See llms.txt for all machine-readable content.

Back to Templates

Turn podcast episodes into captioned YouTube shorts with Klap and OpenAI

Created by

Created by: Andy || banandyy
Andy

Last update

Last update 3 days ago

Categories

Quick overview

Repurpose owned podcasts and webinars into selected captioned portrait clips with Klap. Review exact candidates, archive verified MP4s privately in Google Drive, and create OpenAI metadata drafts from observed clip context. Nothing is automatically published.

How it works

  1. Reads approved ready source URLs from Google Sheets, rejects duplicate job IDs and records one authoritative versioned pipeline state in a native n8n Data Table.
  2. Saves an intent before creating a Klap API-key v2 video-to-shorts task, then checks its status with bounded polling and reads candidate project previews.
  3. Requires an allowlisted signed-in n8n user to select exact candidate IDs with the current state version and hash, and to describe what each selected clip actually contains.
  4. Requests exports only for those selections, records generated Google Drive file IDs before creation, validates downloaded H264/AAC portrait MP4s, and uploads to the configured private folder using a saved resumable session.
  5. Checks the archived file's ID, folder and size, downloads it again and compares SHA-256. OpenAI drafts titles, descriptions and hashtags from the selected name and reviewer-observed clip context.
  6. Mirrors RAW status to Sheets and requires another exact-version final review before marking delivery approved. Lost submission responses stop for explicit reconciliation.
  7. An authenticated abandonment action can retire an unrecoverable job without replaying it. There is no automatic social publication.

Setup

  1. Import setup.json, workflow.json and review.json into the same compatible n8n project. Keep them inactive while configuring. Run the one-time setup utility with Google Sheets OAuth to create a separate PodcastQueue spreadsheet and initialized native Data Table.
  2. Copy the returned spreadsheet/table IDs, the pipeline ID and your own private Drive folder ID into both worker and review configurations. Replace exact source/export host allowlists and real reviewer email addresses.
  3. Connect Google Sheets and Drive OAuth, a Klap API-key v2 Header Auth credential, and an OpenAI API credential. Keys belong in n8n Credentials. Confirm service entitlements, source permissions, costs and a model supporting the documented JSON schema response.
  4. Run the default synthetic preview demo. For approved real work, set demoMode=false and processingEnabled=true, add a unique approved ready source row and run the worker repeatedly after recorded poll times.

Use the authenticated review companion with fresh versions and hashes for candidate selection and final delivery approval. Follow the bundled exact phrases and recovery guide. Review caption accuracy, clip meaning, metadata and private Drive files before any separate publication.

Requirements

  • Compatible n8n with native Data Tables and n8n User Auth Forms; target local version 2.28.6. Google Sheets/Drive OAuth, a private Drive folder, Klap API-key v2 access with sufficient credits and an OpenAI API account. Reviewers need allowlisted n8n accounts. Owned public HTTPS source media, exact trusted source/export hosts, 1–3 requested clips, a 20–60 second maximum and regular 1080×1920 H264/AAC exports with audio within the configured byte limit. Service and hosting charges are separate. Only local/mock tests are included; real provider output and n8n Cloud are not validated.

Customization

  • Set language, optional Klap caption/style preset, requested count, maximum clip duration, source context, OpenAI model and reviewer allowlist. Use your own exact source/export hosts and private resource IDs. Change worker and review policy settings together before starting a job; active jobs reject policy changes. Keep intent-before-effect persistence, exact-version selection, media/read-back verification and n8n User Auth intact. Publishing integrations are a separate customization requiring their own review and tests.

Additional info

This original package processes existing recordings rather than generating new scenes. Klap API-key v2 project data does not promise transcripts; metadata therefore uses the reviewer-observed clip context. Candidate virality scores are provider predictions, not observed views or sales. The default demo is fictional; the included local MP4 is synthetic, not Klap output. No live Klap video/export calls, worker OpenAI metadata calls or actual Google account tests were performed. Lost task/export responses require manual original-ID reconciliation, and interrupted uploads are held rather than blindly repeated. A final approval marks private delivery only; the template never publishes a social post. Metadata completions are capped at 800 tokens. Exact-version abandonment keeps the original intent in an audit and authorizes no provider cancellation, retry or replacement.