### Install and Run Voxtype

Source: https://voxtype.io/docs/USER_MANUAL

Initial setup involves joining the 'input' group, running the setup wizard to download models and verify permissions, and then starting the Voxtype daemon.

```bash
groups | grep input
```

```bash
voxtype setup --download
```

```bash
voxtype
```

--------------------------------

### Verify Installation

Source: https://voxtype.io/docs

Run the built-in setup check to validate system permissions and dependencies.

```bash
voxtype setup
```

--------------------------------

### Quick Install Voxtype from Source (Arch Linux)

Source: https://voxtype.io/docs

Installs dependencies, builds Voxtype from source, and performs initial setup on Arch Linux. Requires logging out and back in after adding user to the 'input' group.

```bash
# Install dependencies, build, and setup (Arch)
sudo pacman -S --needed base-devel rust clang alsa-lib wtype wl-clipboard && \
git clone https://github.com/peteonrails/voxtype && cd voxtype && \
cargo build --release && \
sudo cp target/release/voxtype /usr/local/bin/ && \
sudo usermod -aG input $USER && \
echo "Log out and back in, then run: voxtype setup --download"
```

--------------------------------

### Install Typing Backends

Source: https://voxtype.io/docs

Install wtype or ydotool based on the display environment.

```bash
# Fedora:
sudo dnf install wtype
# Arch:
sudo pacman -S wtype
# Ubuntu:
sudo apt install wtype
```

```bash
# Fedora:
sudo dnf install ydotool
# Arch:
sudo pacman -S ydotool
# Ubuntu:
sudo apt install ydotool

# Enable and start the daemon (Arch)
systemctl --user enable --now ydotool
```

```bash
# Install dotool (check your distribution's package manager or AUR)
# Add user to input group for uinput access
sudo usermod -aG input $USER
# Log out and back in
```

```bash
# Install ydotool (see commands above for your distro)
# Then enable and start the daemon (required!)
systemctl --user enable --now ydotool  # Arch
# For Fedora, see Troubleshooting guide for system service setup
```

--------------------------------

### Download Models

Source: https://voxtype.io/docs/MODEL_SELECTION_GUIDE

Use the `voxtype setup model` command to browse, install, and manage speech recognition models from HuggingFace.

```bash
voxtype setup model
```

--------------------------------

### Install Recommended Display Server Packages

Source: https://voxtype.io/docs

Install necessary utilities for Wayland or X11 support.

```bash
# For Wayland:
sudo dnf install wtype wl-clipboard libnotify
# For X11 or as fallback:
sudo dnf install ydotool
```

--------------------------------

### Voxtype Basic Usage Example

Source: https://voxtype.io/docs/USER_MANUAL

Demonstrates the typical workflow of starting the Voxtype daemon, holding a hotkey to speak, and observing the transcribed text output.

```bash
voxtype
[INFO] Voxtype v0.1.0 starting...
[INFO] Using model: base.en
[INFO] Hotkey: SCROLLLOCK
[INFO] Output mode: type (fallback: clipboard)
[INFO] Ready! Hold SCROLLLOCK to record.

# User holds ScrollLock and says "Hello world"
[INFO] Recording started...
[INFO] Recording stopped (1.2s)
[INFO] Transcribing...
[INFO] Transcribed: "Hello world"
[INFO] Typed 11 characters
```

--------------------------------

### Install Compiled Binaries and Configs

Source: https://voxtype.io/docs

Manually install the built binary, configuration files, and systemd service.

```bash
# Install binary
sudo install -Dm755 target/release/voxtype /usr/local/bin/voxtype

# Install config (optional - will be created on first run)
sudo install -Dm644 config/default.toml /etc/voxtype/config.toml

# Install systemd service (optional)
install -Dm644 packaging/systemd/voxtype.service \
    ~/.config/systemd/user/voxtype.service
```

--------------------------------

### Install Build Dependencies by Distribution

Source: https://voxtype.io/docs

Install required development tools and libraries based on the OS.

```bash
sudo pacman -S base-devel clang alsa-lib
```

```bash
sudo apt install build-essential libclang-dev libasound2-dev pkg-config
```

```bash
sudo dnf install @development-tools clang-devel alsa-lib-devel
```

--------------------------------

### Install Wayland Input Support

Source: https://voxtype.io/docs

Install wtype to enable text injection on Wayland compositors.

```bash
# Install wtype
sudo pacman -S wtype  # or apt/dnf
```

--------------------------------

### Install eitype

Source: https://voxtype.io/docs/USER_MANUAL

Installs the eitype tool via cargo, recommended for GNOME and KDE environments.

```bash
cargo install eitype
```

--------------------------------

### Start Voxtype Recording with a Profile

Source: https://voxtype.io/docs/USER_MANUAL

Specify a profile when starting a recording using the `voxtype record start` command to apply custom settings.

```bash
# Use the slack profile for this recording
voxtype record start --profile slack
voxtype record stop

# Or with toggle mode
voxtype record toggle --profile code

```

--------------------------------

### Configure Speed vs Accuracy Example

Source: https://voxtype.io/docs/USER_MANUAL

Example configuration for switching between a fast model and a high-accuracy model.

```toml
[hotkey]
key = "SCROLLLOCK"
model_modifier = "LEFTSHIFT"

[whisper]
model = "tiny.en"                   # Lightning fast for quick notes
secondary_model = "large-v3-turbo"  # High accuracy when needed

[audio.feedback]
enabled = true  # Helpful audio cues when switching models
```

--------------------------------

### Install Rust Toolchain

Source: https://voxtype.io/docs

Set up the Rust environment required for building from source.

```bash
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh
source ~/.cargo/env
```

--------------------------------

### Enable and Start ydotool User Service (Arch Linux)

Source: https://voxtype.io/docs/TROUBLESHOOTING

Enable and start the `ydotool` user service on Arch Linux using `systemctl --user`. Verify its status afterwards.

```bash
# Enable and start ydotool as a user service
systemctl --user enable --now ydotool

# Verify it's running
systemctl --user status ydotool

```

--------------------------------

### Setup whisper.cpp server

Source: https://voxtype.io/docs/USER_MANUAL

Commands to clone, build, and prepare a model for a self-hosted whisper.cpp server.

```bash
# Clone whisper.cpp
git clone https://github.com/ggerganov/whisper.cpp
cd whisper.cpp

# Build the server
make server

# Download a model
./models/download-ggml-model.sh large-v3-turbo
```

--------------------------------

### Install wl-clipboard

Source: https://voxtype.io/docs/TROUBLESHOOTING

Install the `wl-clipboard` package, which provides `wl-copy` and `wl-paste`, on Arch, Debian, or Fedora systems.

```bash
# Install wl-clipboard
# Arch: sudo pacman -S wl-clipboard
# Debian: sudo apt install wl-clipboard
# Fedora: sudo dnf install wl-clipboard

# Test it works
echo "test" | wl-copy
wl-paste

```

--------------------------------

### Enable and Start ydotool System Service (Fedora)

Source: https://voxtype.io/docs/TROUBLESHOOTING

Enable and start the `ydotool` system service on Fedora using `sudo systemctl`. This requires editing the service file to allow user access.

```bash
# 1. Enable and start the system service
sudo systemctl enable --now ydotool

# 2. Edit the service to allow your user to access the socket
sudo systemctl edit ydotool

```

--------------------------------

### Voxtype Setup and Model Management

Source: https://voxtype.io/docs/USER_MANUAL

The `voxtype setup` command is used for initial configuration, downloading default or specific models, and managing transcription engines like ONNX.

```bash
voxtype setup
```

```bash
voxtype setup --download
```

```bash
voxtype setup model
```

```bash
voxtype setup vad
```

```bash
voxtype setup onnx
```

--------------------------------

### Verify Audio Setup

Source: https://voxtype.io/docs

Check audio sources and perform a recording test.

```bash
# List audio sources
pactl list sources short

# Test recording (speak and listen)
arecord -d 3 -f S16_LE -r 16000 test.wav && aplay test.wav && rm test.wav
```

--------------------------------

### Install Parakeet ONNX Binary

Source: https://voxtype.io/docs/PARAKEET

Download and install the ONNX-enabled voxtype binary for your CPU architecture. This step is necessary to use Parakeet. Remember to restart the daemon after installation.

```bash
# Download the Parakeet binary for your CPU
# Example: AVX-512 capable CPU
curl -L https://github.com/peteonrails/voxtype/releases/download/v0.6.3/voxtype-0.6.3-linux-x86_64-onnx-avx512 \
  -o /tmp/voxtype-onnx

# Make executable and install
chmod +x /tmp/voxtype-onnx
sudo mv /tmp/voxtype-onnx /usr/local/bin/voxtype

# Restart the daemon
systemctl --user restart voxtype

# Verify
voxtype --version
```

--------------------------------

### Build and Install .deb package for Voxtype (Debian/Ubuntu)

Source: https://voxtype.io/docs

Builds the Voxtype .deb package from source and installs it on Debian/Ubuntu systems. Requires build dependencies to be installed first.

```bash
# Install build dependencies
sudo apt install build-essential cargo rustc libclang-dev libasound2-dev \
    pkg-config debhelper devscripts

# Clone and build
git clone https://github.com/peteonrails/voxtype
cd voxtype

# Build the package
dpkg-buildpackage -us -uc -b

# Install
sudo dpkg -i ../voxtype_0.1.0-1_*.deb
```

--------------------------------

### Sway Keybinding Example

Source: https://voxtype.io/docs/CONFIGURATION

Example of binding `voxtype record` commands in Sway. Ensure `enabled = false` in your voxtype config for this to work.

```bash
bindsym $mod+v exec voxtype record start
bindsym --release $mod+v exec voxtype record stop
```

--------------------------------

### Configure Polybar Module

Source: https://voxtype.io/docs/WAYBAR

Integration example for Polybar users.

```ini
[module/voxtype]
type = custom/script
exec = voxtype status --format text
interval = 1
format = <label>
label = %output%
```

--------------------------------

### Install Voxtype from .deb package (Debian/Ubuntu)

Source: https://voxtype.io/docs

Downloads and installs the Voxtype .deb package on Debian/Ubuntu systems. Use 'sudo apt-get install -f' to resolve any missing dependencies.

```bash
# Download the latest release
wget https://github.com/peteonrails/voxtype/releases/download/v0.1.0/voxtype_0.1.0-1_amd64.deb

# Install
sudo dpkg -i voxtype_0.1.0-1_amd64.deb

# Install any missing dependencies
sudo apt-get install -f
```

--------------------------------

### Download Whisper Models

Source: https://voxtype.io/docs

Use the built-in setup command to download speech recognition models.

```bash
# Interactive setup (checks dependencies and offers to download)
voxtype setup

# Download the default model (base.en)
voxtype setup --download

# Interactive model selection (choose from available models)
voxtype setup model
```

--------------------------------

### Install dotool and add user to input group

Source: https://voxtype.io/docs/FAQ

Install the dotool package and add the current user to the 'input' group for necessary permissions. Remember to log out and back in after making these changes.

```bash
sudo usermod -aG input $USER
# Log out and back in
```

--------------------------------

### Build whisper-cli from source

Source: https://voxtype.io/docs/TROUBLESHOOTING

Steps to build and install the whisper-cli binary for the CLI backend.

```bash
git clone https://github.com/ggerganov/whisper.cpp
cd whisper.cpp
cmake -B build
cmake --build build --config Release
sudo cp build/bin/whisper-cli /usr/local/bin/
```

--------------------------------

### Install via Cargo

Source: https://voxtype.io/docs

Install directly using the Rust package manager.

```bash
# Install build dependencies first (see above)

# Install from crates.io (when published)
cargo install voxtype

# Or install from git
cargo install --git https://github.com/peteonrails/voxtype
```

--------------------------------

### Install ydotool for X11

Source: https://voxtype.io/docs/USER_MANUAL

Commands to install ydotool for X11 support on various Linux distributions.

```bash
# Install and start ydotool
# Fedora: sudo dnf install ydotool
# Arch: sudo pacman -S ydotool
```

--------------------------------

### Install Recommended Packages (Debian/Ubuntu)

Source: https://voxtype.io/docs

Installs recommended optional packages for Voxtype on Debian/Ubuntu systems, including Wayland clipboard tools, notifications, and X11 fallback tools.

```bash
# For Wayland:
sudo apt install wtype wl-clipboard libnotify-bin
# For X11 or as fallback:
sudo apt install ydotool
```

--------------------------------

### Install libnotify for Desktop Notifications

Source: https://voxtype.io/docs/TROUBLESHOOTING

Install the `libnotify` package (or `libnotify-bin` on Debian) to enable desktop notifications. Test with `notify-send`.

```bash
# Install libnotify
# Arch: sudo pacman -S libnotify
# Debian: sudo apt install libnotify-bin
# Fedora: sudo dnf install libnotify

# Test
notify-send "Test" "This is a test"

```

--------------------------------

### Hyprland Keybinding Example

Source: https://voxtype.io/docs/CONFIGURATION

Example of binding `voxtype record` commands in Hyprland. Ensure `enabled = false` in your voxtype config for this to work.

```hyprlang
bind = SUPER, V, exec, voxtype record start
bindr = SUPER, V, exec, voxtype record stop
```

--------------------------------

### Complete Waybar Configuration

Source: https://voxtype.io/docs/WAYBAR

Full example of a Waybar configuration file including the Voxtype module.

```json
{
    "layer": "top",
    "position": "top",
    "modules-left": ["sway/workspaces"],
    "modules-center": ["sway/window"],
    "modules-right": ["custom/voxtype", "pulseaudio", "clock"],

    "custom/voxtype": {
        "exec": "voxtype status --follow --format json",
        "return-type": "json",
        "format": "{}",
        "tooltip": true
    },

    "clock": {
        "format": "{:%H:%M}"
    },

    "pulseaudio": {
        "format": "{volume}%"
    }
}
```

--------------------------------

### Run Local Whisper.cpp Server

Source: https://voxtype.io/docs/CONFIGURATION

Start a local whisper.cpp server on your GPU. Ensure the model path is correct.

```bash
./server -m models/ggml-large-v3-turbo.bin --host 0.0.0.0 --port 8080
```

--------------------------------

### Setup and download Whisper models

Source: https://voxtype.io/docs/TROUBLESHOOTING

Commands to download required Whisper models for transcription.

```bash
# Download the model
voxtype setup --download

# Or manually download
mkdir -p ~/.local/share/voxtype/models
curl -L -o ~/.local/share/voxtype/models/ggml-base.en.bin \
    https://huggingface.co/ggerganov/whisper.cpp/resolve/main/ggml-base.en.bin
```

--------------------------------

### Start and Stop a Meeting

Source: https://voxtype.io/docs/SMOKE_TESTS

Use these commands to initiate and conclude a meeting recording. A short delay is included to ensure the meeting starts before stopping.

```bash
voxtype meeting start --title "Label Test"
sleep 10
voxtype meeting stop
```

--------------------------------

### Install wtype for Wayland

Source: https://voxtype.io/docs/USER_MANUAL

Commands to install wtype for Wayland support on various Linux distributions.

```bash
# Install wtype
# Fedora: sudo dnf install wtype
# Arch: sudo pacman -S wtype
# Ubuntu: sudo apt install wtype
```

--------------------------------

### Install and configure ydotool for Voxtype

Source: https://voxtype.io/docs/TROUBLESHOOTING

Set up ydotool as a typing backend for Voxtype. This requires installing the package and enabling/starting its systemd user daemon. Verify the daemon's status to ensure it's running.

```bash
# 1. Install ydotool
# Arch:
sudo pacman -S ydotool
# Fedora:
sudo dnf install ydotool
# Ubuntu/Debian:
sudo apt install ydotool

# 2. Enable and start the daemon (required!)
systemctl --user enable --now ydotool

# 3. Verify it's running
systemctl --user status ydotool
```

```bash
# Ubuntu/Debian:
sudo apt install ydotool

# Start the daemon (see "ydotool daemon not running" section for Fedora)
systemctl --user enable --now ydotool
```

--------------------------------

### Install DankMaterialShell Widget for Voxtype

Source: https://voxtype.io/docs/USER_MANUAL

Instructions for installing and managing the Voxtype QML widget plugin for DankMaterialShell, including uninstalling and outputting raw QML.

```bash
voxtype setup dms
```

```bash
voxtype setup dms --install
```

```bash
voxtype setup dms --uninstall
```

```bash
voxtype setup dms --qml
```

--------------------------------

### Complete Moonshine Configuration

Source: https://voxtype.io/docs/CONFIGURATION

Full configuration example for the Moonshine engine.

```toml
engine = "moonshine"

[moonshine]
model = "base"
quantized = true
on_demand_loading = false  # Keep model loaded for fast response
```

--------------------------------

### Webhook Listener Configuration

Source: https://voxtype.io/docs/CI-SETUP

Setup a webhook listener using Docker and define the build hook.

```bash
# webhook-listener.sh
docker run -d \
  --name voxtype-webhook \
  --restart unless-stopped \
  -p 9000:9000 \
  -v /path/to/voxtype:/repo \
  -v /var/run/docker.sock:/var/run/docker.sock \
  adnanh/webhook \
  -hooks /repo/scripts/webhooks.json \
  -verbose
```

```json
[
  {
    "id": "build",
    "execute-command": "/repo/scripts/ci-build.sh",
    "command-working-directory": "/repo"
  }
]
```

```bash
curl -X POST http://your-build-host:9000/hooks/build
```

--------------------------------

### Configure Audio Input

Source: https://voxtype.io/docs

Install ALSA compatibility layers and verify audio source settings for PipeWire or PulseAudio.

```bash
# If using PipeWire, install ALSA compatibility layer
sudo pacman -S pipewire-alsa  # Arch
sudo apt install pipewire-alsa  # Debian/Ubuntu
sudo dnf install pipewire-alsa  # Fedora

# Check PipeWire/PulseAudio is running
pactl info
# Check default source
pactl get-default-source
```

--------------------------------

### Install xclip for Voxtype clipboard mode

Source: https://voxtype.io/docs/TROUBLESHOOTING

Install xclip on Ubuntu/Debian systems to enable Voxtype's clipboard mode when running on X11.

```bash
# Ubuntu/Debian:
sudo apt install xclip
```

--------------------------------

### Download Moonshine models

Source: https://voxtype.io/docs/MOONSHINE

Use the interactive setup tool or manual curl commands to download the necessary model files.

```bash
voxtype setup model
```

```bash
# Create model directory
mkdir -p ~/.local/share/voxtype/models/moonshine-base

# Download the three required files
cd ~/.local/share/voxtype/models/moonshine-base
curl -L https://huggingface.co/onnx-community/moonshine-base-ONNX/resolve/main/onnx/encoder_model.onnx -o encoder_model.onnx
curl -L https://huggingface.co/onnx-community/moonshine-base-ONNX/resolve/main/onnx/decoder_model_merged.onnx -o decoder_model_merged.onnx
curl -L https://huggingface.co/onnx-community/moonshine-base-ONNX/resolve/main/tokenizer.json -o tokenizer.json
```

--------------------------------

### Enable and start ydotool user service

Source: https://voxtype.io/docs/FAQ

Enable and start the ydotool user service to manage input events. This is required if dotool is not used or available.

```bash
systemctl --user enable --now ydotool
```

--------------------------------

### Install Optional Dependencies (Arch Linux)

Source: https://voxtype.io/docs

Installs recommended optional packages for Voxtype on Arch Linux, including Wayland clipboard tools, notifications, and ALSA compatibility for PipeWire.

```bash
# Install recommended optional packages
sudo pacman -S wtype wl-clipboard libnotify

# If using PipeWire (most Arch systems), install ALSA compatibility:
sudo pacman -S pipewire-alsa

# For X11 or as Wayland fallback:
sudo pacman -S ydotool
```

--------------------------------

### Automatic Compositor Setup for Sway

Source: https://voxtype.io/docs/USER_MANUAL

Run `voxtype setup compositor sway` to automatically configure modifier key blocking for Sway. This prevents typed text from triggering shortcuts.

```bash
# For Sway
voxtype setup compositor sway
swaymsg reload
systemctl --user restart voxtype
```

--------------------------------

### Autostart Voxtype with Sway/i3

Source: https://voxtype.io/docs/USER_MANUAL

Add this command to your Sway or i3 configuration file to automatically launch Voxtype when the window manager starts.

```bash
exec --no-startup-id voxtype
```

--------------------------------

### Configure River compositor keybindings

Source: https://voxtype.io/docs/USER_MANUAL

Commands for mapping recording start, stop, and toggle actions in the River compositor.

```bash
riverctl map normal Super V spawn 'voxtype record start'
riverctl map -release normal Super V spawn 'voxtype record stop'

# Toggle mode (press to start/stop)
riverctl map normal Super V spawn 'voxtype record toggle'
```

--------------------------------

### Install Voxtype using AUR helper (Arch Linux)

Source: https://voxtype.io/docs

Installs Voxtype on Arch Linux using an AUR helper like paru or yay.

```bash
# Using paru
paru -S voxtype

# Using yay
yay -S voxtype
```

--------------------------------

### Autostart Voxtype with Hyprland

Source: https://voxtype.io/docs/USER_MANUAL

Include this line in your Hyprland configuration file to ensure Voxtype starts once when Hyprland is launched.

```bash
exec-once = voxtype
```

--------------------------------

### Enable GPU Acceleration (Fedora)

Source: https://voxtype.io/docs

Installs the Vulkan loader and enables GPU acceleration for Voxtype on Fedora.

```bash
sudo dnf install vulkan-loader
sudo voxtype setup gpu --enable
```

--------------------------------

### Complete Parakeet Configuration

Source: https://voxtype.io/docs/CONFIGURATION

Full configuration example for the Parakeet engine.

```toml
engine = "parakeet"

[parakeet]
model = "parakeet-tdt-0.6b-v3"
on_demand_loading = false  # Keep model loaded for fast response
```

--------------------------------

### AI Summarization Command Example

Source: https://voxtype.io/docs/USER_MANUAL

Demonstrates how to trigger AI summarization for the latest meeting and export the summary to a Markdown file. Ensure the summary backend is configured.

```bash
# Requires [meeting.summary] backend set to "local" or "remote"
voxtype meeting summarize latest
voxtype meeting summarize latest --format markdown --output summary.md
```

--------------------------------

### Start Voxtype Daemon

Source: https://voxtype.io/docs

Run the Voxtype daemon in the foreground with varying levels of verbosity.

```bash
# Run in foreground (for testing)
voxtype

# With verbose output
voxtype -v

# With debug logging
voxtype -vv
```

--------------------------------

### Configure meeting audio

Source: https://voxtype.io/docs/CONFIGURATION

Example configuration for audio loopback and echo cancellation settings.

```toml
[meeting.audio]
loopback_device = "auto"
echo_cancel = "auto"  # GTCRN enhancement + transcript dedup
```

--------------------------------

### Setup Modifier Interference Fix for River

Source: https://voxtype.io/docs/TROUBLESHOOTING

Run this command to automatically install a fix that blocks modifier keys during text output in River. This command does not set up keybindings, only the interference fix. A restart of the voxtype service is required.

```bash
voxtype setup compositor river
# Restart River or source the new config
systemctl --user restart voxtype
```

--------------------------------

### Setup Modifier Interference Fix for Sway

Source: https://voxtype.io/docs/TROUBLESHOOTING

Run this command to automatically install a fix that blocks modifier keys during text output in Sway. This command does not set up keybindings, only the interference fix. A restart of the voxtype service is required.

```bash
voxtype setup compositor sway
swaymsg reload
systemctl --user restart voxtype
```

--------------------------------

### Clone and Build from Source

Source: https://voxtype.io/docs

Download the repository and compile the binary.

```bash
git clone https://github.com/peteonrails/voxtype
cd voxtype
cargo build --release
```

--------------------------------

### Setup Modifier Interference Fix for Hyprland

Source: https://voxtype.io/docs/TROUBLESHOOTING

Run this command to automatically install a fix that blocks modifier keys during text output in Hyprland. This command does not set up keybindings, only the interference fix. A restart of the voxtype service is required.

```bash
voxtype setup compositor hyprland
hyprctl reload
systemctl --user restart voxtype
```

--------------------------------

### Initialize Configuration File

Source: https://voxtype.io/docs/USER_MANUAL

Commands to copy the default configuration or generate a new one in the user's home directory.

```bash
cp /etc/voxtype/config.toml ~/.config/voxtype/config.toml

# Or generate one
mkdir -p ~/.config/voxtype
voxtype config > ~/.config/voxtype/config.toml
```

--------------------------------

### Test Input Simulation

Source: https://voxtype.io/docs/FAQ

Commands to verify that input simulation tools are installed and functioning correctly.

```bash
which wtype
```

```bash
wtype "test"
```

```bash
systemctl --user status ydotool
```

```bash
ydotool type "test"
```

```bash
voxtype --clipboard
```

--------------------------------

### Enable GPU Acceleration (Arch Linux)

Source: https://voxtype.io/docs

Installs the Vulkan loader and enables GPU acceleration for Voxtype on Arch Linux.

```bash
sudo pacman -S vulkan-icd-loader
sudo voxtype setup gpu --enable
```

--------------------------------

### Configure On-Demand Model Loading

Source: https://voxtype.io/docs/SMOKE_TESTS

Enable loading the transcription model only when a recording starts to save idle memory.

```bash
# 1. Enable on_demand_loading in config.toml:
#    [whisper]
#    on_demand_loading = true

# 2. Restart daemon
systemctl --user restart voxtype

# 3. Check memory before recording (model not loaded):
systemctl --user status voxtype | grep Memory

# 4. Record and transcribe
voxtype record start && sleep 3 && voxtype record stop

# 5. Check logs for model load/unload:
journalctl --user -u voxtype --since "1 minute ago" | grep -E "Loading|Unloading"
```

--------------------------------

### Example Output Hook for Logging

Source: https://voxtype.io/docs/USER_MANUAL

Use `post_output_command` to log the current date and time to a file after each transcription.

```toml
post_output_command = "echo $(date) >> ~/voxtype.log"
```

--------------------------------

### Pre/Post Output Commands for Notifications

Source: https://voxtype.io/docs/CONFIGURATION

Use pre-output and post-output commands to send notifications when typing starts and finishes.

```toml
[output]
# Notification when typing starts/finishes
pre_output_command = "notify-send 'Typing...'"
post_output_command = "notify-send 'Done'"
```

--------------------------------

### Verify Multilingual Models

Source: https://voxtype.io/docs/SMOKE_TESTS

Commands to test language auto-detection and verify model availability in the setup menu.

```bash
# Use a multilingual model (without .en suffix)
voxtype --model small record start
sleep 3
voxtype record stop

# Check logs for language auto-detection:
journalctl --user -u voxtype --since "30 seconds ago" | grep "auto-detected language"

# Verify model menu shows multilingual options:
echo "0" | voxtype setup model  # Should show tiny, base, small, medium (multilingual)
```

--------------------------------

### Install Voxtype via RPM

Source: https://voxtype.io/docs

Download and install the specific RPM package manually.

```bash
# Download the latest release
wget https://github.com/peteonrails/voxtype/releases/download/v0.1.0/voxtype-0.1.0-1.fc39.x86_64.rpm

# Install
sudo dnf install ./voxtype-0.1.0-1.fc39.x86_64.rpm
```

--------------------------------

### Manual AUR Build for Voxtype (Arch Linux)

Source: https://voxtype.io/docs

Manually builds and installs Voxtype from the AUR on Arch Linux.

```bash
git clone https://aur.archlinux.org/voxtype.git
cd voxtype
makepkg -si
```

--------------------------------

### Quick Smoke Test Script

Source: https://voxtype.io/docs/SMOKE_TESTS

A standalone shell script to verify basic installation health and core functionality.

```bash
#!/bin/bash
# quick-smoke-test.sh - Run after new build install

set -e
echo "=== Voxtype Smoke Tests ==="

echo -n "Version: "
voxtype --version

echo -n "Status: "
voxtype status

echo "Recording 3 seconds..."
voxtype record start
sleep 3
voxtype record stop
echo "Done."

echo ""
echo "Check logs:"
journalctl --user -u voxtype --since "30 seconds ago" --no-pager | tail -10
```

--------------------------------

### Configure Whisper Engine

Source: https://voxtype.io/docs/MODEL_SELECTION_GUIDE

Configuration examples for the Whisper engine, demonstrating settings for English-only performance and multilingual GPU usage.

```toml
# English-only, good CPU performance
[whisper]
model = "small.en"
language = "en"
```

```toml
# Multilingual with GPU
[whisper]
model = "large-v3-turbo"
language = "auto"    # or "ja", "zh", "ko", "ar", etc.
```

--------------------------------

### Configure Hyprland Keybindings for Voxtype Profiles

Source: https://voxtype.io/docs/USER_MANUAL

Set up compositor keybindings, using Hyprland as an example, to easily switch between different Voxtype profiles during recordings.

```hyprlang
# Different keybindings for different profiles
bind = SUPER, V, exec, voxtype record start
bindr = SUPER, V, exec, voxtype record stop

bind = SUPER SHIFT, V, exec, voxtype record start --profile slack
bindr = SUPER SHIFT, V, exec, voxtype record stop

bind = SUPER CTRL, V, exec, voxtype record start --profile code
bindr = SUPER CTRL, V, exec, voxtype record stop

```

--------------------------------

### Autostart Voxtype with River

Source: https://voxtype.io/docs/USER_MANUAL

Add this command to your River window manager's init file to spawn Voxtype when River starts.

```bash
riverctl spawn voxtype
```

--------------------------------

### Manage DankMaterialShell Voxtype Plugin

Source: https://voxtype.io/docs/USER_MANUAL

Commands to manage the DankMaterialShell Voxtype plugin, including showing setup instructions, uninstalling, or outputting raw QML.

```bash
voxtype setup dms              # Show setup instructions
voxtype setup dms --uninstall  # Remove the widget
voxtype setup dms --qml        # Output raw QML (for scripting)
```

--------------------------------

### Install Voxtype via COPR

Source: https://voxtype.io/docs

Use the Fedora COPR repository for a streamlined installation process.

```bash
sudo dnf copr enable pete/voxtype
sudo dnf install voxtype
```

--------------------------------

### Full Configuration Reference

Source: https://voxtype.io/docs/USER_MANUAL

Complete configuration options for hotkeys, audio input, whisper models, and output modes.

```toml
[hotkey]
# The key to hold for push-to-talk recording
# Common choices: SCROLLLOCK, PAUSE, RIGHTALT, F13-F24
key = "SCROLLLOCK"

# Optional modifier keys that must be held with the main key
# Examples: ["LEFTCTRL"], ["LEFTCTRL", "LEFTALT"]
modifiers = []

# Enable built-in hotkey detection (default: true)
# Set to false when using compositor keybindings (Hyprland, Sway) instead
# enabled = true

[audio]
# Audio input device
# "default" uses the system default microphone
# List devices with: pactl list sources short
device = "default"

# Sample rate in Hz (whisper expects 16000)
sample_rate = 16000

# Maximum recording duration in seconds (safety limit)
# Recording automatically stops and transcribes after this time
max_duration_secs = 60

[whisper]
# Model to use for transcription
# Options: tiny, tiny.en, base, base.en, small, small.en, medium, medium.en, large-v3
# .en models are English-only but faster/smaller
# Can also be an absolute path to a .bin model file
model = "base.en"

# Language for transcription
# "en" for English, "auto" for auto-detection
# See: https://github.com/openai/whisper#available-models-and-languages
language = "en"

# Translate non-English speech to English
translate = false

# Number of CPU threads for inference
# Omit to auto-detect optimal thread count
# threads = 4

# Load model on-demand (saves memory/VRAM, slight delay per recording)
# on_demand_loading = true

[output]
# Primary output mode
# "type" - Simulates keyboard input at cursor (wtype/dotool/ydotool)
# "clipboard" - Copies text to clipboard (requires wl-copy)
# "paste" - Copies to clipboard then simulates Ctrl+V (for non-US keyboard layouts)
mode = "type"

# Fall back to clipboard if typing fails
fallback_to_clipboard = true

# Delay between typed characters in milliseconds
# 0 = fastest, increase if characters are dropped
type_delay_ms = 0

[output.notification]
# Show notification when recording starts (hotkey pressed)
on_recording_start = false

# Show notification when recording stops (transcription begins)
on_recording_stop = false

# Show notification with transcribed text
on_transcription = true
```

--------------------------------

### Enable Voxtype Systemd Service

Source: https://voxtype.io/docs/USER_MANUAL

Automatically start Voxtype on system login by enabling and starting its user systemd service.

```bash
systemctl --user enable --now voxtype
```

--------------------------------

### Start Meeting Without Title

Source: https://voxtype.io/docs/SMOKE_TESTS

Start a meeting without providing a title; the system will auto-generate one based on the date and time.

```bash
# Start without a title (should auto-generate one from the date)
voxtype meeting start
sleep 5
voxtype meeting stop

# Verify auto-generated title in list
voxtype meeting list
# Expected: title like "Meeting 2026-02-16 14:30"
```

--------------------------------

### Start and Stop Voxtype Recording

Source: https://voxtype.io/docs/SMOKE_TESTS

Basic command to start and stop Voxtype recording. Check logs to confirm Wayland is used.

```bash
voxtype record start && sleep 2 && voxtype record stop
journalctl --user -u voxtype --since "30 seconds ago" | grep -E "wtype|Text output"
```

--------------------------------

### Start and Control Voxtype Meetings

Source: https://voxtype.io/docs/USER_MANUAL

Manage meeting transcriptions using commands like `voxtype meeting start`, `pause`, `resume`, and `stop`.

```bash
# Start a new meeting
voxtype meeting start
voxtype meeting start --title "Weekly standup"

# Control a running meeting
voxtype meeting pause
voxtype meeting resume
voxtype meeting stop

```

--------------------------------

### Build with GPU Support

Source: https://voxtype.io/docs/FAQ

Compile from source with specific GPU backend features.

```bash
cargo build --release --features gpu-cuda
```

--------------------------------

### Install DankMaterialShell Voxtype Plugin

Source: https://voxtype.io/docs/USER_MANUAL

Automatically install the Voxtype QML plugin for DankMaterialShell. This adds a Voxtype widget to your KDE Plasma panel.

```bash
voxtype setup dms --install
```

--------------------------------

### Switch GPU Backends

Source: https://voxtype.io/docs/SMOKE_TESTS

Manage transitions between CPU and GPU (Vulkan/CUDA) backends by updating system symlinks.

```bash
# Check current status
voxtype setup gpu

# Whisper mode (symlink points to voxtype-vulkan or voxtype-avx*)
# --enable switches to Vulkan, --disable switches to best CPU
ls -la /usr/bin/voxtype  # Verify current symlink
sudo voxtype setup gpu --enable   # Switch to Vulkan
ls -la /usr/bin/voxtype  # Should point to voxtype-vulkan
sudo voxtype setup gpu --disable  # Switch to best CPU (avx512 or avx2)
ls -la /usr/bin/voxtype  # Should point to voxtype-avx512 or voxtype-avx2

# ONNX mode (symlink points to voxtype-onnx-*)
# --enable switches to CUDA, --disable switches to best ONNX CPU
sudo ln -sf /usr/lib/voxtype/voxtype-onnx-avx512 /usr/bin/voxtype
sudo voxtype setup gpu --enable   # Switch to ONNX CUDA
ls -la /usr/bin/voxtype  # Should point to voxtype-onnx-cuda
sudo voxtype setup gpu --disable  # Switch to best ONNX CPU
ls -la /usr/bin/voxtype  # Should point to voxtype-onnx-avx512

# Restore to Whisper Vulkan for normal use
sudo ln -sf /usr/lib/voxtype/voxtype-vulkan /usr/bin/voxtype
```

--------------------------------

### Dual Audio Sources (Loopback)

Source: https://voxtype.io/docs/SMOKE_TESTS

Configure and test dual audio sources by enabling loopback to capture both microphone input and system audio. Verify speaker attribution in show and export commands.

```bash
# Verify loopback detection
# 1. Configure loopback in config.toml:
#    [meeting.audio]
#    loopback_device = "auto"

# 2. Start a meeting while in a video call (Zoom, Teams, etc.)
voxtype meeting start --title "Video Call Test"

# 3. Speak into mic and wait for remote participants to speak
sleep 30
voxtype meeting stop

# 4. Check speaker attribution
voxtype meeting show latest
# Expected: segments attributed to "You" (mic) and "Remote" (loopback)

# 5. Verify export includes speaker labels
voxtype meeting export latest --format text --speakers
# Expected: "You:" and "Remote:" labels in output

# Disable loopback (mic-only mode)
#    [meeting.audio]
#    loopback_device = "disabled"
systemctl --user restart voxtype
voxtype meeting start --title "Mic Only Test"
sleep 10
voxtype meeting stop
voxtype meeting show latest
# Expected: all segments attributed to "You" or "Unknown"
```

--------------------------------

### Troubleshooting Build Issues

Source: https://voxtype.io/docs/CI-SETUP

Commands to manage build resources and system permissions.

```bash
export CARGO_BUILD_JOBS=1
./scripts/ci-build.sh vulkan
```

```bash
sudo usermod -aG docker $USER
```

```bash
export CARGO_BUILD_JOBS=1
export CMAKE_BUILD_PARALLEL_LEVEL=1
```

--------------------------------

### Example TOML configuration with profiles

Source: https://voxtype.io/docs/CONFIGURATION

This TOML configuration demonstrates setting a default post-processing command and timeout, along with several profiles for different use cases like Slack, code comments, and meeting notes, each with custom settings.

```toml
# Default post-processing
[output.post_process]
command = "ollama run llama3.2:1b 'Clean up:'"
timeout_ms = 30000

# Profile: casual chat
[profiles.slack]
post_process_command = "ollama run llama3.2:1b 'Rewrite casually for Slack:'"

# Profile: code comments, output to clipboard
[profiles.code]
post_process_command = "ollama run llama3.2:1b 'Format as code comment:'"
output_mode = "clipboard"

# Profile: meeting notes with longer timeout
[profiles.notes]
post_process_command = "ollama run llama3.2:1b 'Convert to bullet points:'"
post_process_timeout_ms = 60000
```

--------------------------------

### Speaker Labeling Example

Source: https://voxtype.io/docs/USER_MANUAL

Example of using the `voxtype meeting label` command to assign names to speakers. This command can use either the full speaker ID or just the numerical index.

```bash
voxtype meeting label latest SPEAKER_00 "Alice"
voxtype meeting label latest 1 "Bob"  # Short form: just the number
```

--------------------------------

### Automatic Compositor Setup for River

Source: https://voxtype.io/docs/USER_MANUAL

Run `voxtype setup compositor river` to automatically configure modifier key blocking for River. This prevents typed text from triggering shortcuts.

```bash
# For River
voxtype setup compositor river
# Restart River or source the new config
systemctl --user restart voxtype
```

--------------------------------

### Basic Voxtype CLI Commands

Source: https://voxtype.io/docs/SMOKE_TESTS

Verify installation and check help documentation for core commands. Use `voxtype config` to view current settings and `voxtype status` to check the daemon's operational state.

```bash
# Version and help
voxtype --version
voxtype --help
voxtype daemon --help
voxtype record --help
voxtype setup --help

# Show current config
voxtype config

# Check status
voxtype status
```

--------------------------------

### Automatic Compositor Setup for Hyprland

Source: https://voxtype.io/docs/USER_MANUAL

Run `voxtype setup compositor hyprland` to automatically configure modifier key blocking for Hyprland. This prevents typed text from triggering shortcuts.

```bash
# For Hyprland
voxtype setup compositor hyprland
hyprctl reload
systemctl --user restart voxtype
```

--------------------------------

### Apply Custom Configuration File

Source: https://voxtype.io/docs/USER_MANUAL

Use the -c flag to specify a custom path for the configuration file.

```bash
voxtype -c /path/to/my/config.toml
```

--------------------------------

### Voxtype Status JSON Output Example

Source: https://voxtype.io/docs/USER_MANUAL

Example of JSON output from `voxtype status --format json --extended`, showing the current state, model, device, and backend information.

```json
{
  "text": "🎙️",
  "class": "idle",
  "tooltip": "Voxtype ready\nModel: base.en\nDevice: default\nBackend: CPU (AVX-512)",
  "model": "base.en",
  "device": "default",
  "backend": "CPU (AVX-512)"
}
```

--------------------------------

### Configure Hyprland Keybindings

Source: https://voxtype.io/docs/USER_MANUAL

Use bindr for push-to-talk functionality in Hyprland.

```hyprlang
# Push-to-talk (hold to record, release to transcribe)
bind = , SCROLL_LOCK, exec, voxtype record start
bindr = , SCROLL_LOCK, exec, voxtype record stop

# Or with a modifier
bind = SUPER, V, exec, voxtype record start
bindr = SUPER, V, exec, voxtype record stop

# Toggle mode (press to start/stop)
bind = SUPER, V, exec, voxtype record toggle
```

--------------------------------

### Install dotool for Voxtype

Source: https://voxtype.io/docs/TROUBLESHOOTING

Install dotool, a recommended tool for Voxtype output on KDE Plasma and GNOME Wayland, and X11. Ensure your user is added to the 'input' group and log out/in for changes to take effect. Configure 'dotool_xkb_layout' in 'config.toml' for non-US keyboards.

```bash
# 1. Install dotool (check your distribution's package manager)
# Arch (AUR):
yay -S dotool
# From source: https://sr.ht/~geb/dotool/

# 2. Add user to input group (required for uinput access)
sudo usermod -aG input $USER
# Log out and back in for group change to take effect

# 3. Configure keyboard layout if needed (non-US keyboards)
# Add to config.toml:
# [output]
# dotool_xkb_layout = "de"  # German, French ("fr"), etc.
```

```bash
# Ubuntu/Debian (from source):
sudo apt install libxkbcommon-dev
git clone https://git.sr.ht/~geb/dotool
cd dotool && ./build.sh && sudo cp dotool /usr/local/bin/

# Arch (AUR):
yay -S dotool

# Add user to input group
sudo usermod -aG input $USER
# Log out and back in
```

```bash
# 1. Install dotool
# Arch (AUR):
yay -S dotool
# Ubuntu/Debian (from source):
# See https://sr.ht/~geb/dotool/ for instructions
# Fedora (from source):
# See https://sr.ht/~geb/dotool/ for instructions

# 2. Add user to input group (required for uinput access)
sudo usermod -aG input $USER
# Log out and back in for group change to take effect
```

--------------------------------

### Switch Models via CLI

Source: https://voxtype.io/docs/USER_MANUAL

Commands to trigger recording with specific models using the CLI.

```bash
# Use primary model
voxtype record start

# Use secondary model
voxtype record start --model large-v3-turbo

# Use any available model
voxtype record start --model medium.en
```

--------------------------------

### Clone and Build Voxtype

Source: https://voxtype.io/docs/CI-SETUP

Initial steps to clone the repository and execute the build script.

```bash
git clone https://github.com/peteonrails/voxtype.git
cd voxtype
```

```bash
./scripts/ci-build.sh
```

--------------------------------

### Configure CLI backend

Source: https://voxtype.io/docs/USER_MANUAL

Update configuration to use the CLI backend instead of FFI.

```toml
[whisper]
backend = "cli"
model = "base.en"
language = "en"

# Optional: specify path if not in PATH
# whisper_cli_path = "/usr/local/bin/whisper-cli"
```

--------------------------------

### Configure On-Demand Loading

Source: https://voxtype.io/docs/CONFIGURATION

Loads the model only when recording starts to save memory.

```toml
[moonshine]
model = "base"
on_demand_loading = true  # Free memory when not transcribing
```

--------------------------------

### Build whisper.cpp with GPU support

Source: https://voxtype.io/docs/USER_MANUAL

Compile the server with hardware acceleration for CUDA, Vulkan, or Metal.

```bash
# CUDA (NVIDIA)
make server GGML_CUDA=1

# Vulkan (AMD/NVIDIA/Intel)
make server GGML_VULKAN=1

# Metal (macOS)
make server GGML_METAL=1
```

--------------------------------

### Configure Sway Keybindings

Source: https://voxtype.io/docs/USER_MANUAL

Use --release to handle key release events in Sway.

```bash
# Push-to-talk
bindsym Scroll_Lock exec voxtype record start
bindsym --release Scroll_Lock exec voxtype record stop

# With a modifier
bindsym $mod+v exec voxtype record start
bindsym --release $mod+v exec voxtype record stop

# Toggle mode
bindsym $mod+v exec voxtype record toggle
```

--------------------------------

### Download and Switch Voxtype Models

Source: https://voxtype.io/docs/SMOKE_TESTS

Instructions for downloading and switching between different Voxtype models. Models can be selected interactively or specified directly via the command line.

```bash
# Download a different model if not present
voxtype setup model  # Interactive selection

# Or specify directly
voxtype setup model small.en

# Test with different models (edit config.toml or use --model flag)
```

--------------------------------

### Override status icons

Source: https://voxtype.io/docs/CONFIGURATION

Example of overriding specific icons while using a base theme.

```toml
[status]
icon_theme = "emoji"

[status.icons]
recording = "🔴"  # Override just the recording icon
```

--------------------------------

### Select Multi-GPU Devices

Source: https://voxtype.io/docs/SMOKE_TESTS

Identify available GPUs and force selection using environment variables.

```bash
# Check detected GPUs
voxtype setup gpu
# Expected: lists all detected GPUs with vendor names

# Test GPU selection via environment variable
VOXTYPE_VULKAN_DEVICE=amd voxtype setup gpu | grep "GPU selection"
# Expected: "GPU selection: AMD (via VOXTYPE_VULKAN_DEVICE)"

VOXTYPE_VULKAN_DEVICE=nvidia voxtype setup gpu | grep "GPU selection"
```

--------------------------------

### Configure GPU Isolation

Source: https://voxtype.io/docs/SMOKE_TESTS

Enable GPU memory release for hybrid graphics setups.

```bash
# 1. Enable gpu_isolation in config.toml:
#    [whisper]
#    gpu_isolation = true

# 2. Restart daemon
systemctl --user restart voxtype

# 3. Record and transcribe
voxtype record start && sleep 3 && voxtype record stop

# 4. Check logs for subprocess spawning:
journalctl --user -u voxtype --since "1 minute ago" | grep -i subprocess

# 5. Verify GPU memory is released after transcription:
#    (AMD) watch -n1 "cat /sys/class/drm/card*/device/mem_info_vram_used"
#    (NVIDIA) nvidia-smi
```

--------------------------------

### Find Audio Input Devices

Source: https://voxtype.io/docs/CONFIGURATION

Use the `pactl list sources short` command in the terminal to list available audio input devices and their names. This is useful for configuring the `device` option in the `[audio]` section.

```bash
pactl list sources short
```