Changelog

We're thrilled to announce the public beta of SpeakToCode! After months of development and testing, voice-to-text for developers is here.

New Features

• Voice-to-text transcription with OpenAI Whisper running 100% locally
• 7 Whisper model sizes, from Tiny (75 MB) to Large (2.9 GB)
• Global hotkeys for push-to-talk and toggle modes
• Auto-paste to any active window
• Multi-monitor overlay to see recording status across all screens
• Transcription history with search and statistics
• Cloud sync for optional backup of transcriptions and settings
• Dark and light themes

Technical Details

• Built with Tauri v2 for native Windows performance
• whisper.cpp for optimized local inference
• Supabase for authentication and cloud sync
• 15-day free trial, no credit card required

This release adds AI-powered text cleanup and cross-device settings sync.

New Features

• AI text cleanup: Use Groq API to clean up and format your dictations
• Settings sync: Your preferences now sync across devices via cloud
• Improved overlay: Customizable position and display duration
• Better hotkey support: More key combinations available

Bug Fixes

• Fixed overlay flickering on some multi-monitor setups
• Fixed clipboard paste delay on certain applications
• Improved Whisper model download reliability

Performance

• 15% faster transcription with optimized whisper.cpp build
• Reduced memory usage for Tiny and Base models