Changelog
What’s new
Every improvement, feature, and fix — documented release by release.
v1.0.0LatestMarch 2026
General Availability
- Full support for all 22 Arabic dialects with dedicated language models
- AI Auto Edits: filler words, repetitions, and pronunciation typos removed automatically
- Personal dictionary learns names, technical terms, and dialect expressions
- Voice snippets: create shortcuts for frequently used phrases
- Tone adaptation: professional in email, casual in chat, technical in code editors
- Real-time translation: speak in one language, output in another across 100+ languages
- Multi-device sync for dictionary and preferences
- Performance: under 700ms latency on Apple Silicon with Metal GPU acceleration
v0.9.0BetaJanuary 2026
Public Beta
- Added 15 new Arabic dialects (total: 22)
- Introduced code dictation mode for developers
- Custom dictionary with auto-learning from dictation sessions
- New hotkey configuration options
- Improved accuracy for mixed Arabic-English speech
- Fixed BiDi text rendering in VS Code and Terminal
- Reduced model download size by 40%
v0.8.0AlphaNovember 2025
Private Alpha
- Initial release for macOS with Apple Silicon optimization
- Core dictation engine powered by whisper.cpp
- 7 Arabic dialects: Egyptian, Gulf, Levantine, Moroccan, Iraqi, Sudanese, Tunisian
- Basic filler word removal
- Global hotkey: hold-to-speak in any app
- 100% local processing — zero network calls
- Privacy-first architecture: no telemetry, no analytics, no cloud
v0.7.0InternalSeptember 2025
Internal Testing
- Proof of concept with whisper.cpp integration
- Metal GPU acceleration for Apple Silicon
- Basic Arabic MSA transcription
- Hotkey-based activation system
- Latency benchmark: 680ms average on M1 MacBook Pro