Dict runs on macOS, Linux, and Windows. We said we would, and then we actually did it.
Pick any shortcut you like. The default was never going to fit everyone.
Every transcription is logged so you can grab something you said five minutes ago.
Optional managed cleanup if you do not want to run your own model, plus over-the-air updates so you never re-download.
Right now it is English and Portuguese. Whisper speaks plenty more, so Spanish, French, German, and friends are next. Auto-detection would be nice too.
The 15-second cap keeps things snappy, but sometimes you have more to say. We want a sane way to lift it.
Automatically adjust LLM correction style based on the active app. Casual in Slack, formal in Mail, code-aware in VS Code.
A mode where you don't have to keep pressing the hotkey. Start once, keep talking, automatic sentence detection and pasting.
Trigger Dict from Raycast or Alfred. Search your snippets and commands. Power user stuff.
Sync config, snippets, commands, and dictionary across your machines. Bring your own storage, no server of ours required.
Export your snippets and commands to share with others or backup. Import from other dictation apps.
A lightweight iPhone app that shares your snippets and lets you dictate on the go. Big maybe.
Drop an audio file onto Dict to transcribe it. Handy for meeting recordings and voice memos.
Identify who's speaking in multi-person recordings. "Person A said X, Person B said Y." Very ambitious.
Speak in one language, paste in another. Probably slow and impractical, but would be cool.
We're building this for us, but we're open to suggestions.
Please do not ask for more things though. The backlog is already terrifying. 😅
Send me an email anyway