Clip a podcast into Shorts without uploading it anywhere

ShortsFlow is a free, open-source tool that cuts a long podcast or interview into captioned vertical clips on your own computer. No account, no API key, no watermark, no credits. Your video is not uploaded.

Download for Windows Mac, Linux, pip, Colab

Open source (MIT). Windows installer, or pip install "shortsflow[clip]" on any system.

How it works

Everything below runs on your machine.

1. Read the speech

Whisper runs locally and writes a transcript with word timings. Nothing is sent anywhere.

2. Pick the moments

Clips start and end on whole sentences, skip a mic check or a mumbled opening, and never overlap. Each clip says why it was picked.

3. Frame and caption

A face-tracked 9:16 crop, a cut to whoever is speaking in a two-person talk, and animated word-by-word captions you can restyle.

pip install "shortsflow[clip]"
shortsflow doctor
shortsflow clip my_episode.mp4 --captions

ShortsFlow compared with hosted clip tools

An honest comparison. Hosted tools such as Opus Clip, Klap and Vizard are polished and need no setup, and ShortsFlow does not claim to beat them on every video.

Where your video goes

ShortsFlow keeps it on your machine. Hosted tools upload it to their cloud.

Price and watermark

ShortsFlow is free (MIT) with no watermark. Hosted tools use subscriptions or credits, and the free tiers usually add a watermark.

Account

ShortsFlow needs none. Hosted tools require one.

Choosing the moments

Without an AI key, ShortsFlow picks moments with transparent text and audio rules, which is weaker than an AI that reads the whole transcript. Adding a key (or a free local Ollama) lets an AI pick and score them, and then the transcript text goes to that provider.

Convenience

Hosted tools win here: nothing to install, and a web editor. ShortsFlow asks you to install a program.

Results from public-domain and Creative Commons videos, including the weak spots, are on the benchmarks page.

Questions

Is it really free?

Yes. It is MIT licensed, with no credits and no watermark. Optional AI providers have their own pricing if you choose to add a key. Remotion, used for animated captions, has its own licence for companies above its size threshold.

Does anything leave my computer?

Not by default. Only optional features send data: an AI text provider receives the transcript if you add a key, and Google Drive upload happens only when you ask. A pasted YouTube link is downloaded to your PC.

Do I need a GPU?

No. A GPU makes transcription faster, but it runs on a CPU as well, only more slowly.

Which formats come out?

Vertical 9:16 MP4 clips with burned-in captions, sized for YouTube Shorts, TikTok and Instagram Reels. You review and post them yourself.

Something broke. What now?

Run shortsflow doctor and open an issue with its output.