BetterWispr: local Mac dictation with cleanup and meeting notes

Dictate into the app you already use

BetterWispr is a free, open-source dictation app for macOS 14 or later. Its basic workflow is to place a cursor in a text field, hold Option-Space while speaking and release the shortcut to insert the transcription.

Built-in speech models run on the Mac after their initial download. The publisher describes no account requirement and no cloud fallback for that local path. Users can also choose their own external API provider, so the selected model configuration matters when assessing where speech is processed.

Control editing as well as recognition

The app offers vocabulary hints and replacements for names or specialized terms. Its default light cleanup removes English filler words, repetitions and false starts. Medium cleanup also revises for clarity using Apple Intelligence or Ollama on the Mac; disabling cleanup preserves the spoken words.

App-specific tones distinguish personal messages, work applications and email. The site says these tone controls apply to English. History stores raw and corrected text locally, with an option to restore the original transcription or turn history off.

A notetaker separates microphone speech from other application audio and places the transcript beside the user's notes. Apple Intelligence or Ollama can then produce summaries, decisions and action items locally.

The landing page's typing-versus-speaking demonstration is explicitly an illustration rather than a benchmark. Its displayed speed should not be read as a measured productivity gain for a new user.

Compare the edited text with what you said

Franklin analysis: test a few sentences containing names, dates and deliberate self-corrections in the actual applications you use. Compare raw transcription with the cleaned version before relying on automatic insertion.

For meeting notes, inspect speaker separation and whether an action item reflects an agreed decision. A fluent summary can still overstate a tentative suggestion. Keep the model choice and cleanup level visible in the trial so differences in recognition are not confused with later rewriting.