TypeWhisper/typewhisper-mac

Local speech-to-text for macOS on-device AI, fully private, optional cloud

TypeWhisper for Mac – What It Is

TypeWhisper is a native macOS application that lets you dictate speech, have it transcribed, and then run AI‑powered text transformations – all from the menu bar. It supports both on‑device speech models (e.g., WhisperKit, Parakeet, Apple SpeechAnalyzer) and cloud APIs (OpenAI, Groq, xAI/Grok, etc.). The transcribed text can be sent through configurable workflows that invoke LLMs for translation, rewriting, extraction, or any custom prompt you define. Everything is wrapped in a polished UI with hot‑key support, iCloud sync, a plugin system, and a CLI/HTTP API for automation.

Core Capabilities

Area What You Get
Transcription • Choose from dozens of local or cloud ASR engines (WhisperKit, Parakeet, Apple SpeechAnalyzer, Qwen3, Voxtral, etc.)
• Real‑time streaming preview while you speak
• Drag‑and‑drop batch transcription of audio/video files, Finder integration, and subtitle export (SRT/WebVTT)
• Import custom speech models from Hugging Face or a local folder
Dictation • System‑wide push‑to‑talk hotkey (single‑modifier or custom)
• Automatic paste into any app, with undo and “restore raw transcript” actions
• Recovery of the last three successful dictations for up to 24 h, plus cancellation options (double‑Esc, single‑Esc, immediate)
AI Processing (Workflows) • Build reusable pipelines that can translate, rewrite, extract data, format text, or trigger app‑specific automation
• Select LLM providers per workflow (Apple Intelligence, Groq, OpenAI, xAI/Grok, Gemini, local MLX models) with fallback ordering

• Local on‑device LLM support via the MLX plugin (Gemma 4, Qwen3.5, LFM2.5) for privacy‑first prompt processing | | Personalisation | • Per‑app / per‑website workflow triggers, hotkey‑driven or manual palette • Dictionary with term packs and auto‑learning of corrections • Snippets with placeholders ({{DATE}}, {{CLIPBOARD}}) • Searchable History with device grouping, bulk export, and inline editing | | Premium Add‑ons | • Calendar‑linked meeting automation (countdowns, structured minutes) • Cloud‑folder sync for dictionaries/snippets (iCloud Drive, Dropbox, OneDrive, Syncthing) • iCloud sync of History/Inbox across Macs | | Extensibility | • Plugin SDK for custom LLM providers, transcription engines, TTS, post‑processors, memory back‑ends, and action plugins • Local REST API (default port 8978) with bearer‑token auth for scripts • CLI (typewhisper export/import) for backup/restore | | General UI/UX | • Visual keyboard shortcut editor that respects the current input source • Multiple indicator styles (Notch, Overlay, Minimal) with live preview • Widgets for stats, last transcription, and activity chart • Multilingual UI (English, German, Japanese, Simplified Chinese) |

Typical Use Cases

  • Hands‑free writing: Dictate into any document, email, or code editor with a single hotkey, then let a workflow automatically clean up formatting or translate the text.
  • Meeting minutes: Record a Zoom/Teams call, have the audio transcribed locally, and let a premium workflow generate a structured summary and export it to your notes app.
  • Accessibility: Users who need on‑device speech‑to‑text can run WhisperKit or Parakeet without sending audio to the cloud.
  • Automation pipelines: Combine the local HTTP API with scripts (e.g., Raycast, custom shell tools) to trigger transcription from other apps or CI jobs.
  • Developers & power users: Extend the app with custom plugins or use the CLI to batch‑process large media libraries.

Getting Started

  1. Install – brew install --cask typewhisper/tap/typewhisper or download the DMG from the GitHub releases page.
  2. Grant permissions – Microphone + Accessibility (System Settings → Privacy & Security).
  3. Pick a transcription engine – Open Settings → Engine, download a local model if desired, or configure a cloud API key.
  4. Set a global hotkey – Use the visual shortcut editor to assign a single‑modifier key.
  5. Dictate – Press the hotkey, speak, and the text appears where your cursor is. Use the workflow palette or hotkeys to run post‑processing.

License & Community

  • License: GNU GPL v3 (as shown by the badge).
  • Source: The repo contains the full macOS Swift project, plugin SDK, and documentation (release notes, support matrix, build instructions).
  • Contribution: Development occurs on the main branch (targeting 1.8). Daily builds are published as v1.8.0‑daily.*. Issues, feature requests, and release‑checklist templates are provided in the repository.

TL;DR

TypeWhisper is a full‑featured, privacy‑focused macOS dictation and speech‑to‑text tool that couples on‑device or cloud transcription with customizable AI workflows, plugin extensibility, and system‑wide hotkey control. It’s suitable for anyone who wants a powerful, scriptable voice‑typing experience on a Mac.

Related

  • Project
    mylxsw/typeflux

    Typeflux is a free, open‑source macOS app that lets you dictate or talk to an AI assistant using the `Fn` key. It injects transcribed text directly into the active application, supports both cloud and on‑device speech models, and includes features like custom personas, streaming previews, and session history.

  • Project
    cyx2333hhh/talk-type

    A macOS voice input tool that combines real-time speech recognition, local Whisper fallback, and AI-driven text cleanup for seamless bilingual input across applications.

  • Project
    tover0314-w/opentypeless

    An open-source AI voice input tool for macOS, Windows, and Linux that provides app-aware dictation, voice Q&A, and text rewriting.

  • Project
    tornikegomareli/Talkify

    A private, on-device voice dictation and transcription tool for macOS that uses local Apple frameworks and LLMs to provide speech-to-text, translation, and text refinement.

  • Project
    opennookorg/betterwispr

    BetterWispr is an open‑source macOS dictation tool that transcribes speech offline using Whisper, Parakeet, or Apple’s on‑device recognizer. It lets you dictate into any app via a shortcut, supports custom vocabularies, learns from corrections, offers voice commands, and can integrate external speech‑to‑text APIs while keeping data local.