CS student from Lahore, Pakistan - building MoodCutter, an AI-powered multimodal emotion detection engine for video content. MoodCutter reads the full emotional signature of a video - laughter in the audio, energy in the voice, expression on the face - and automatically clips the moments that make people feel something.
Focused on systems-level understanding and mathematical foundations before specialization. Building in public.
"Other tools clip what you said. MoodCutter clips what you felt."
A multimodal AI virality engine for content creators. Six-signal emotion detection pipeline: speech sentiment, laughter detection, voice intensity, facial emotion, crowd reaction, and a fusion layer that combines all signals into a single Virality Score with plain-language reasoning.
No competitor does this. Existing tools - OpusClip, TuBoost - detect what you said. MoodCutter detects what you felt.
| Repository | Status | Description |
|---|---|---|
| MoodCutter | Archived ✅ | POC - audio-based emotional moment detection |
| MoodCutter-Community | Active 🔧 | Open source - three-signal pipeline, free, self-hosted |
| MoodCutter-Pro | Private 🔒 | Full product - six signals, hosted, coming soon |
Stack: Python · WhisperX · PyTorch · HuggingFace · librosa · OpenCV · FastAPI
Acoustic guitar · Film photography · Painting · Chess
