TECHNOLOGY

BetterWispr, a free open-source Mac dictation app, runs speech models on the device

BetterWispr is a free, open-source dictation app for macOS that turns speech into text in any app. It transcribes on the Mac itself, with no account needed.

MacBook on a desk with a text document open, a hand resting near the Option and Space keysTECHNOLOGY

Image: BetterWispr

BetterWispr is a free, open-source voice dictation app for Mac. You hold a shortcut, speak, let go, and the text appears in whatever app you were typing in.

The default shortcut is Option + Space. You can switch from hold-to-talk to press-to-toggle, or record your own shortcut using Command, Option, Control or an F key.

The app describes itself as a private, offline speech-to-text tool and an open-source alternative to Wispr Flow. It transcribes on your Mac using Whisper, NVIDIA Parakeet or Apple's on-device speech recognition. According to the project, your voice does not leave your computer unless you add an API connection yourself. There is no account, and no cloud fallback.

Four on-device models are listed: Apple speech, Parakeet v3 (25 European languages), Parakeet v2 (English) and Whisper large-v3 turbo. Apple's model needs no download. The others download once and then work offline.

The app also cleans up what you say. It drops the ums and false starts, and lets you set spelling hints and phrase replacements for names and jargon. If it mishears a word, you can fix it once, either in History or right after it is pasted, and it should spell it correctly next time.

Voice commands are supported too. You can say "comma", "new line", "scratch that" or "sorry, remove that". Saying "at the rate KV" gives you "@KV". The project also says you can set a writing style for each app.

A meetings feature shows a live transcript of you and the other side. It can also produce titles, summaries and action items on the device, using Apple Intelligence or a local Ollama model. History keeps both raw and corrected transcripts on your Mac, and you can turn that off.

If you want to use a cloud service, you can add your own model. The listed options are Sarvam AI, Smallest AI, or any OpenAI-compatible endpoint. Once a connection is in use, your audio goes to that endpoint, and its pricing and retention policies apply. API keys are stored in the macOS Keychain. The project says BetterWispr never switches to a cloud provider on its own.

The app needs macOS 14 or later, and Apple Silicon is recommended. To set it up, download the DMG from the project's Releases page and drag the app to Applications. Then pick a model and give it Microphone and Accessibility access, so it can paste for you.

If the app cannot safely paste into the previous app, it copies the text to your clipboard instead. Recordings stop automatically after two minutes.

Sources and further readingBetterWispr: local dictation for macOS ↗
ABOUT THE DESK

Harsh DV

IntraGoals reports on important changes in technology and work. We check each story for clear writing, trusted sources and useful information before it is published.

KEEP READING

Latest from IntraGoals.

All latest stories ↗