v3.3.0 — Linux builds, hardened transcription, Wayland shortcuts

SPEAK_EXECUTE

Voice-controlled terminal for developers. Speak your commands naturally and watch them run. Multi-tab terminals, split panes, and your choice of AI provider.

WINDOWS macOS (Apple Silicon)
macOS (Intel) Linux (AppImage) VIEW SOURCE
zsh Terminal 2 +
AUDIOBASH
~/Projects ❯
[VOICE] "list all files in this directory"
→ ls -la
drwxr-xr-x 8 user staff 256 Jan 2 10:00 src
drwxr-xr-x 42 user staff 1344 Jan 2 09:45 node_modules
-rw-r--r-- 1 user staff 1842 Jan 2 10:15 package.json
-rw-r--r-- 1 user staff 512 Jan 2 10:15 tsconfig.json
READY
Gemini 2.0 Flash
Option+S: Voice

NEW IN v3.3.0

Linux builds, hardened transcription, and Wayland shortcuts

///

Linux downloads

AppImage and .deb builds now ship alongside Windows and macOS. One voice terminal, every desktop.

///

Hardened transcription

Batch cloud transcription requests now run in the main process instead of the renderer.

///

Wayland shortcuts

Global hotkeys register under Wayland, not just X11, so the Alt bindings work on modern Linux desktops.

SYSTEM_CAPABILITIES

Everything you need to control your terminal with voice

VOICE_RECOGNITION

Multiple AI providers: Gemini 2.0/2.5 Flash, ElevenLabs Scribe (real-time or batch), or fully offline with bundled local Whisper. Parakeet is also supported if you run your own NVIDIA GPU server. Choose your engine.

SPLIT_VIEW

Up to 4 terminals visible at once. 5 layout modes: single, horizontal, vertical, 2x2 grid, 1+2. Resizable panes.

QUICK_NAV

Jump to recent or favorite directories with one click. Star your project folders for instant access.

CONTEXT_AWARE

Agent mode understands your environment: current directory, recent output, errors. Smarter commands.

CLAUDE_CODE

Detects Claude Code /voice mode and hands off cleanly: shows a [CC /voice] badge and pauses the built-in mic so the two never fight.

PREVIEW_PANE

Embedded web preview for localhost dev servers. Auto-refresh on file changes. Screenshot capture.

OPERATION_SEQUENCE

Three steps to voice-powered productivity

01

ACTIVATE

Press your configured hotkey (Alt+S on Windows, Option+S on Mac) to start voice capture. The floating overlay appears.

02

SPEAK

Say your command naturally. "List all JavaScript files" or "Git status". The AI transcribes and converts.

03

EXECUTE

The command runs in your active terminal and the results come back in place. Hands stay on the keyboard.

INITIALIZE_NOW

Free and open source. Windows, macOS, and Linux ready.

WINDOWS

Windows 10/11

DOWNLOAD .EXE

macOS

macOS 12+

APPLE SILICON INTEL

LINUX

Linux x64

APPIMAGE .DEB
STAR ON GITHUB USER MANUAL

Windows Installation

  1. Download the .exe installer
  2. Run the installer (may need to click "More info" → "Run anyway")
  3. Launch AudioBash from Start Menu
  4. Add your API key in Settings (gear icon)

macOS Installation

  1. Download the .dmg for your Mac type
  2. Open DMG, drag AudioBash to Applications
  3. Right-click → Open (first time only, bypasses Gatekeeper)
  4. Grant microphone permission when prompted