Description
n8n-nodes-transcriptfetch
This is an n8n community node for source=npm&utmmedium=referral&utm_campaign=n8n-readme”>transcriptfetch.com/docs/n8n
Quick links: source=npm&utmmedium=referral&utmcampaign=n8n-readme#key”>add your API key · source=npm&utmmedium=referral&utmcampaign=n8n-readme#templates”>templates · source=npm&utmmedium=referral&utmcampaign=n8n-readme#troubleshooting”>troubleshooting
Fetch transcripts from YouTube, TikTok & Instagram – with automatic AI transcription when captions don’t exist – directly inside your n8n workflows. Structured JSON output with per-segment timestamps, built for RAG, agents, and LLM pipelines.
n8n is a fair-code licensed workflow automation platform.
Installation
Follow the source=npm&utmmedium=referral&utmcampaign=n8n-readme#key”>add your API key and source=npm&utmmedium=referral&utmcampaign=n8n-readme”>transcriptfetch.com and create an API key in the source=npm&utmmedium=referral&utm_campaign=n8n-readme#key”>step by step.
Billing is per-credit: a successful caption fetch costs 1 credit, failed or blocked fetches are never charged, and every account gets 50 free credits a month. Videos without captions are transcribed from audio automatically (short videos typically finish in about 30 seconds) and are billed once, on delivery of the finished transcript, at the audio-transcription rate (see source=npm&utmmedium=referral&utmcampaign=n8n-readme”>pricing).
Trigger
TranscriptFetch Trigger starts a workflow when a YouTube channel, TikTok profile or Instagram account publishes a new video, and hands you the transcript in the same step, so you don’t need an RSS Feed Trigger plus a separate transcript call. Setup and examples: source=npm&utmmedium=referral&utm_campaign=n8n-readme”>TranscriptFetch monitor: activating the workflow creates a monitor that checks the channel on TranscriptFetch’s schedule and POSTs each new video to the workflow’s webhook URL, signed with the monitor’s secret (the trigger rejects anything unsigned). Deactivating the workflow deletes the monitor. Changing the node’s settings replaces the monitor on the next activation.
Billing. Every check costs 1 credit, whether or not it finds a new video, and so does the first check when the workflow is activated, which records what is already on the channel so the back catalogue is not replayed. A daily check uses about 30 credits a month. A caption transcript costs 1 credit; a video without captions is transcribed from its audio at 1 credit per started 5 minutes. Brand-new YouTube uploads usually get captions a little after they go live, so the monitor waits for them before falling back to audio.
The webhook URL must be public HTTPS. n8n Cloud always is. A self-hosted instance needs WEBHOOK_URL set to a public https:// address (a reverse proxy or a tunnel); otherwise activation fails with a message saying so. Without a public URL, run the TranscriptFetch node’s List Channel Videos on a Schedule Trigger instead.
Each item carries an event field and the video metadata (videoId, url, title, channel, duration, publishedAt, platform, eventId, monitorId, creditsSpent), plus a transcriptStatus ( and the same videoId |
| unavailable | The transcript couldn’t be fetched; see reason and message |
| skipped | Include Transcript was off |
Items with event: "video" are new uploads, oldest first. Items with event: "transcript" are transcripts that finished after their video was reported.
Upgrading from 0.4.x: the trigger is now version 2. Re-add it to existing workflows; the old polling version is gone.
Operations
Every operation with its inputs and outputs: source=npm&utmmedium=referral&utm_campaign=n8n-readme#agents”>use it with an AI Agent.
