The short answer
HeyCaption transcribes, styles and renders your captions on your Mac instead of on a server. There's no upload to wait for, nothing of yours stored in someone else's cloud, and the app keeps working with no internet connection at all.
It's not a privacy setting you have to find and switch on. It's how the app is built.
The upload tax
Most caption tools work the same way: you upload your video, it waits in a queue, a server transcribes it, and you download the result. For a 30-second clip that's a minor annoyance. For an hour of 4K podcast footage, the upload alone can take longer than the captioning.
Home and office connections are often much slower going up than coming down, so the slowest step is the one you can't speed up. Captioning on-device removes it.
Client footage stays put
If you edit for clients, the footage usually isn't yours to share. Some agreements limit where raw footage can be stored or which services may process it, and unreleased products, interviews, and medical or legal content often come with extra rules.
With on-device captioning there's no third party in the loop. The video, the transcript and the exported captions stay on the Mac you're working on. Our privacy policy spells out exactly what we do and don't collect.
Offline is a feature
Once the app and your language model are installed, HeyCaption doesn't need a connection. You can caption on a plane, on location, or on hotel Wi-Fi that can barely load a web page.
Why now: Apple Silicon
Not long ago, accurate speech recognition meant server hardware. Apple Silicon changed that: M-series chips are fast enough to run a capable speech model locally, so captioning a clip takes seconds instead of a round trip to the cloud. That's why HeyCaption is built for Apple Silicon Macs.
The trade-offs
Doing the work locally isn't free, and we'd rather be upfront about it:
- A one-time download. On first launch, HeyCaption downloads the speech model for your language. After that it stays on your Mac.
- Mac requirements. HeyCaption needs macOS 14 Sonoma or later on Apple Silicon.
- Audio still matters. On clear audio, accuracy is 95% or better. Heavy background noise or crosstalk makes any transcription harder, so a quick proofread is always worth it.
We think that's a fair deal for captions that are fast, private and work anywhere. The setup guide gets you from install to first export in a couple of minutes.