OpenWhispr
Open-source voice-to-text dictation with local and cloud processing
Overview
OpenWhispr is an open-source, privacy-first voice-to-text dictation tool that works across any application on macOS, Windows, and Linux. Press a hotkey, speak naturally, and transcribed text appears at your cursor. It supports both local processing via Whisper and NVIDIA Parakeet for fully offline use, and cloud engines including OpenAI, Anthropic Claude, Google Gemini, and Groq. Features include an AI agent mode for text refinement, support for 58+ languages with auto-detection, custom dictionary for domain-specific terms, and a transcription history stored locally in SQLite. Licensed under MIT.
Pricing: Free (2,000 words/week) | Pro: $8/month or $80/year
Minimum macOS: 10.15 (Catalina)
Architecture: Apple Silicon, Intel
Key Features
- Local and cloud speech-to-text with Whisper and NVIDIA Parakeet engines
- System-wide hotkey activation for push-to-talk dictation
- AI agent mode for intelligent text refinement after transcription
- 58+ languages supported with automatic language detection
- Custom dictionary for domain-specific terms and proper nouns
- Fully offline capability with local model processing
- Multi-provider AI: OpenAI, Anthropic Claude, Google Gemini, Groq
- Transcription history stored locally in SQLite database
- Cross-platform: macOS, Windows, and Linux
- Open-source under MIT license
Tags
Similar Apps
FluidVoice
Speak naturally, write instantly, stay private
SpeakMac
Type with voice anywhere on your Mac - private, offline, lightning fast
VoiceInk
Transform speech into text instantly with advanced AI voice recognition
BetterDictation
You speak, we type - offline Whisper-powered dictation for Mac