novavoice 0.1.0

Local, offline voice dictation daemon for Linux — hold Space, speak, release

NovaVoice

Local, offline voice dictation for Linux. Hold Space anywhere on your desktop, speak, release — the transcribed text appears in whatever app is focused. No cloud, no GPU required.

---

How it works

Hold Space (> 0.5s) → microphone opens → speak → release Space → text appears

Powered by faster-whisper (CPU/int8). Works in browsers, terminals, VS Code, and any other app.

---

Install

Option 1 — One-line installer (recommended)

bash <(curl -fsSL https://raw.githubusercontent.com/novafabric/novavoice/main/install.sh)

Installs system dependencies, sets up auto-start on login, and guides you through the one-time setup.

---

Option 2 — pipx

# System dependencies
sudo apt install libportaudio2 xdotool xclip pipx

# Keyboard access (requires logout)
sudo usermod -aG input $USER
logout

# Install
pipx install novavoice
pipx ensurepath

# Auto-start on login
mkdir -p ~/.config/systemd/user
curl -fsSL https://raw.githubusercontent.com/novafabric/novavoice/main/contrib/novavoice.service \
  -o ~/.config/systemd/user/novavoice.service
systemctl --user enable --now novavoice.service

---

Option 3 — .deb package

Download the latest .deb from Releases, then:

sudo apt install ./novavoice_*.deb

# Keyboard access (requires logout)
sudo usermod -aG input $USER
logout

# Auto-start on login
systemctl --user enable --now novavoice.service

---

Usage

Once running, NovaVoice works silently in the background.

Action	Result
Hold Space (0.5s+), speak, release	Text injected into focused app
novavoice status	Check if daemon is running
novavoice stop	Stop the daemon
novavoice start	Start the daemon
novavoice doctor	Check all prerequisites
novavoice inject "hello"	Test injection without recording

---

Configuration

Copy the example config and customize:

mkdir -p ~/.config/novavoice
cp examples/config.example.toml ~/.config/novavoice/config.toml

Key settings:

[stt]
model = "tiny.en"   # faster; try "base.en" for better accuracy

[hotkey]
hold_threshold_ms = 500   # how long to hold Space before recording starts

[audio]
max_record_seconds = 90

---

System requirements

Requirement	Notes
Ubuntu 22.04+ (or any Linux with systemd)	X11 or Wayland
Python 3.11+
libportaudio2	sudo apt install libportaudio2
xdotool (X11) or ydotool/wtype (Wayland)	For text injection
User in input group	sudo usermod -aG input $USER then re-login

Run novavoice doctor to check everything at once.

---

Troubleshooting

evdev access fails

sudo usermod -aG input $USER
# Log out and back in (required)

Nothing happens when holding Space

novavoice status    # is it running?
novavoice doctor    # check all prerequisites
journalctl --user -u novavoice.service -f   # view daemon logs

Poor transcription quality Set model = "base.en" in ~/.config/novavoice/config.toml. Slower but more accurate.

Wayland injection not working Install ydotool and ensure ydotoold is running:

sudo apt install ydotool
sudo ydotoold &

---

License

Apache 2.0 — see LICENSE.