300 TB of Spotify to build a better Shazam: the plan challenging the original app

300 TB of Spotify to build a better Shazam: the plan challenging the original app

The idea sounds like an engineer with too much free time and a huge hard drive going off the rails, but it has its logic: download all 300 terabytes of Spotify's catalog – some 86 million songs – to build a music recognition system that outperforms today's Shazam. The project, making the rounds in technical circles, proposes using Spotify's metadata and YouTube audio to generate fingerprints (hashes) that identify songs with greater accuracy.

Why Shazam fails and how to improve it

Shazam, the go-to music recognition app, has blind spots. With little-known songs or noisy snippets, it sometimes returns wrong results, leading users to complain on YouTube that the app redirects them to strange tracks. The proposed alternative involves building a hash database from Spotify's most popular songs – those with over 95% popularity – supplemented with popular YouTube tracks not available on the streaming platform. The math is simple: 300 TB / 86 million songs = about 3.4 MB per file in mp3 or AAC format. It's not FLAC, but it's enough to generate hashes.

The business behind the idea

While some see the project as an eccentricity with no market, the entrepreneurial approach is clear: a more reliable Shazam that can also identify songs the original app doesn't recognize. The key is not to store the 300 TB of audio, but only the hashes – far lighter – and prioritize the most searched-for tracks. The main stumbling block, however, is not technical but legal: downloading Spotify's metadata database and YouTube audio may abusa terms of service and copyright. The debate over the legality of using OnlyFans content as an example makes it clear that, in Spain, the severity of the offense varies depending on the origin of the pirated material.

A colossal project built on sand

With 300 TB of data and 86 million songs, storage and bandwidth are only the first obstacle. The quality of audio extracted from YouTube – generally compressed – and the need to keep the database updated with weekly releases make viability even harder. But the idea persists, as a determination to prove that a better Shazam can be built with open resources and a bit of technical ingenuity.

---

The project remains in the speculation phase, with no sign that anyone has started downloading the 300 TB. But the discussion has left an uncomfortable question on the table for Shazam: why does such a popular app keep failing with songs outside the top 40?