Drop in a YouTube link → get back scored, captioned clips (16:9 or 9:16) — 100% on your own machine.

---

Chopify is a free, open-source, fully local alternative to Opus Clip, Vizard, Klap, 2Short and Submagic. It automatically turns long videos — YouTube videos, podcasts, interviews and webinars — into short, ready-to-post viral clips in 16:9, 9:16 (vertical) or 1:1, with AI virality scoring, automatic face-tracking reframe, and burnt-in word-by-word captions. Everything runs 100% on your own machine: no subscription, no cloud upload, and no API keys.

Tip

New in v1.1 — output is now 16:9 full-frame by default (the whole screen, no crop). Add --aspect 9:16 for vertical or --aspect 1:1 for square, and every clip auto-saves a PNG poster next to the .mp4.

Demo

How it works

Features

• AI virality scoring — every segment is rated on hook, shock, humour, controversy, insight, emotion, energy and complete-arc; only 8+/10 clips are kept.
• Local transcription — faster-whisper with word-level timestamps. Nothing leaves your PC.
• Multi-format output — 16:9 (default, full frame), 9:16 (speaker-tracking auto-reframe via OpenCV YuNet, snap-on-cut + edge guard), or 1:1.
• PNG poster per clip — a ready thumbnail saved next to every clip.
• Word-by-word captions — bold CapCut style, burnt in with ffmpeg.
• Zero paid APIs, zero cloud — yt-dlp + faster-whisper + ffmpeg + OpenCV.
• Content decides the count — a 1-hour video might yield 2 clips or 20.

Chopify vs. the paid tools

Chopify does the core of what these subscription products do — AI-scored clips, auto-reframe to 9:16, and burnt word-by-word captions — except it runs free, on your own machine, and fully open-source.

Product	Pricing	Runs	Open source
Chopify	Free	Local (your PC)	MIT
Opus Clip	Paid subscription	Cloud	No
Vizard.ai	Paid subscription	Cloud	No
Klap	Paid subscription	Cloud	No
2Short.ai	Freemium + paid	Cloud	No
Munch	Paid subscription	Cloud	No
Spikes Studio	Freemium + paid	Cloud	No
Submagic	Paid subscription	Cloud	No
SendShort	Paid subscription	Cloud	No

_{Independent products and trademarks of their respective owners; pricing and features change over time. Comparison covers the long-video to short-clip workflow only.}

Requirements

• Windows, Python 3.11+
• ffmpeg + ffprobe on PATH — winget install Gyan.FFmpeg
• deno on PATH (yt-dlp JS-challenge solving) — winget install DenoLand.Deno

Install

python -m venv .venv
.venv\Scripts\python -m pip install -r requirements.txt

Usage

1. Download + transcribe → writes work/transcript.json:

.venv\Scripts\python download_and_transcribe.py "<YOUTUBE_URL>"

2. Score → write work/segments.json with the segments to clip:

{ "start": 134.2, "end": 187.6, "hook": "short title for the filename", "overall": 8.4 }

Scoring reads transcript.json and rates segments on the eight criteria above. In the reference setup an LLM agent does this in-loop (zero API cost); score it however you like.

3. Render → cut + reframe (16:9 default; --aspect 9:16 / 1:1) + burnt captions + a .png poster → C:\clips:

.venv\Scripts\python render_clips.py work

FAQ

Is there a free alternative to Opus Clip? Yes — Chopify is a free, open-source, self-hosted alternative to Opus Clip, Vizard and Klap. It runs entirely on your own computer, with no subscription and no API keys.

Can I turn long videos into short clips without paying? Yes. Chopify downloads a video, transcribes it locally, AI-scores every segment for virality, and exports the best moments as ready-to-post clips — for free.

Does it run locally, offline and privately? Yes. Download, transcription, scoring and rendering all happen on your machine. Nothing is uploaded to the cloud.

What formats / aspect ratios does it export? 16:9 (landscape, default), 9:16 (vertical for TikTok, Reels and YouTube Shorts) and 1:1 (square). Each clip also gets a PNG poster.

Do I need an OpenAI, Gemini or other paid API key? No. Chopify uses only open-source tools: yt-dlp, faster-whisper, ffmpeg and OpenCV.

What is it good for? Repurposing podcasts, interviews, webinars, lectures and long YouTube videos into short-form clips for TikTok, Instagram Reels and YouTube Shorts.

Notes

• GPU: faster-whisper (CTranslate2) is CUDA / NVIDIA-only; on AMD it runs on CPU (whisper.cpp + Vulkan was attempted for AMD but is unstable on RDNA3 — crashes or returns corrupted output).
• Captions use a built-in ffmpeg ASS renderer (the PyPI pycaps is an empty stub).
• Output folder is C:\clips — change OUT_DIR in render_clips.py.

_{Personal / educational use — you are responsible for the rights to any video you process.}