FreqBlog vs MeloData

If you are replacing Spotify's discontinued audio-features endpoint, the honest question is not "which one is real" — it is which service computes the numbers itself, and which one fits the way your data is keyed and the tools you actually need to build. MeloData and FreqBlog both analyse real audio with the open-source Essentia engine rather than re-serving Spotify's retired values, so the differences that matter are inputs, notation, tooling, and who stands behind the service. Every MeloData claim below is taken from MeloData's own homepage and API docs, checked on 14 July 2026.

At a glance

Both services back-fill on a miss: FreqBlog analyses uncovered tracks on demand, and MeloData returns a 202 for a never-seen ISRC, then analyses and caches it. Both also offer a free tier of 1,000 lookups per month with no credit card required.

When MeloData is the better fit

MeloData is a genuine Essentia-based analyser, and for some teams it is the cleaner choice:

• Your pipeline is already keyed entirely by ISRC and you want a minimal ISRC-in, features-out contract with nothing else to map.
• You specifically need valence, acousticness or instrumentalness and your catalogue is well covered by the AcousticBrainz archive rows, where MeloData exposes those values.
• You would rather drop a copy-paste agent-skill file into Cursor or Claude Code than point a client at a hosted MCP endpoint.
• You value MeloData's granular, per-field accuracy notes (which fields are computed over the 30-second preview window) for your own calibration.

Why teams pick FreqBlog

• Look up by name, not just ISRC. Pass "Blue Monday" + New Order and get features back in one call — no Spotify ID, ISRC or MBID required (though all are accepted).
• Built for harmonic mixing. Every key ships as name, Camelot and Open Key, and the set-builder (/transition, /next-track, /setlist) orders a crate into a beat-matched, key-compatible arc.
• Agent-native. A hosted MCP server at mcp.freqblog.com/mcp lets LLMs and agents call the catalogue directly, alongside /charts and /similar discovery endpoints.
• Deeper response. 42 fields per track — BPM, key, energy, mood, genre, ISRC, MBID and more — computed with Essentia on real audio.
• Bigger warm cache. ~576,000 tracks already analysed means fewer on-demand waits, with on-miss backfill for anything not yet covered.
• A named, accountable operator. Run by Steven Birring in Nuneaton, UK — an ICO-registered sole trader trading as Stackbase/FreqBlog, not an anonymous brand.
• Transparent, low pricing. A free tier of 1,000 requests per month with no card, and paid usage from roughly £0.17 per 1,000 requests.

FAQ

Do I need a Spotify ID or an ISRC to use FreqBlog?

No. FreqBlog resolves a track by name + artist directly, and also accepts ISRC, MusicBrainz ID or Spotify ID if you already have one. MeloData's feature endpoint is keyed by ISRC, with title/artist available through a separate search call.

Are FreqBlog's numbers just re-served Spotify values?

No — and neither are MeloData's. Both compute their features with the open-source Essentia engine from audio rather than re-serving Spotify's discontinued values. The practical differences are in inputs, harmonic notation (Camelot/Open Key), set-building tools and the number of fields returned.

Can I use it with an LLM or coding agent?

Yes. FreqBlog runs a hosted MCP server at mcp.freqblog.com/mcp, so agents can query the catalogue without any local wiring. MeloData instead publishes a copy-paste "agent skill" file for Claude Code and Cursor.

What happens when a track isn't in the catalogue yet?

Both back-fill on a miss. FreqBlog analyses the track on demand and serves it once ready; MeloData returns a 202, analyses the ISRC, then caches the result. FreqBlog starts from a larger warm cache of ~576,000 pre-analysed tracks.

Compare with others

• FreqBlog vs Musicae
• FreqBlog vs ReccoBeats
• FreqBlog vs AudD
• Full music-API comparison — every provider