typewhisper-mac
Open-source speech-to-text for macOS with local models, optional cloud APIs, system-wide dictation and on-device AI text processing. GPL-3.0 licensed.
TypeWhisper is a macOS dictation and transcription app for people who want to keep voice recordings on their own machine and use the resulting text across apps. Local engines keep audio on your Mac; optional cloud providers such as Groq, OpenAI and xAI/Grok process audio remotely.
System-wide dictation inserts speech into the active app, while WhisperKit can show a transcript as you speak. For existing recordings, the app batch-transcribes audio and video files and exports timestamped SRT or WebVTT subtitles. Searchable history lets you edit and export past transcripts, and dictionary corrections fix recurring recognition mistakes.
Its reusable AI workflows can rewrite dictated text, extract information or apply formatting based on the app or website you're using. They can also process selected text or clipboard contents. Apple Translate handles translation on-device, and the Local LLM plugin uses MLX on Apple Silicon for prompt processing with Gemma 4 E2B and E4B models. Cloud LLMs and OpenAI-compatible providers are optional alternatives.
Plugins connect workflows to Obsidian, Linear and MCP tools. A local HTTP API and command-line tool let other software use its transcription capabilities.
The project is open source under GPL-3.0. It runs natively on Intel and Apple Silicon Macs with macOS 14 or later and at least 8 GB RAM; Apple Silicon is recommended, with 16 GB or more for larger models. Apple Intelligence and SpeechAnalyzer require macOS 26 or later.
