voxd 0.1.2

Voice dictation daemon for Linux with Sarvam AI STT

voxd

Voice dictation daemon for Linux. Press keybind, talk, press again. Text goes to clipboard.

Install

pip install voxd

Dependencies:

# X11 (dwm, i3, bspwm, etc)
sudo pacman -S ffmpeg xclip

# Wayland (hyprland, sway, etc)
sudo pacman -S ffmpeg wl-clipboard

Setup

Set your Sarvam AI API key:

voxd config set api_key YOUR_KEY

Optional - set language:

voxd config set language hi-IN  # Hindi
voxd config set language en-IN  # English (default)

Start daemon

Add to your startup:

dwm/i3/bspwm - ~/.xinitrc:

voxd-daemon &

hyprland - ~/.config/hypr/hyprland.conf:

exec-once = voxd-daemon

Keybind

dwm - config.h:

{ MODKEY, XK_semicolon, spawn, SHCMD("voxd toggle") },

hyprland - hyprland.conf:

bind = SUPER, semicolon, exec, voxd toggle

Usage

Press keybind → talk → press keybind → paste (Ctrl+V)

Terminal:

voxd toggle    # Start/stop recording
voxd status    # Check if recording
voxd quit      # Kill daemon

Config

voxd config list              # Show all settings
voxd config set key value     # Change setting
voxd config get key           # Get value

Settings:

• api_key - Sarvam AI key (required)
• language - Language code (default: en-IN)
• model - STT model (default: saaras:v3)

Config stored at ~/.config/voxd/config.json

Advanced

WM-specific commands (if auto-detect fails):

voxd-dwm toggle     # Force X11 mode
voxd-hypr toggle    # Force Wayland mode

Troubleshooting

Daemon not running:

ps aux | grep voxd
cat /run/user/$UID/voxd/daemon.log

No clipboard:

# Install clipboard tool
sudo pacman -S xclip          # X11
sudo pacman -S wl-clipboard   # Wayland

Wrong language:

voxd config set language hi-IN
voxd quit && voxd-daemon &