MBA student, and quotes Inghelbrecht saying that Mukherjee brought product and operational expertise that he himself knew very little about. The team then recruited Avery Wang, who had earned a PhD at Stanford in auditory scene analysis, through its network of professors at Stanford and Berkeley; the Haas post counts him among the co-founders, and Wikipedia describes him as the digital signal processing specialist the company hired.

Wang set out the problem himself. In a paper titled “An Industrial-Strength Audio Search Algorithm”, presented at the ISMIR music information retrieval conference in 2003 under the Shazam Entertainment name, he wrote that the company had been started in 2000 to connect people to music by recognising it in their surroundings through their mobile phones. The sample would be short, mixed with heavy background noise and reverberation, captured by a small phone microphone, squeezed by voice compression and interrupted by network dropouts before it reached the company’s servers. Wang wrote that no algorithm known to the team met all of those constraints, so it developed its own.

Four digits, thirty seconds and a text message.

The first product was a phone number. The Haas post says Shazam launched in the UK only in 2002 under the name 2580, the number users dialled; after a 30-second call they received a text message with the song title and artist. Wikipedia adds that the call ended automatically after 30 seconds and that later messages carried hyperlinks for downloading the song. Wang’s 2003 paper described the same experience from the inside: a listener calls the service, samples up to fifteen seconds of audio, the identification runs on the company’s servers and the title and artist come back by SMS, with the tagged tracks also listed on a website. He wrote that the service was then live in Germany, Finland and the UK, and Wikipedia records a US launch on AT&T Wireless in 2004.

The service moved onto handsets as phones changed. Wikipedia lists a first smartphone app in 2006 on the Amp’d Mobile service, an iPhone app that appeared with the launch of the App Store on 10 July 2008, an Android version on 30 October 2008 and a paid edition, Shazam Encore, for iPhone in November 2009. It also records an expansion into tagging television for US users in 2012 and integration with Apple’s Siri when iOS 8 launched in September 2014. Apple’s own announcement in 2018 called Shazam one of the first apps on the App Store.

Ownership changed at the end of 2017. Wikipedia says Apple announced the acquisition in December 2017, that the European Commission said on 23 April 2018 that it would review the deal and that it approved it on 6 September 2018. On 24 September 2018 Apple’s Newsroom said it had completed the acquisition and that Shazam would soon be ad-free for all users. Oliver Schusser, then Apple’s vice president of Apple Music, said in the release that Apple and Shazam had a long history together. Wikipedia records that Apple announced ShazamKit, a framework bringing Shazam’s recognition to other developers’ apps, in June 2021, and Apple’s design guidance for ShazamKit refers to a Music Recognition control alongside the Shazam app itself.

From a dial-in number to an app inside Apple.

For a music streaming or discovery service, the app icon does most of the work a logo does elsewhere. It sits on a phone’s home screen among dozens of others, is seen at the size of a fingertip and has to be recognised before it is read, so these marks tend to reduce to a single symbol, a simple shape or a letter on a flat field. Colour carries even more of the load. A single strong hue can identify a service on a lock screen, in a share card, on a web player and in a corner of a festival poster, and many services treat that colour as their main asset, with typography and illustration arranged to support it rather than compete with it. Because the catalogue itself belongs to artists and labels, the brand usually shows up in the frame around the music: the player, the buttons, the empty states and the motion between screens.

Playlist covers are where a streaming identity is tested at scale. Editorial teams publish hundreds or thousands of playlists, so covers are built as templates, with a fixed type treatment, a colour rule and a slot for a photograph or pattern that can change weekly without breaking the family. Year-end recap campaigns stretch the same system into a shareable format, turning a listener’s own data into cards designed to be posted, and short sonic logos do a similar job for sound in adverts and app openings. Services that court musicians also need an artist-facing layer, with dashboards, promotional assets and upload tools that borrow the consumer palette but speak in a plainer, more practical voice.

Constellation maps, a star field and a strip of plastic.

The most vivid picture of Shazam comes from its engineering. In the 2003 paper Wang explained that the system reduces a recording to a spectrogram, a map of frequency against time, and keeps only its strongest peaks. The team called the resulting sparse lists of coordinates “constellation maps”, he wrote, because the scatter plots often resemble a star field. To explain matching, he asked readers to imagine a song’s constellation map on a strip chart and a few seconds of a sample drawn on a transparent piece of plastic, slid along until the dots line up.

The paper also describes how the dots become a fingerprint. Points are chosen as anchors, each with a target zone, and every anchor is paired with the points inside its zone; each pair yields two frequencies and the time between them, which become a searchable hash. Wikipedia describes the app recording about ten seconds of audio, creating a fingerprint from it and matching it against a database of millions of songs to return the artist, title and album. Wang wrote that the method could pick out music behind a loud voice-over, and that several tracks mixed together could each be identified.

The name has company in the wider world.

For a design guide, Shazam shows how a brand can be built on a single gesture, from a four-digit number dialled near a speaker to a button inside an Apple control, while its technical story offers a ready-made visual language of peaks, dots and star fields. Compare Apple Music, the streaming service of Shazam’s owner, and Spotify, the Stockholm streaming service. For this guide we use tag blue, star field and peak white with a sans, an unofficial interpretation; no product or logo is reproduced.

Keep Shazam’s identity on hand — save it to a board in the Made Good inspiration library.