Whisper Desktop v2.0.0
Whisper Desktop v2.0.0
A ground-up redesign plus AI translation!
π€ AI Translation β The headline feature
Translate your transcriptions through OpenAI, Anthropic, Gemini, or any OpenAI-compatible provider. Manage multiple providers with keyring-backed API keys, configure per-provider models with reasoning levels (None/Low/Medium/High), and let the system auto-detect context windows and recover from token-limit errors automatically. Original subtitle formatting (SRT, VTT, LRC, TXT) is preserved throughout. Preview lines before translating, and cancel mid-request with immediate HTTP abort.
π Other Major Features
Premium UI Overhaul
- Complete glassmorphic redesign with royal-blue gradient accents
- Collapsible sidebar with staggered stagger animations
- 3-step wizard for transcription flow with auto-convert
- Custom-styled dropdowns, toggles, radio buttons, scrollbars throughout
- Responsive layout stacking for small viewports
- Disabled context menu on non-input elements
Voice Activity Detection (VAD)
- Built-in Silero VAD model download and auto-usage
- Configurable threshold, speech tolerance, minimum segment duration
- Reduces processing time on silent sections
Model Hub Enhancements
- Integrated model downloader with pause/resume/delete
- Real-time speed, ETA, and progress tracking
- Search and category filtering
- Spec-guided recommendations based on GPU/hardware
Batch Processing
- Batch queue with drag reordering and sort controls
- Sequential per-file progress with detailed status
- Error dialogs for per-file retry/skip/abort
π οΈ Engineering & Hardening
- 35+ correctness fixes across Rust backend and JS frontend
- Replaced all
.unwrap()with.expect()on regex/path invariants - XSS hardening β DOM builder pattern replacing
innerHTMLwithtextContent - Tauri event listener lifecycle management with scoped cleanup
- TDZ fixes for timer/handler/state declarations
- IPC rate limiter removed (caused log line drops during FFmpeg bursts)
- Sequential batch probing (fixed parallel race condition)
- Error notifications for settings load/save failures
- Strip ANSI codes from log output
- Null-safety guard for non-string inputs
π¦ Downloads
Assets below β .deb, .rpm, and .AppImage for Linux.
Enjoy local, private speech-to-text transcription!
Full Changelog: v1.0.1...v2.0.0