Version2.0.3
Revision16
Size46.5 MB
Licenseunset
Confinementstrict
Basecore24

Privacy-first AI voice dictation for Linux

ScreenshotScreenshotScreenshot

OpenDictate is a native GTK4 and Libadwaita voice dictation application for Linux written in 100% pure Rust. It turns spoken voice into clean, polished text across both Wayland and X11 desktops using either local on-device Whisper models or cloud AI providers.

Key Features:
  • Dual AI Engine: Transcribe 100% offline with local whisper.cpp models (Tiny, Base, Small, Medium) or use low-latency Cloud AI (Groq, OpenAI, Google Gemini, Alibaba Qwen, DeepSeek).
  • Floating Live Waveform Minibar: Compact overlay bar with real-time audio waveform, word counter, tone selector (Clean, Professional, Casual, Concise, Raw), and instant clipboard copy.
  • Hardware-Aware Audio Capture: Exclusive single-device capture with automatic detection of Bluetooth Handsfree headsets, wired headphones, and system microphones via PipeWire, PulseAudio, and ALSA.
  • 28-Language UI Localization: Instant in-place interface translation across 28 desktop Linux languages.
  • Persistent Dictation History: Local SQLite history (~/.local/share/opendictate/history.db) with one-click copy and deletion inside the header menu popover.
  • Privacy-Safe Crash Diagnostics: Opt-in, user-editable diagnostic report viewer with automatic path and API key redaction.

Update History

2.0.2 (15) → 2.0.3 (16)
20 Oct 2026, 19:00 UTC
2.0.1 (14) → 2.0.2 (15)
13 Oct 2026, 07:16 UTC
2.0.0 (13) → 2.0.1 (14)
12 Oct 2026, 19:31 UTC
2.0.0 12 → 13
9 Oct 2026, 18:31 UTC
2.0.0 11 → 12
8 Oct 2026, 12:01 UTC
2.0.0 10 → 11
8 Oct 2026, 11:16 UTC
2.0.0 9 → 10
8 Oct 2026, 10:46 UTC
2.0.0 8 → 9
8 Oct 2026, 10:01 UTC
2.0.0 7 → 8
8 Oct 2026, 09:16 UTC
2.0.0 4 → 7
8 Oct 2026, 09:01 UTC
2.0.0 2 → 4
8 Oct 2026, 06:31 UTC
2.0.0 (2)
7 Oct 2026, 18:16 UTC

Published7 Oct 2026, 18:14 UTC

Last updated20 Oct 2026, 18:58 UTC

First seen7 Oct 2026, 18:16 UTC