This n8n workflow template turns a video into a narrated voiceover. It extracts key frames from the video, and a multimodal LLM writes a script based on what the frames show. The script is then converted into a voiceover clip, and the finished audio is uploaded to Google Drive.
The workflow downloads the video with the n8n HTTP node. A Python code node using OpenCV extracts frames and converts the video into images the LLM can read. The Loop node batches the frames so partial scripts are written for different sections of the video, which makes longer inputs more feasible. The partial scripts are combined into one complete script. That script is sent back to the same multimodal LLM to create the narration audio.
Suggested uses include turning training or demo videos into narrated explainers, turning product walkthrough footage into voiceover content, producing structured narration for caption or audio workflows, and documenting system behavior from screen-capture videos. The template is inspired by the OpenAI Cookbook example "Processing and narrating a video with GPT's visual capabilities and the TTS API."
- Type
- n8n workflow template
- Input
- Video downloaded with the n8n HTTP node
- Output
- Voiceover audio clip uploaded to Google Drive
- Frame extraction
- Python code node using OpenCV
- AI model
- OpenAI multimodal LLM for script and voice generation
- n8n nodes
- HTTP, Code, Loop, Wait, Split Out, Aggregate, Edit Image, Sticky Note, Google Drive
- Integrations
- OpenAI, Google Drive
- Hardware
- Mid-range machine with 16GB RAM recommended