Paste a YouTube URL or upload a video file for processing.
AMD Developer Hackathon · Track 3
Turn long videos into short-form clips.
ElevenClip.AI uses Whisper, Qwen, Hugging Face, and AMD ROCm on MI300X to find highlight moments, render subtitles, and give creators a human-AI editor for TikTok, Shorts, and Reels.
Pipeline
"The moment viewers stop scrolling"
"A practical takeaway in 60 seconds"
How It Works
Whisper Large V3 creates timestamped multilingual transcripts.
Qwen2.5 ranks highlights using creator profile and engagement signals.
Creators trim, edit subtitles, approve, regenerate, and download clips.
Hackathon Fit
Backend target is AMD Developer Cloud with Instinct MI300X.
Designed for PyTorch ROCm, vLLM ROCm backend, and Optimum-AMD.
Uses HF model hub for Whisper, Qwen2.5, and Qwen2-VL.
Combines audio, text, video frames, subtitles, and rendered clips.
Project Status
Local MVPWorking
Upload to clipsWorking
Subtitle renderingWorking
Human editorWorking
AMD Cloud creditsRequested
Real Whisper inferencePending
Real Qwen inferencePending
MI300X benchmarkPending