TL;DR
Get business pricing on monitors, keyboards and dev gear
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
BetterWispr has been presented on Hacker News as a free, open-source dictation app for macOS 14 and later. It uses downloadable speech models on the Mac, offers adjustable transcription cleanup and app-specific writing styles, and does not require an account or cloud transcription service.
BetterWispr has been introduced as a free, open-source dictation app for macOS 14 and later, with speech transcription handled on the Mac using installed local models. The app is aimed at people who want to dictate into other applications without sending audio to a transcription service, while choosing how much the resulting text is cleaned up.
The app’s developer describes a workflow based on a keyboard shortcut: hold Option + Space while speaking, then release to transcribe and insert the text into the active app. Users can choose a different shortcut or switch to a press-to-toggle mode. The site says transcription begins after recording stops, rather than appearing word by word as someone speaks.
BetterWispr includes nine built-in speech models from Apple Speech, Parakeet and Whisper, according to its website. Language coverage depends on the selected model: the site lists Parakeet options for English, Japanese and 25 European languages, and describes Whisper as covering more languages. Users can also configure compatible services from Sarvam AI, Smallest AI or OpenAI-compatible APIs; choosing an external provider changes the local-processing arrangement.
Text handling is configurable. The default Light cleanup removes English filler words, repetitions and false starts; None keeps the spoken wording, while Medium also edits for clarity using a selected notes model. Users can set tones for different apps, add names and terms, and correct words in the app’s history. BetterWispr also offers meeting notes that distinguish the user’s microphone from audio coming from other apps, plus summaries and action items produced with Apple Intelligence or Ollama on the Mac.
Local Dictation Without a Subscription
BetterWispr’s pitch combines local speech processing with a free, open-source distribution model. That may appeal to people who dictate sensitive work, prefer not to create accounts, or want to avoid a recurring dictation subscription. The developer says the app has no cloud transcription fallback and that installed local models can work offline.
The project is licensed under Apache 2.0, and its site directs users to the source code on GitHub, where they can inspect, report issues or build the app. That gives technically inclined users a route to examine how the software is made, though the license and local-processing claims alone do not independently establish the app’s security or performance.
The app also targets a common friction point in dictation: turning spoken language into text that fits the destination. App-specific styles, optional cleanup and voice commands are intended to adapt the output without requiring users to edit every transcript manually. Those features are described by the developer; the source does not provide independent accuracy tests or user-performance measurements.
As an affiliate, we earn on qualifying purchases.
How BetterWispr Works
BetterWispr is presented through a Show HN launch post and its product website. The site describes an install-and-configure process that includes granting permissions and downloading a local speech model. It lists Apple Silicon as recommended, while stating that supported Macs must run macOS 14 or later. Model availability and speed will depend on the computer, selected model and language.
Local use does not mean the app never needs internet access. Model files must be downloaded before they can be used, and the developer says those downloads happen when users choose to install them. External API connections are also available as an option. The site says any use of those connections sends dictation to the selected provider; users can instead choose local models and local note-writing tools.
To put text into another app, BetterWispr uses accessibility permissions for automatic insertion and may rely on microphone and, for Apple Speech, speech-recognition permissions. The developer says that if an app blocks pasting, the transcript remains available to copy from the dashboard. Secure fields and some other apps may prevent automatic insertion.
“Speech stays on your Mac.”
— BetterWispr website
local speech recognition software for Mac
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Performance and Privacy Details
The available source is the product website, which describes the app’s features and privacy design but does not include independent testing of transcription accuracy, latency, model quality or resource use. The site’s sample timing illustration is explicitly identified as an illustration, not a benchmark. Actual results may vary by Mac, model, language, speaking style and recording conditions.
The source also does not provide a publication date for the launch post, download figures, user numbers or a detailed security audit. While the project is open source and says speech remains local when local models are selected, readers may want to inspect the code and configuration themselves. Some optional API connections send dictation to an outside provider, and support for languages and cleanup features varies by model and feature.
open-source speech transcription Mac
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Installation and Project Updates
People interested in trying BetterWispr can download it for a Mac running macOS 14 or later, follow the setup guide, grant the required permissions and install a compatible model. Apple Silicon is recommended by the developer. The source does not state a date for a future release or describe a product roadmap, so further milestones have not been announced in the material provided.
The website says updates are distributed through GitHub Releases using Sparkle and are installed only when their EdDSA signature matches. The project also invites users to report issues or build the software from its source code. Its practical performance and reliability will become clearer through testing across different Macs, models and supported languages.
As an affiliate, we earn on qualifying purchases.
Key Questions
Is BetterWispr free to use?
Yes. The developer says it is free and licensed under Apache 2.0, with no dictation subscription, account or API key required for its local-model workflow. Downloaded model files use storage on the Mac.
Does BetterWispr send speech to the cloud?
The developer says speech processed with installed local models stays on the Mac and that there is no cloud transcription fallback. However, users can select external API providers; the site says dictation is sent to a chosen provider when that option is used.
Which Macs and operating systems are supported?
BetterWispr requires macOS 14 or later. Apple Silicon is recommended for local model processing. The source says model availability and performance depend on the Mac and selected model.
Does transcription appear while I speak?
No. According to the product FAQ, recording ends when the user releases the shortcut, and transcription begins after recording stops. The site does not describe live, word-by-word transcription during speech.
Can users keep their exact spoken words?
Yes. Users can select None for cleanup to keep every word, or use Light cleanup to remove English filler words, repetitions and false starts. Medium cleanup also edits English for clarity using a selected notes model.
Source: hn
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
