v0.2.02026-09-19
Sturdier hotkeys and wake word, more cloud models, real-time sync
- Hotkeys keep working after a screen lock, an RDP reconnect, or a stuck modifier key, and each hotkey can now be turned off on its own
- Pressing Escape cancels dictation without the keystroke leaking into the app underneath
- Wake word: say a stop word to end a session, and it no longer goes deaf after a dead microphone stream or a rejected command
- Idle CPU use while listening for the wake word is a fraction of what it was
- New cloud speech models: Grande Pro, Grande XPro, and Venti Pro, with filler words removed from transcripts
- Cloud transcription now receives your workspace context, vocabulary, and language hints, and rides out flaky connections
- Vocabulary entries are limited to 50 characters with an inline hint, and the phrase-limit notice follows the model you selected
- Settings and workspaces sync to your other devices the moment they change instead of on a timer
- The continuous-dictation balloon stays where you put it, and each app can have its own text-injection method
- The tray microphone list refreshes when you plug a device in, and app-management voice commands work on Windows
- Mid-recording capture stalls are detected and reported instead of silently producing gaps
- Installers are roughly half the size
v0.1.02026-07-30
First public release
- Press a hotkey, speak, and your words land in whatever app the cursor is in — no copy and paste
- Transcription runs on your machine by default, with GPU acceleration on Windows and Apple Silicon
- Optional cloud transcription and refinement when you want more accuracy or speed
- Hands-free wake word so you can start dictating without touching the keyboard
- Custom commands turn a spoken phrase into an action, a snippet, or a rewrite
- Your own vocabulary for the names, jargon, and acronyms generic models get wrong
- Tone presets and custom prompts to rewrite your words as you speak
- Your clipboard is always restored exactly as you left it
- Available in English, French, and Vietnamese
- Native builds for macOS (Apple Silicon), Windows, and Linux, with automatic updates
v0.0.52026-03-15
Cross-device Sync & BYOK Providers
- Added cross-device sync for workspaces and settings
- BYOK provider support: connect your own OpenAI, Anthropic, or Google API keys
- New provider management UI with key validation
- Sync conflict resolution with last-write-wins strategy
- Bug fix: workspace matching engine handles empty context gracefully
v0.0.42026-02-01
Workspaces & Context Awareness
- Workspace system for per-app and per-project context
- Automatic workspace detection based on active window
- Custom vocabulary per workspace
- Tag-based workspace organization with conflict validation
- Improved LLM refinement prompts with workspace context injection
v0.0.32026-01-10
LLM Refinement Engine
- Local LLM refinement via LlamaSharp integration
- Cloud LLM support through Semantic Kernel
- Tone presets: Professional, Casual, Technical, Creative
- Refinement pipeline with configurable processing stages
- Response streaming for lower perceived latency
v0.0.22025-11-20
Multi-platform Audio & Overlay
- Cross-platform audio capture via PortAudioSharp2
- Listening overlay with real-time waveform visualization
- Silence detection and automatic recording stop
- macOS CoreML acceleration for Whisper inference
- Windows CUDA support for faster transcription
v0.0.12025-10-01
Acoustic Core
- System tray application with global hotkey activation
- Whisper-based speech-to-text with multiple model sizes
- Text injection into any active application via SharpHook
- Basic settings UI with model selection
- macOS and Windows platform support