n8n Video Narration Workflow: Multimodal AI Script + Voiceover

Turn any video into a narrated voiceover—automatically (with multimodal AI)

This n8n template takes a video, extracts key frames, uses a multimodal LLM to write a script from the visuals, then generates a matching voiceover clip from that script and uploads the finished audio to Google Drive.

What this workflow does

• Downloads your video using the n8n HTTP node.
• Extracts frames with a Python code node (using OpenCV), converting the video into images the LLM can understand.
• Batches frames with the Loop node to generate partial scripts for different sections of the video (improves feasibility for longer inputs).
• Combines partial scripts into a single complete script.
• Generates narration audio by sending the full script back to the same multimodal LLM for voiceover creation.
• Uploads the finished voiceover clip to Google Drive.

Use cases

• Content repurposing: convert training or demo videos into a narrated explainer with a clean script.
• Podcast-style marketing: turn product walkthrough footage into voiceover content you can publish or reuse.
• Accessibility & localization: generate structured narration that can be further processed for captions/audio workflows.
• Automation engineering demos: document system behavior from screen-capture videos with AI-generated narration.

Technical details

• Python + OpenCV for frame extraction.
• n8n nodes include Code, Wait, Split Out, Aggregate, Edit Image, and Sticky Note.
• OpenAI for LLM and voice generation.
• Google Drive upload for the resulting voiceover clip.
• Hardware tip: ideally a mid-range (16GB RAM) machine for acceptable performance.

Inspired by Processing and narrating a video with GPT's visual capabilities and the TTS API (OpenAI Cookbook).