n8n transcription node: turn audio and video into text in any workflow
JustTranscribe's community node is verified by n8n, so you add it from the nodes panel and every recording that enters a workflow — a voice note from Telegram, a meeting dropped in Google Drive, a new YouTube video — comes out as a timestamped transcript. One node creates the transcript from a public link or a binary file and waits until it is ready; the same node exports subtitles or a Word document, and an AI Agent can call it as a tool.
Verified by n8n and live on n8n Cloud
n8n approved the node on 2026-09-16 and published it on n8n Cloud the next day. Verified means n8n reviewed the package and lets instance owners install it straight from the canvas — no terminal, no custom image. It runs on the same account, the same REST API and the same limits as the app.
How to add JustTranscribe to n8n
About two minutes. The API key is the only thing you paste.
- 1Create an API key. Sign in to JustTranscribe and create a key at justtranscribe.ai/api-keys. Copy it once — it is shown a single time.
- 2Install the node. On n8n Cloud, or a recent self-hosted version, open the nodes panel on the canvas, search for JustTranscribe and choose Install under “More from the community” (instance owners and admins can install). On any self-hosted instance you can also go to Settings → Community Nodes → Install and enter n8n-nodes-justtranscribe.
- 3Add the credential. Create a JustTranscribe API credential and paste the key. n8n tests it against your account before saving.
- 4Pick an operation and run it. Choose Transcript → Create, set Source to Video URL or to the binary field that holds your file, and leave Wait Until Finished on: the node hands the finished transcript to the next node.
n8n's own instructions: installing verified community nodes · installing from Settings.
Resources and operations
- TranscriptCreate — transcribes a public video link or a binary file from an earlier node. With Wait Until Finished on (the default) it returns the finished transcript; Max Wait sets how long it holds, from 30 to 3,600 seconds, 15 minutes by default.
- TranscriptGet — returns one transcript by ID: status, timestamped segments, detected language and the AI analysis.
- TranscriptList — lists the account's transcripts newest first, with limit, offset and a status filter.
- TranscriptExport — downloads a finished transcript as SRT, WebVTT, TXT, CSV or Markdown into the item JSON, or as PDF or Word (DOCX) into a binary field. Timestamps and speaker labels are optional.
- TranscriptDelete — removes the transcript and its stored media.
- AccountGet — returns the account behind the API key; it is also what the credential test calls.
If Max Wait runs out, the node stops with a clear message while the transcript keeps processing on our side — fetch it later with Get. A failed transcription surfaces the real reason (a private video, a file over the limit) as the node's error, so an Error Trigger or the Continue On Fail setting can handle it.
Workflow ideas
Each chain is three or four nodes. The repository also ships an example workflow you can import.
- A file lands in a Google Drive folderGoogle Drive Trigger → Download file → JustTranscribe: Create (Binary File)→ the transcript saved as a Google Doc
- A voice note arrives in TelegramTelegram Trigger → Get file → JustTranscribe: Create (Binary File)→ the text sent back as a reply
- A channel publishes a new YouTube videoRSS Feed Trigger → JustTranscribe: Create (Video URL)→ a summary from your LLM node posted to Slack
- An email arrives with a recording attachedGmail Trigger → JustTranscribe: Create (Binary File) → Export (DOCX)→ the Word document emailed back
- Someone uploads a video in an n8n Formn8n Form Trigger → JustTranscribe: Create → Export (SRT)→ subtitles stored next to the video
- An AI Agent is asked about a videoAI Agent with JustTranscribe attached as a tool→ the agent transcribes the link and answers from the text
n8n audio to text, speech to text and YouTube transcripts — what it handles
Audio to text in n8n
MP3, WAV, M4A, AAC, OGG, OPUS and FLAC files from any node that outputs binary data: Google Drive, Dropbox, S3, email attachments, form uploads, Telegram. WhatsApp voice notes work exactly as they come off the phone (.opus).
YouTube and social video transcripts in n8n
Public links from YouTube, TikTok, Instagram, Facebook, Pinterest and Google Drive, plus MP4, MOV, WEBM and MKV files. The audio track is extracted for you; nothing to convert first.
Speech to text with timestamps, subtitles and speakers
Every transcript carries word-level timestamps, sentence-merged segments and automatic language detection (Spanish first, 100+ languages). Export gives you SRT or WebVTT subtitles directly; speaker labels and translation into 111 languages are available on the finished transcript in the app.
Compared with transcribing through the OpenAI node
n8n's OpenAI node can transcribe a recording too, within OpenAI's 25 MB per-file limit and without fetching links. JustTranscribe takes files up to 500 MB and 150 minutes, splits long recordings for you, fetches public video links, and returns subtitles and documents as well as text.
Limits and behavior
- Public links only — no private or login-required videos, no live streams. Up to 150 minutes per recording, 500 MB per file.
- Same brakes as the app and the REST API: 3 new transcripts per hour and one processing at a time during the beta, so a loop over several files should run them one after another.
- 120 API requests per minute per key. While it waits, the node polls every 5 seconds, far below that.
- One API key, one account, one set of limits everywhere — the key you use in n8n is the same one Zapier, Make and the REST API take.
- The first transcript is free with no account (up to 40 minutes); after that, a free account raises the limit to 150 minutes and 500 MB per file — unlimited transcripts during the beta, no card needed.
FAQ
Is the JustTranscribe node verified by n8n?
Yes. n8n approved n8n-nodes-justtranscribe on 2026-09-16 and published it on n8n Cloud on 2026-09-17. Instance owners and admins can install it from the nodes panel, and it is on npm for any self-hosted instance.
How do I install it on n8n Cloud?
Open the nodes panel on the canvas, search for JustTranscribe, select it under “More from the community” and choose Install. If it does not appear, check that verified community nodes are enabled in the Cloud Admin Panel.
How do I install it on self-hosted n8n?
Go to Settings → Community Nodes → Install and enter n8n-nodes-justtranscribe. Recent versions also show verified nodes in the nodes panel, the same way n8n Cloud does.
How do I transcribe audio in an n8n workflow?
Add JustTranscribe → Transcript → Create after the node that produces the file, set Source to Binary File and point Input Binary Field at it (usually data). With Wait Until Finished on, the output item carries the transcript: use the segments directly, or add Export for plain text, SRT or a Word file.
Can I transcribe WhatsApp or Telegram voice notes?
Yes — an .opus or .ogg voice note is accepted exactly as it arrives. Fetch the file with your messaging node and pass its binary field to Create.
Does the node wait for the transcript?
By default, yes: it polls until the transcript is complete or failed and returns it, up to Max Wait. Turn Wait Until Finished off to get the pending job back immediately and collect it later with Get.
Is there a trigger node?
No. Inside n8n the Create operation already waits for the result. For event-driven setups, the REST API accepts a webhook URL when you create a transcript, so an n8n Webhook node can receive the completion event.
Can an AI Agent use it as a tool?
Yes. The node is marked as usable as a tool, so you can connect it to an AI Agent node and let the agent transcribe a link when the conversation calls for it. On self-hosted n8n, whether community nodes may run as tools is an instance setting.
Is it free?
The first transcript is free with no account (up to 40 minutes); after that, a free account raises the limit to 150 minutes and 500 MB per file — unlimited transcripts during the beta, no card needed.
Prefer a different tool? See the Zapier integration, Make integration, REST API, MCP for Claude and ChatGPT.