Create a transcript

Transcribes an audio or video file, from a public URL or from a file sent with POST /v1/transcripts/upload.

The call waits up to about 20 seconds: a short file comes back 200 with status: "done" and the text. Anything longer answers 202 with status: "queued" or "processing"; read the result with GET /v1/transcripts/{id} or receive it on webhookUrl. Send ?wait=false to always answer at once.

Pricing: US$ 0.02 per hour of audio, charged per started minute of the audio actually transcribed, only when the transcript is done. Failed transcripts are not charged. Requires paid balance or auto-recharge; trial credit does not apply. The price of each transcript is held from the balance once its length is known, before it is processed: a transcript the balance cannot cover ends in error with insufficient_balance: ... and costs nothing.

Limits: files up to 2 GB and 6 hours of audio. Results are kept for 7 days.

Recent Requests
Log in to see full request history
TimeStatusUser Agent
Retrieving recent requests…
LoadingLoading…
Query Params
string
enum

false answers at once with the transcript queued

Allowed:
Body Params

Send exactly one of url or uploadKey.

uri

Public http(s) URL of the audio or video. Private and internal addresses are refused.

string

Key returned by POST /v1/transcripts/upload, after the file was PUT to its uploadUrl.

string

ISO 639-1 code (pt, en, es, ...) or auto (default). About 99 languages are supported.

uri

Called once with POST { "data": { "id", "status" } } when the transcript ends (done or error). One attempt, no retries: read the result with GET /v1/transcripts/{id}.

glossary
object

Names and terms that must come out right: { "what it hears": "what to write" }. Up to 100 entries, 100 characters each. Used as a hint to the model and applied to the text afterwards.

Responses

Language
Credentials
URL
LoadingLoading…
Response
Click Try It! to start a request and see the response here! Or choose an example:
application/json