Summary

TypeWhisper/typewhisper-mac provides on-device, privacy-focused speech-to-text and AI text processing for macOS. It empowers users with local transcription and LLM-based text manipulation, offering a secure alternative to cloud-dependent services. The application targets macOS users seeking enhanced productivity through advanced, private AI capabilities.

How It Works

The project offers a flexible architecture with eight distinct transcription engines, including local options like WhisperKit and Apple SpeechAnalyzer, alongside cloud integrations (Groq, OpenAI). Users can choose between fully on-device processing for maximum privacy or leverage cloud APIs for faster results. Post-transcription, text can be refined using custom LLM prompts across various providers (Apple Intelligence, Groq, OpenAI, Gemini), enabling tasks like translation, summarization, and grammar correction. This hybrid approach prioritizes user data privacy while offering high performance and extensive customization.

Quick Start & Requirements

• Install: Via Homebrew (brew install --cask typewhisper/tap/typewhisper) or by downloading the latest DMG from GitHub Releases.
• Prerequisites: macOS 14.0 (Sonoma) or later. Apple Silicon (M1+) recommended. Minimum 8 GB RAM, 16 GB+ recommended for larger models. macOS 15+ required for some features; macOS 26+ for Apple Intelligence and Apple SpeechAnalyzer.
• Links: GitHub Releases page for DMGs.

Highlighted Details

• Diverse Transcription Engines: Supports WhisperKit (99+ languages, streaming, translation), Parakeet TDT v3, Apple SpeechAnalyzer, Qwen3 ASR, Voxtral, Groq, OpenAI, and OpenAI Compatible APIs.
• On-Device AI: Prioritizes local processing for privacy, with optional cloud acceleration.
• Advanced AI Processing: Enables custom LLM prompts for translation, summarization, grammar correction, and more, supporting multiple providers.
• System-Wide Dictation: Features global hotkeys, microphone selection, and auto-pasting into any application.
• Extensibility: Offers a plugin system, a local HTTP API for automation, and a CLI tool for shell integration.
• Personalization: Per-app and per-website profiles allow granular control over language, engine, prompts, and hotkeys.

Maintenance & Community

The provided README does not detail specific community channels (e.g., Discord, Slack), notable contributors, or a public roadmap. Update channels (stable, release-candidate, daily) are configurable within the application.

Licensing & Compatibility

• License: GPLv3.
• Compatibility: Commercial licensing is available separately. The GPLv3 license imposes copyleft requirements.

Limitations & Caveats

Certain advanced features, such as Apple Intelligence integration and Apple SpeechAnalyzer, are restricted to macOS 26+. Some features also require macOS 15+. Running multiple local transcription models concurrently can significantly increase memory consumption.