Short answer: n8n needs no special node. Three HTTP Request nodes submit the link, wait for the job and export the transcript. For an n8n AI Agent, connect the MCP Client Tool node to the Transcript Dock MCP server instead.
1. Add the credential#
- Create an API key on the API keys page.
- In n8n, create a credential of type
Header Authwith nameAuthorizationand valueBearer td_live_YOUR_KEY.
2. Submit the link#
Add an HTTP Request node: method POST, URL https://www.transcriptdock.com/v1/jobs, authentication Generic Credential Type with the Header Auth credential. Add a header Idempotency-Key with an expression such as transcriptdock-{{ $execution.id }}-{{ $itemIndex }}, and send this JSON body:
{"source": { "url": "{{ $json.url }}" },"mode": "captions_only"}
3. Wait for the job#
A second HTTP Request node: method GET, URL https://www.transcriptdock.com/v1/jobs/{{ $json.id }}?wait=25, same credential. It returns as soon as the job finishes. Caption jobs finish in a few seconds; for long AI transcriptions, loop back through an IF node until status is succeeded, failed or cancelled.
4. Export the transcript#
A third HTTP Request node: method GET, URL https://www.transcriptdock.com/v1/transcripts/{{ $json.result_id }}/export?format=txt, response format Text. Use srt or vtt for subtitles, or call /v1/transcripts/{{ $json.result_id }} for JSON with timed segments. Exports are free.
For an n8n AI Agent#
Add an MCP Client Tool node to the agent with endpoint https://www.transcriptdock.com/mcp, transport HTTP Streamable, and the same Header Auth credential. The agent gets get_video_transcript, YouTube search, channel and playlist listing, and TikTok profile listing. Tool details are on the MCP page.