BetterWispr is an open-source dictation application for macOS. According to the maker, it transcribes speech directly on your Mac using local models—such as Apple speech, Parakeet, and Whisper—without requiring an account or sending audio to cloud servers. Users can press or hold a keyboard shortcut to dictate into text fields across any app. The software also removes filler words, customizes tone per application, learns custom vocabulary, and includes a meeting notetaker.
Who it’s for
Mac users who want private, offline speech-to-text dictation and meeting transcription without subscriptions or cloud processing.
What it does, according to the maker
- Shortcut-based dictation (such as Option + Space) that inserts text directly into active text fields.
- Local speech transcription using built-in models from Apple speech, Parakeet, and Whisper, with optional support for external APIs like Sarvam AI or OpenAI-compatible endpoints.
- Adjustable English cleanup settings (None, Light, Medium) to strip out filler words, stuttering, and false starts.
- App-specific writing styles that adapt capitalization and punctuation for personal chats, work communication, or formal writing.
- Vocabulary learning that remembers custom names, technical terms, and corrections made in history.
- Notetaker feature that captures both microphone and system audio to generate meeting summaries and action items using local models like Ollama or Apple Intelligence.
- Local dictation history that retains both raw and corrected transcripts, with the option to restore original wording.
How to get started
- Download BetterWispr for Apple Silicon or Intel from the website.
- Install the application and follow the welcome setup to grant Microphone, Accessibility, and optional Speech Recognition permissions.
- Download your chosen local speech model.
- Click into any text field, hold Option + Space, speak, and release the keys to transcribe.
Pricing
Free and open source under the Apache 2.0 license.
Availability
macOS 14 or later (Apple Silicon recommended; Intel supported).
Good to know
- Models download once to local storage and do not require an active internet connection to transcribe.
- Tone adjustments, filler word cleanup, and voice editing commands currently apply to English.
- If an app blocks automatic pasting, BetterWispr copies the transcribed text to your clipboard instead.
- Meeting audio separation ('Them' audio) requires macOS 14.2 or later.
- Software updates are delivered via GitHub Releases through Sparkle with verified EdDSA signatures.