BetterWispr 0.0 Release – Free Open‑Source Local Dictation for macOS

BetterWispr 0.0 delivers offline, local‑only dictation for macOS

BetterWispr 0.0 is a free, Apache‑2.0‑licensed macOS app that transcribes speech entirely on the user’s machine. It requires macOS 14+ (Apple Silicon recommended) and does not need an account, cloud API key, or internet connection after the initial model download.

Core capabilities are ready‑to‑use out of the box

  • Hold‑to‑talk – Press ⌥ Space (or a custom shortcut) while speaking; release to insert cleaned text.
  • Nine built‑in models – Apple on‑device speech, Parakeet (English, Japanese, 25 European languages), and OpenAI Whisper for broader multilingual support. Users can also configure Sarvam AI, Smallest AI, or any OpenAI‑compatible endpoint.
  • Automatic cleanup – Light mode removes English filler words ("um", repeats, false starts); Medium adds AI‑driven clarity edits; None preserves the raw transcript.
  • Per‑app styles – Each app can have a tone (casual, very casual, formal) that controls punctuation and capitalization.
  • Meeting‑notes workflow – The Notetaker records the user as Me and other apps as Them, then uses Apple Intelligence or a local Ollama model to generate summaries, decisions, and action items.
  • Vocabulary learning – Adding names or custom terms in Settings or fixing a word in History teaches the engine to spell them correctly in future dictations.
  • History & insights – Raw and cleaned transcripts are stored locally; the dashboard shows word count, speaking speed, cleanup savings, and daily streaks.
  • Secure updates – Releases are delivered via Sparkle with EdDSA signatures, ensuring tamper‑free upgrades.

Why local processing matters

All transcription runs on the Mac, eliminating round‑trips to external servers. This design protects user privacy, reduces latency, and works without an internet connection once a model is installed. The project’s GitHub repo (https://github.com/opennookorg/betterwispr) provides full source code and a make dev build script for contributors.

Community reaction highlights

  • Comparison with existing tools – Users repeatedly asked how BetterWispr differs from Handy, VoiceInk, Wispr Flow, and other OSS projects. The consensus is that BetterWispr’s unique mix of local models, built‑in cleanup levels, and per‑app styling sets it apart, though many see it as part of a crowded space.

    "This reminds me very much in terms of features as https://github.com/cjpais/Handy … what's the use case for this over that?" – Zizizizz

  • Model choice – Several commenters noted that Whisper (2022) remains the de‑facto baseline for multilingual STT, praising its continued relevance.

    "What’s really shocking is this and many of the competitors are all using OpenAI Whisper – which came out in 2022. Kudos to the early OpenAI team…" – Royce‑CMR

  • Local‑first alternatives – Google’s AI‑Edge Eloquent and the open‑source Biscotti project were cited as other offline‑only options, underscoring a growing demand for privacy‑preserving dictation.
  • User experience – Some users reported flawless early testing with Parakeet v3, while others expressed fatigue with voice input compared to typing.

    "I’ve been trying to figure out how to code and build while on the bike … Want a remote terminal into my Mac that I can voice control." – honkycat

  • Future feature wishes – Wake‑word activation, dual‑button commands for LLM‑driven edits, and deeper meeting‑recording diarization were popular suggestions.

    "What about adding a wake word detection so I can just say ‘Hey wakeWord’ to activate the STT?" – handfuloflight

  • Performance on Apple Silicon – Users confirmed that the small Whisper model (~200 MB) runs with negligible GPU load on M‑series chips, making the app viable for everyday use.

    "I thought the small model won't be any good, but it's actually great and barely causes the GPU any usage." – starik36

How to get started

  1. Install – Download the app from https://betterwispr.com/ and follow the welcome guide to grant microphone, accessibility, and speech‑recognition permissions.
  2. Choose a model – In Settings, download a local model (Parakeet v3 is recommended for English) and set your preferred cleanup level.
  3. Dictate – Place the cursor in any text field, hold ⌥ Space, speak, and release. The cleaned text appears instantly; if automatic paste fails, the transcript is copied to the clipboard.
  4. Fine‑tune – Add custom vocabulary or adjust per‑app styles to match your workflow.

Outlook

BetterWispr demonstrates that a fully offline dictation pipeline is feasible on modern Macs, combining high‑quality speech models with user‑controlled cleanup and styling. The vibrant HN discussion shows strong community interest in expanding local speech capabilities—especially wake‑word activation, diarization for meetings, and tighter integration with local LLMs. As Apple continues to improve on‑device speech assets (macOS 27+), projects like BetterWispr are likely to become even more capable and mainstream.

Sources

Related

  • Project
    TypeWhisper/typewhisper-mac

    A system-wide speech-to-text and AI text processing utility for macOS that supports local and cloud transcription engines with customizable AI workflows.

  • Project
    OpenWhispr/openwhispr

    A privacy-first, open-source voice-to-text application providing local AI dictation, meeting transcription with speaker diarization, and AI-powered notes.

  • Project
    Beingpax/VoiceInk

    VoiceInk is a macOS‑only, offline speech‑to‑text app that uses local Whisper‑based models (via whisper.cpp and transcribe‑cpp‑swift) to provide near‑instant, privacy‑preserving transcription, context‑aware modes, custom dictionaries, global shortcuts, and a built‑in AI assistant. It’s open‑source under GPL‑v3, installable via Homebrew or from source, with optional paid licensing for updates and support.

  • Project
    ideaplexa/voicetypr

    Voicetypr is an open‑source, offline‑first dictation app for macOS and Windows. It captures speech via a global hot‑key, transcribes locally with Whisper (and Parakeet on Apple Silicon) or optionally via cloud STT services, can clean up the text with LLMs, and inserts the result at the current cursor. A companion CLI lets scripts or AI agents use the same transcription pipeline, returning plain text or JSON. The app is built with Tauri (Rust backend) and React, offers LAN‑based private transcription servers, and is released under AGPL‑3.0.

  • Project
    watzon/pindrop

    Pindrop is a discontinued, native‑macOS menu‑bar dictation app that transcribes speech offline using on‑device models (Whisper‑Kit, Parakeet, SenseVoice, Apple Speech) or optional OpenAI cloud models. It offers real‑time streaming, multilingual support, speaker diarization, a searchable transcript library, AI‑enhanced cleanup, and privacy‑first defaults. Built in Swift/SwiftUI, it runs on macOS 14+ (Apple Silicon recommended) and is replaced by the Sayso app.