Show HN: BetterWispr – Free, Open-source Dictation For Mac

TL;DR

Before you orderOffer from Amazon

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

Start your free Prime trial →Free trial for eligible customers · Cancel anytime

As an affiliate, we earn on qualifying purchases.Hide

BetterWispr has introduced a free, open-source dictation app for Macs running macOS 14 or later. It uses locally installed speech models to transcribe recordings on the Mac, then can clean up and insert text into the app the user was working in. Apple Silicon is recommended, and the product’s feature and privacy descriptions come from its developer.

BetterWispr, a free, open-source dictation app for Mac, has been introduced through a Show HN post. The developer says it records speech and transcribes it using models running on the user’s Mac, then inserts cleaned-up text into the active app, offering an alternative to dictation services that process speech through the cloud.

The app requires macOS 14 or later, with Apple Silicon recommended for local model processing. Users place the cursor in a text field, hold Option + Space while speaking, and release the shortcut to transcribe. The developer says transcription starts after recording stops, rather than appearing word by word during speech. Users can change the shortcut or set it to toggle recording.

BetterWispr lists nine built-in models from Apple Speech, Parakeet and Whisper. Language coverage depends on the selected model: the product page says Parakeet supports English, Japanese and 25 European languages, while Whisper supports a broader range. Users can also connect certain external services or compatible APIs, but those options are separate from the app’s local-model setup.

Other listed features include selectable cleanup, per-app writing styles, voice commands and meeting notes. The default Light cleanup removes English filler words, repetitions and false starts; Medium can also edit for clarity using a selected notes model. The developer says local options such as Apple Intelligence or Ollama can produce meeting summaries, decisions and action items. These are product descriptions, not independent performance tests.

A Local Option for Mac Dictation

Dictation tools can send speech to remote services for recognition, which may raise privacy and data-handling concerns for people composing work messages, notes or other sensitive text. BetterWispr’s stated local-processing design gives Mac users another option: once a compatible model and its required files are installed, transcription can work offline without a cloud transcription fallback, according to the developer.

The app also combines dictation with text cleanup and insertion into other applications. That could reduce the steps between speaking and getting usable text into a document or message. The practical value will depend on recognition accuracy, speed and hardware, none of which the supplied material establishes through independent testing. The product page’s typing comparison is explicitly described as an illustration, not a benchmark.

Its Apache 2.0 license makes the source code available for inspection, modification and self-building. That gives technically inclined users a way to review how the app works, though the license alone does not verify every privacy claim or guarantee that the software will suit a particular workflow.

How the App Handles Speech

BetterWispr’s workflow is based on record, release and transcribe: users hold a shortcut as they speak, then wait for the transcript after recording ends. The product says text is inserted into the app where the cursor was placed. If automatic insertion is blocked, the transcript can be retrieved from the dashboard and copied manually; secure fields and some applications may prevent pasting.

The developer says dictation history stores both raw and corrected text on the Mac and can be turned off. Users can choose to keep every word, apply light cleanup, or select more extensive editing. The page limits styles, cleanup and voice commands to English, while speech recognition language support varies by model. The app also requires permissions for microphone access and, for automatic insertion, Accessibility access; Apple Speech may require Speech Recognition permission.

The launch material says BetterWispr is free, requires no account and has no dictation subscription or API key requirement for its built-in options. Models must be downloaded before use, taking storage on the Mac. The developer says updates are distributed from GitHub Releases and installed only when their EdDSA signature matches.

“Speech stays on your Mac.”

— BetterWispr, on its product website

Accuracy and Performance Remain Untested

The supplied launch material does not include independent tests of transcription accuracy, latency, battery use or model performance across different Mac hardware. It also gives no comparative results against other dictation apps. The website warns that model availability and performance depend on hardware, language and selected model.

The privacy description is the developer’s account; the supplied material does not include an external audit of data handling or source-code review. It also does not specify the size of every model download or the storage and processing demands users should expect. Whether particular languages, accents or specialist vocabulary are handled well will need to be tested by users with their own speech.

Some features have narrower limits than the headline description may suggest. Cleanup, styles and voice commands are English-only, and external API options may send dictated text to the selected provider. The source does not give details on pricing or data retention for those third-party services.

Users Can Test the Released App

Interested Mac users can check the project’s GitHub source and release information, install the app and select a compatible model. They will need to grant the relevant macOS permissions and download the model assets they plan to use. BetterWispr says its source is available under Apache 2.0 and invites users to report issues or build the app themselves.

The next useful evidence will come from real-world testing and any updates from the developer: how accurately the models handle different voices and languages, how quickly they produce text on supported Macs, and how reliably text reaches other apps. No independent review, test results or development roadmap is provided in the source material.

Key Questions

Is BetterWispr free?

Yes. The developer describes it as free and open source under the Apache 2.0 license, with no dictation subscription or account required for its built-in options. Downloaded model files use storage on the Mac.

Does BetterWispr send speech to the cloud?

The developer says built-in local models transcribe speech on the Mac, with no cloud transcription fallback, once the required model files are installed. If users choose an external API or provider, the handling of that data depends on the selected service.

Which Macs and operating systems does it support?

The app requires macOS 14 or later. Apple Silicon is recommended for local model processing, and performance depends on the Mac and model selected.

Does it show the transcript as I speak?

No. The product page says users hold a shortcut while recording and that transcription begins after recording stops. It does not describe live, word-by-word transcription.

Can users control how much the app edits?

Yes. The listed options include None, Light and Medium cleanup. None keeps every word; Light removes English filler words, repetitions and false starts; Medium can also edit for clarity using a selected notes model.

Source: hn