AI Video Narrator Agent using OpenAI Vision & TTS

Unlock Automated Video Narration with this AI Agent

This AI Agent transforms your raw video footage into engaging, narrated content automatically. It starts by downloading a video, then intelligently extracts a series of representative frames using OpenCV within a Python environment. These frames are processed in batches and fed to OpenAI's powerful GPT-4o vision model, which analyzes the visual content to generate a descriptive and coherent narration script. The agent is designed to build the script iteratively, ensuring context is maintained across different segments of the video. Once the full script is complete, it utilizes OpenAI's Text-to-Speech (TTS) capabilities to create a natural-sounding audio voiceover. The final MP3 voiceover is then conveniently uploaded to your Google Drive.

This agent empowers you to add professional voiceovers to marketing materials, product demos, educational content, or social media videos with minimal effort, saving significant time and resources.

Key Features & Benefits

• Video Understanding: Leverages OpenCV for frame extraction and OpenAI's GPT-4o vision to analyze and interpret video content.
• AI Script Generation: Automatically creates compelling narration scripts tailored to the video's visuals, with support for iterative script building for longer videos.
• High-Quality Voiceover Creation: Employs OpenAI's advanced Text-to-Speech (TTS) to produce clear, natural-sounding audio narrations in MP3 format.
• Batch Processing: Efficiently handles videos by processing frames in manageable batches to stay within LLM token limits and manage resources.
• Automated Upload: Seamlessly uploads the final voiceover audio file to a specified Google Drive folder.
• Streamlined Content Creation: Drastically reduces the manual effort and cost associated with video narration and voiceover production.

• An n8n instance (Cloud or self-hosted).
• OpenAI API Key with access to a vision-capable model (e.g., GPT-4o like gpt-4o-2024-08-06 used in the template) and TTS models.
• Google Drive credentials configured in n8n for the upload functionality.
• Ensure your n8n instance has sufficient memory, as video frame processing can be resource-intensive, especially for longer videos or a high number of captured frames.

1. Download the n8n workflow JSON file.
2. Import the workflow into your n8n instance.
3. Optional: Modify the 'Download Video' (HTTP Request node) URL to use your own video source. Ensure it's publicly accessible or adjust authentication as needed.
4. Configure your OpenAI credentials:
   • Select your OpenAI API credential in the 'OpenAI Chat Model' node. This model is used by the 'Generate Narration Script' node.
   • Select your OpenAI API credential in the 'Use Text-to-Speech' node (which is an OpenAI node type configured for audio generation).
5. Configure the 'Upload to GDrive' node: select your Google Drive credential, specify the target folder, and adjust the output filename pattern if desired.
6. Review and optionally adjust parameters in the 'Capture Frames' (Code node), such as max_frames, to control the number of frames extracted from the video. Note that the Python code uses OpenCV.
7. Customize the main prompt within the 'Generate Narration Script' node (an LLM Chain node) to alter the narration style (e.g., the default is David Attenborough style), tone, or specific instructions for the AI.
8. If needed, adjust the batch size in the 'For Every 15 Frames' (SplitInBatches node) based on your video length, LLM context window limits, and performance considerations.
9. Activate the workflow. Be aware that processing can take some time and consume resources, particularly for large video files or high frame counts.