ViralCutter RafaelGodoyEbert/ViralCutter

Local Gradio app that cuts long YouTube videos into vertical shorts with LLM moment picking, WhisperX captions, face tracking and split screen.

Creator kit + Compare GitHub ↗ Install guide (official) ↗

ViralCutter at a glance

Checked 16 Oct 2026

ViralCutter (0.8 alpha) is a Python/Gradio WebUI and CLI that picks hooks from a long video with Gemini, GPT-4 or a local GGUF model, transcribes with WhisperX, auto-crops to 9:16 with face tracking or a two-person split screen, and burns word-by-word captions; it can also translate captions and export Premiere Pro XML. GPL-3.0. Install scripts are Windows .bat files; a Colab notebook is offered. NVIDIA GPU strongly recommended. No virality score and no direct social upload (on the roadmap). No tagged releases.

Cite this pageRepoQuarry. “ViralCutter.” Checked 16 Oct 2026. https://repoquarry.dev/projects/viralcutter

How much of Opus Clip does it replace?

3 of 5 jobs. Covers picking, reframing and captions locally; no virality score or social publishing, and it is alpha software.

AI picks highlight clips from a long video Partial replacement

Auto-reframe to vertical 9:16 (speaker or face tracking) Partial replacement

Animated burned-in captions on clips Partial replacement

Virality score per clip Not covered

Schedule and publish clips to social platforms Not covered

For virality score per clip, schedule and publish clips to social platforms, see other Opus Clip alternatives. You may also need to build an interface, an API server and hosting yourself.

Facts, with sources

Every answer links to where we found it. We read the documentation; we did not install it.

Report a correction

QuestionAnswerSourceChecked

Runs on my machine? Yes, local WebUI or CLI; the LLM can be a local GGUF model or a cloud API All running on your machine. github.com · README_en.md ↗ 1d ago

Code licence GPL-3.0 (copyleft) github.com · LICENSE ↗ 1d ago

Model weights licence InsightFace pretrained models are for non-commercial research; commercial licensing must be obtained separately. Whisper weights are MIT, and user-selected local GGUF or hosted LLM options retain their own terms The training data containing the annotation (and the models trained with these data) are available for non-commercial research purposes only. github.com · insightface#license ↗ today

Commercial use GPL-3.0 code allows commercial use with copyleft conditions; licences of the models it uses are not stated github.com · LICENSE ↗ 1d ago

Paid services needed? None required; Gemini (free key) or GPT-4 optional, or run fully offline with local GGUF models **Flexible**: Gemini (Free), GPT-4, **Local GGUF (Offline)** github.com · README_en.md ↗ 1d ago

Hardware NVIDIA GPU with CUDA 12.4+ strongly recommended; exact VRAM and RAM not stated **NVIDIA GPU** is highly recommended for speed and local AI operations. github.com · README_en.md ↗ 1d ago

Operating systems Windows install scripts documented; Google Colab notebook; macOS and Linux not documented Open the ViralCutter folder and double-click **one of the installers** below github.com · README_en.md ↗ 1d ago

Setup effort Medium; needs Python 3.10/3.11, FFmpeg, NVIDIA drivers and C++ Build Tools, then a .bat installer Required to compile `insightface` and avoid "Cpp/Visual Studio" setup errors. github.com · README_en.md ↗ 1d ago