TL;DR
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
BetterWispr has introduced a free, open-source dictation app for macOS 14 and later. It records speech through a keyboard shortcut, transcribes it using a selected speech model, and can insert the result into the active app; the company says local models keep speech on the Mac.
BetterWispr has introduced a free, open-source dictation app for Macs running macOS 14 or later, offering local speech transcription and configurable text cleanup. The launch matters to users seeking an alternative to cloud-based dictation: the developer says recordings are processed on the Mac, though users must install a compatible model before using offline transcription.
The app is designed to dictate into other applications. A user focuses a text field, holds Option + Space while speaking, then releases the shortcut to finish recording. BetterWispr transcribes after recording stops, rather than displaying words live as the user talks, and attempts to place the text in the active app. If an app blocks insertion, users can retrieve and copy the transcript from the dashboard.
BetterWispr lists nine built-in speech models from Apple speech, Parakeet and Whisper. The available languages depend on the selected model: the product says Parakeet supports English, Japanese and 25 European languages, while Whisper offers broader multilingual recognition. Users can also configure connections to Sarvam AI, Smallest AI or an OpenAI-compatible API; using an external provider may send dictation to that provider.
Cleanup settings range from None, which retains the spoken words, to Light, which removes English filler words, repetitions and false starts. Medium can also edit English for clarity using a selected notes model. BetterWispr says Apple Intelligence or Ollama can perform that work locally. Its other features include app-specific writing styles, spoken punctuation commands, editable vocabulary hints, local transcript history and meeting notes that separate the user’s microphone from audio captured from other apps.
A Local Option for Mac Dictation
BetterWispr brings together dictation, configurable cleanup and app-specific styles in a free Mac application. Users can choose whether to preserve speech verbatim or remove some disfluencies, and can set different tones for different apps. That may suit people who dictate messages, email or notes but want control over how spoken language appears in writing.
The privacy distinction is central to the launch. The developer says built-in models run on the Mac, with no account and no cloud transcription fallback. That design limits the need to send speech to a transcription server, but it is not the only available configuration: the product also supports external APIs, and some model or feature choices may involve third-party processing. Users should check which provider and processing option they have selected.
The product’s page includes a typing-time illustration, but labels it “not a benchmark” and says it is simulated. It should not be read as independent evidence of speed or accuracy. Actual results may vary with hardware, model, language, microphone and speech habits.
As an affiliate, we earn on qualifying purchases.
How the Dictation Workflow Works
BetterWispr describes a three-step workflow: install the app and a speech model, place the cursor in a text field, then hold a shortcut while speaking. Releasing the key ends recording and starts transcription. Users can change the shortcut, including choosing press-to-toggle behavior, and the app requires permissions for microphone access and, for automatic insertion, Accessibility access. Apple speech also requires Speech Recognition permission.
The source says history is stored on the Mac and can be turned off; it can show both raw and corrected text. Users can restore the original wording or correct terms so the app learns names and vocabulary. The application is licensed under Apache 2.0, and the developer says the source code is available on GitHub. The source also says software updates are delivered through GitHub Releases using Sparkle and are installed only when their EdDSA signature matches.
“Speech stays on your Mac.”
— BetterWispr website
Performance and Provider Choices
The launch material does not provide independent accuracy tests, latency measurements or comparisons across the listed models. It also does not give a release date, adoption figures or a detailed account of how each optional external API handles speech and transcripts. The typing illustration is expressly not a benchmark, so it cannot establish time saved in everyday use.
Compatibility and results depend on the Mac, selected model and language. The developer recommends Apple Silicon, while saying macOS 14 or later is required. Offline use is available only after the relevant model and assets are installed; Apple speech also depends on supported languages and operating-system assets. The source says styles, cleanup and voice commands work in English, so support for those features should not be assumed for other languages.
Automatic text insertion may fail in secure fields or apps that block paste. The product says users can copy the transcript from its dashboard in those cases. Its materials do not provide independent verification of the privacy and security claims, so users handling sensitive speech may want to review the code and selected model or provider configuration.
Installation and Real-World Testing
Users can install BetterWispr, grant the permissions needed for their chosen features, and download a compatible local model. The app’s next practical test is whether transcription quality, language coverage and insertion behavior meet individual needs across their Mac and applications. Those results will depend on local hardware and the model selected.
The developer points users to GitHub to read the source, build the app or report issues. Future updates are described as arriving through signed GitHub releases. The source does not announce a release schedule or further product milestones, so the next confirmed developments will depend on updates from the project and feedback from users.
Key Questions
Is BetterWispr free?
Yes. The developer says the app is free and licensed under Apache 2.0, with no dictation subscription, account or API key required for its built-in local models. Model files use storage on the Mac.
Does BetterWispr send recordings to the cloud?
The developer says built-in models transcribe speech locally and that there is no cloud transcription fallback. The app also offers optional external API connections; users choosing one should check that provider’s terms and data practices.
Which Macs and languages does it support?
BetterWispr requires macOS 14 or later, and Apple Silicon is recommended. Language support varies by model: the developer lists Japanese and 25 European languages for Parakeet models and broader multilingual recognition for Whisper. Cleanup, styles and voice commands are described as English-only.
Does text appear as I speak?
No. According to the product information, transcription starts after recording stops. Users hold Option + Space while speaking and release it to finish.
Can I keep my words exactly as spoken?
Yes. The app offers a None cleanup setting to keep every word. Light cleanup removes English filler words, repetitions and false starts; Medium can also edit English for clarity using a notes model.
Source: hn
Columbus Day / Indigenous Peoples' Day Picks
long weekend sales
As an affiliate, we earn on qualifying purchases.
