Quick start
Request
Response
PollGET /transcribe/status?id={jobId}, or wait for the webhook. A completed job has the full text and timestamped segments (times in seconds):
data (data.text, data.data.segments) instead of under result.
Captions or speech-to-text
Setting
provider (or providerOptions) always runs speech-to-text and skips captions, so leave it out unless the video has none. Speech-to-text downloads the media to extract its audio. That download is billed at $0.01 per MB and itemized as downloadCostInDollars in the result.
When a video has no captions
Without aprovider, a video with no captions fails with TRANSCRIPTION_PROVIDER_REQUIRED and a quote for speech-to-text. Nothing is charged:
"provider": "importly" to accept the quote.
Languages
Passlanguage as a two-letter code, such as es.
Set it for non-English media whenever you use speech-to-text.
Provider-specific options such as AssemblyAI’s
speaker_labels go in providerOptions. See provider options.
Transcripts inside an import
To get the file and its transcript from one job, addincludeTranscript: true to POST /import. The result gains a transcript object, billed like a separate transcription. Use transcriptProvider, transcriptApiKey, transcriptLanguage, and transcriptOptions the same way as provider, apiKey, language, and providerOptions here. If no transcript can be produced, the import still succeeds and transcript is { "status": "skipped", "reason": "…" }.
Pricing
Failed transcriptions are refunded automatically. See Pricing & credits.
Full reference: POST /transcribe
Every parameter, provider options, and the status endpoint.