An n8n workflow that turns podcasts, talks, and interviews into a daily stream of short-form video. Submit a long video through a form node; an FFmpeg job extracts the audio, OpenAI Whisper transcribes it with word-level timestamps, Gemini mines the transcript for 3–6 clips of 15–60 seconds and writes titles and descriptions, and Upload-Post’s GPU FFmpeg pipeline handles the cutting and cropping before scheduling one clip per day to TikTok, Instagram Reels, and YouTube Shorts.
It complements templates we already list such as Generate AI viral videos with Seedance and upload to TikTok: those generate new footage from scratch, while this one repurposes long content you already have — the right fit if you are sitting on recordings and stream archives.
Template page stats: author carlosgracia, published 2025-10-18, roughly 203 views (collected 2026-07-27) — the newest template in this batch.
What it does
- Accepts a video of any length through a form node, vertical or horizontal
- Extracts audio via an FFmpeg job and transcribes with Whisper for word-level timestamps
- Has Gemini select 3–6 hook-worthy segments (15–60 seconds), the count scaling with video length and material quality
- Cuts on absolute-second timecodes so clips do not start or end mid-word
- Respects source resolution rather than upscaling, cropping or padding intelligently for 9:16 delivery
- Auto-generates per-platform titles and descriptions
- Schedules one clip per day on consecutive days: 3 clips fill 3 days, 6 clips fill 6
- Polls job status with built-in waits and retries before retrieving finished clips
Prerequisites
- n8n, cloud or self-hosted
- An OpenAI API key for Whisper transcription
- Google Gemini credentials for the AI Agent node
- An Upload-Post account and API token — the same token covers both FFmpeg jobs and publishing
- TikTok / Instagram / YouTube accounts connected on the Upload-Post side
How to use
- Import the template and configure OpenAI and Gemini credentials.
- Generate an Upload-Post API token, connect your social accounts there, and add the token as an n8n credential.
- Change the timezone. The template posts at 20:00 Europe/Madrid by default — set it to your own timezone and your audience’s active hours.
- Adjust clip count, clip length, and target platforms as needed.
- Run a shorter video end to end first and check the cut points, cropping, and copy quality.
- Review the generated titles and descriptions; add your voice and hashtag conventions to the Gemini prompt if needed.
Use cases
- Podcast and long-interview creators splitting one episode into a week of shorts
- Agencies and social teams converting webinars and stream replays into a reliable short-form pipeline
- Brands and founders keeping a daily cadence without a dedicated editor
- Teams with a large recording archive who want to start a short-form channel cheaply
Notes
- The dependency chain is long (Whisper, Gemini, Upload-Post FFmpeg, plus each publishing platform). A bad credential anywhere stops the run — verify node by node on first setup.
- The default schedule is Europe/Madrid, which means overnight posting for most other regions. Change it first.
- Upload-Post offers a free trial (no credit card, per its own description), but sustained use is paid. Cost also includes Whisper transcription, Gemini inference, and FFmpeg compute — see each provider’s official pricing page.
- The AI picks “viral” moments from transcript text alone; it does not see the picture. Review before publishing, especially for demos, charts, or visual punchlines.
- Publishing to live social accounts is not reversible. Schedule as drafts or private first, then open up once you trust the output.
- OpenAI, Gemini, Upload-Post, and all three publishing platforms are overseas services; mainland China access typically requires an international network.