One Song, One Record: How Beatra's ISRC-Verified Music Catalogue Works

Music bots usually store whatever the source hands them. Play the same song from YouTube and from Spotify and you get two records, two play counts and two different names. Beatra stores it once.

This system replaced the earlier "MasterTrack" database. The idea survived, the implementation did not: everything below describes what runs today.

The problem with raw platform data

One recording can appear on the internet under half a dozen names:

• "Tarkan - Kuzu Kuzu (Official Music Video)"
• "Tarkan · Kuzu Kuzu"
• "TARKAN - Kuzu Kuzu [Lyrics Video]"

Counted separately, those split your play counts, pollute search, and make recommendations worse. Cleaning song titles with string tricks does not fix it either — it just produces a different set of wrong answers.

ISRC as the source of truth

An ISRC is the industry identifier for a specific recording, assigned by the rights holder. Beatra treats it as the gate into the catalogue: a track that has been played is verified against public metadata, and only a verified recording becomes a catalogued song with a real title, a real artist and real artwork.

That gate is why the catalogue has no near-duplicate junk. If a recording cannot be verified — an unofficial edit, a rare upload, a mashup — it still plays perfectly, it just does not become a catalogue entry.

What one record holds

Each song is stored once, with links to every platform it has been seen on:

• YouTube and YouTube Music video ids
• Spotify, Deezer, Apple Music, Tidal and SoundCloud links, when found
• The canonical title and artist, artwork and duration
• Play counts, listeners and skip behaviour, aggregated across every source

Artists are canonical too, so "ILLIT", "HYBE LABELS" uploads and featured credits all resolve to one artist page rather than three.

What it changes for you

Accurate stats. Your play counts, listening time and top tracks are counted once per recording, no matter which platform anyone played it from.

Faster playback. A song already in the catalogue resolves from Beatra's own store instead of a fresh external search, so it starts quicker — and it starts as the right recording.

Better search. A search checks the catalogue first. Beatra also learns from what actually gets played, so a query that used to return the wrong upload gets corrected over time.

Song and artist pages. Every catalogued song has a page at beatra.app/songs with its stats, artwork and platform links; every artist has one at beatra.app/artists with their catalogue, similar artists and top listeners.

Real recommendations. Autoplay and the recommendation engine work on the catalogue's own similarity graph — built from what people actually play together — rather than a third-party API.

Cleaner playlists. A playlist track that matches a catalogue entry links straight to its song page, so you see the clean title and artist rather than an upload's marketing headline.

It grows on its own

Every track played through Beatra — in Discord or in the web player — is queued for verification in the background. Nothing is asked of you, and playback is never delayed by it. Recordings that fail verification once are retried later, so the archive keeps filling in rather than dead-ending on one unlucky check.

Browse what has been catalogued so far at beatra.app/songs.