GitHub - zachlatta/freeflow: Free & fast alternative to Wispr Flow

Open original

FreeFlow is a free, open source dictation app for Mac that uses AI transcription and context-aware cleanup to turn speech into text, replacing paid alternatives like Whisper Flow and Superwhisper. You bring your own API key (Groq or compatible), so there's no subscription or data retention on FreeFlow's servers.

  • Hold Fn to dictate or use customizable keyboard shortcuts; Edit Mode lets you highlight text and transform it with voice commands like "make this shorter"
  • Context-aware cleanup reads your current app to spell names and terms correctly, with support for custom vocabulary and simpler post-processing if you prefer
  • Works with any OpenAI-compatible provider (Groq, Ollama, LM Studio) and supports local models with configurable timeouts for slower hardware

FreeFlow is a free Mac dictation app that copies the core features of paid alternatives like Whisper Flow and Superwhisper. You hold or toggle a keyboard shortcut to dictate, and the app transcribes your speech using Groq's API (which you need to set up yourself with a free key). It then cleans up the raw transcription using an AI model, fixing grammar, removing filler words, and correcting spelling based on context from whatever app you're typing into. You can add custom vocabulary for names and jargon you use frequently, and you can edit existing text by selecting it and giving spoken instructions like "make this shorter" or "turn this into bullets." The app has no server component, so nothing gets stored on FreeFlow's infrastructure—only your audio and transcripts go to whatever API provider you configure (Groq by default, but you can swap in OpenAI-compatible alternatives or run local models like Ollama). The cleanup process is customizable; if you want less aggressive processing, you can paste in a simpler prompt that just fixes obvious errors without inferring meaning. For local models that run slower, you can extend the default 20-second timeout up to 120 seconds or longer through macOS terminal commands, with separate timeout settings for transcription, cleanup, and context requests. The project is maintained by Marc Bodea and released under the MIT license, so you can inspect the code and modify it yourself. It requires either a Groq API key or access to another transcription and LLM provider, which means there's a setup step before you can start using it. The shortcut system is flexible—you can customize both the hold-to-talk and toggle modes, and if your toggle shortcut extends your hold shortcut, you can start holding and then press extra modifier keys to switch into latched mode without stopping the recording.

Ask about this article