For developers
Dictate commits, PR descriptions, Slack messages, and docs. Keep your hands on the keyboard for the actual code.
No account needed. Works offline.
The reality
Look at any honest engineer's day:
Use cases
"feat(auth): replace session-cookie path with the OAuth refresh token flow we discussed yesterday — context in PR-1843, breaking change for any client still on /v1/login." A long commit message is a few seconds of speech instead of a couple of minutes of typing.
The PR template asks for context, screenshots, test plan, breaking changes. By PR five, you give up. Voice helps. Speak the description while the build is running.
Long Slack replies are friction. Voice makes them less of one — you answer instead of leaving the thread on read.
A thorough ticket and a sloppy ticket end up costing roughly the same effort once you stop typing them. Voice removes the reason to skip.
ADR drafts, README rewrites, JSDoc on a complex function. Speak the explanation while the logic is fresh — clean it up after.
Every dev tool
Same hotkey, every text field. SnailText pastes the same way Ctrl+V does — anywhere a keyboard works, dictation works.
IDE & terminal
Code review & PM
Comms
Plus every other text field — no integration list to maintain.
Custom dictionary
Out of the box Whisper handles English well, but it has no idea what your stack is called. "Cubernetes". "G-R-P-C". "Async, A-wait." That part is on you — but only once. Add the words to the custom dictionary and SnailText replaces them on the way to the text field:
Word-boundary aware. Case-preserving. No regex. The dictionary is a manual list — nothing happens until you add an entry.
Privacy
Some commit messages contain sensitive context. Internal architecture. Customer names. Auth flow details. The whole point of dictation is that the words go straight from voice to text — but if those words go through a cloud STT service first, every dictation is a tiny exfiltration.
SnailText runs Whisper locally. Audio is processed in RAM and discarded as soon as the text is ready. Nothing on disk, nothing on a server. Verifiable in your network tab — no outbound STT traffic during dictation.
If your company has a policy about where source code or customer data can travel, this matters.
Pricing
The compact local models handle commits, PRs, Slack, and most docs without trouble. Pro is for the harder cases — long technical paragraphs, non-English transcription, or accuracy on rare terms.
Free
$0 always
Compact local models. Unlimited dictation. No account.
Pro
$7.49 / month or $89/yr
Advanced local models. Up to 3 devices. 30-day refund.
FAQ
Common ones, mostly yes — better than OS-built-in dictation. Less common ones (kubectl, gRPC, your team's service names) it will mishear until you add them to the custom dictionary. Most developers end up with a list of 20–30 terms after the first week and rarely add to it.
Copilot Voice in VS Code is scoped to inline code commands. SnailText is system-wide dictation — works in commits, PRs, Slack, terminals, anywhere. Different tools, different jobs. Many developers use both.
Not in MVP. Mac and Windows day-one. Linux is on the roadmap but no committed date. If you want a Linux-first option, OpenWhispr is the best bet — it's MIT-licensed and has Linux builds.
Compact models on a modern CPU: typically 1–3 seconds for a short phrase, several seconds for a full minute of speech. The larger Pro models benefit from a GPU — we auto-detect Vulkan on Windows and Metal on Mac.
Try it
Free to start. About thirty seconds to install.
⌘ Shift Space — that's the only thing you need to remember.
