Identify (audio recognition)

Recognise a track from audio. Send the raw audio file and you get back the best-matching catalog tracks, ranked, each with a confidence score. Costs 10 + ceil(MB) credits. A request that would cost more credits than you have left this month is refused with a 429 before any matching runs. Paid plans only.

Requires a paid plan: the free tier returns 403 for this endpoint.

How much audio to send, cost & retries

• Send the whole track (or a generous section), not a short clip. The part that matches our catalog reference can sit anywhere (often the main hook a few minutes in, not the intro), so a snippet from the wrong section misses the match even though the track is in the catalog. ~10 s is the minimum; more audio is more reliable.

• You're billed on the bytes you upload (10 + ceil(MB)), so a full ~4-min track is ~15–20 credits. On quality: 320 kbps gives the best result, 128 kbps is the best value, and lossless is no better than 320 while costing more. To shrink the upload, downsample to mono 16 kHz (the measured safe floor, do not go lower), accepting that downsampling trades some accuracy for the saving. Trimming to a few seconds is cheaper but risks a miss.

• On a miss we automatically scan deeper windows of your audio and vote across them before returning no match, so there is nothing to retry by hand.

Response

• matched (boolean) and results - the matching catalog tracks, best match first. Each result is slim: id, title, artists, and confidence (a 0–1 score; higher means a more certain match).

• Need ISRC, genre, release info, or cross-platform links? Look the track up by id with GET /v1/tracks/:id (or /v1/tracks/links). Identify keeps its response small on purpose.

• credits_charged - credits billed for this call.