Detect BPM, key and scale from any audio file. Moderate text with a tiered cascade that can tell a quoted lyric from a threat. One HTTP call each, built for products that ship on a schedule.
Send a file or a URL, get back tempo and tonality with confidence scores and a versioned engine string, so you always know which answer came from which pipeline. Async, with webhooks.
A cascade of local checks and language models returns per-category scores and a one-line reason a human can verify. Synchronous — fast enough to sit in a comment or chat path.
Idempotency keys on every write, request IDs on every response, stable error codes, signed webhooks with retries, per-key rate limits. 202 where it should be, 200 where it can be.
Bearer key in, JSON out. Analysis returns a job you can poll or receive by webhook; moderation answers inline.
Early access is open to a small number of teams while the API is being finished. If that's you, get in touch.
# audio → async curl -X POST https://api.irate.tech/v1/audio/analyze \ -H "Authorization: Bearer vk_live_…" \ -F file=@track.mp3 { "job_id": "01J…", "status": "queued" } # moderation → sync curl -X POST https://api.irate.tech/v1/moderation/check \ -H "Authorization: Bearer vk_live_…" \ -d '{"text":"…"}' { "flagged": true, "categories": ["harassment"], "scores": { "harassment": 0.92 }, "tier": "llm" }
Vibed is a social network for artists — live rooms, unsigned records, and no algorithm deciding who gets heard. It runs on this API in production, over the same public endpoints any customer gets. No private path, no special treatment: if something is awkward for Vibed, the contract is what gets fixed.