The split is unusual and useful: the catalogue tools answer keyless, and only submission, status, export and balance need SCRIPTIVOX_API_KEY. The lifecycle is worked out rather than assumed — transcribe_cancel releases reserved balance and is idempotent, transcribe_delete returns 409 CONFLICT while a job is still in flight so you cancel first, and export_transcript requires the job to be in completed status. The aliases transcription_url and transcription_status are kept for @scriptivox/mcp-server@1.0.x and are marked for removal in 2.0.0, so a new integration should not build on them.
A client for the Scriptivox transcription API, covering job submission, job lifecycle, export, and the account questions that come before either.
- Transcription of a public URL, or of a local file up to 5 GB driven through the upload flow internally, across the 119 supported languages
- Per-job options: explicit language, speaker count, diarization, alignment, an idempotency key and a webhook URL
- Job control — status polling for long runs, cancellation that releases the reserved balance, and soft delete of the record
- A cursor-paginated list of past transcriptions filtered by status or date range, with the transcript body left out
- Export as SRT subtitles, WebVTT or plain text, with segmentation controlled by max_words, max_chars, max_duration, sentence_aware, include_speakers and strip_chars
- Credit balance and available hours, plus supported languages, plans and API documentation
A Scriptivox account and its key in SCRIPTIVOX_API_KEY for anything that touches a job. get_supported_languages, get_pricing, get_product_info and get_api_docs answer without one, so the language list and the rates can be read before signing up.
One command plus a key — npx -y @scriptivox/mcp-server, then supply credentials
