voice_linux (C99)

Local hotkey dictation for Linux/X11 using:

PulseAudio capture
whisper.cpp transcription
X11/XTest text injection

Build deps (Debian/Qubes AppVM)

sudo apt update
sudo apt install -y build-essential cmake git wget libpulse-dev libx11-dev libxtst-dev libgtk-3-dev pkg-config

Build app

Use build.sh:

# CPU whisper build
bash ./build.sh

# CUDA whisper build
WHISPER_CUDA=1 bash ./build.sh

# build without whisper integration
WITH_WHISPER=0 bash ./build.sh

./voice_linux is the single app binary and always launches the GTK window.

Download whisper models

Download all supported test models:

bash ./scripts/download_models.sh

This fetches:

ggml-tiny.en.bin
ggml-base.en.bin
ggml-small.en.bin
ggml-medium.en.bin
ggml-large-v3-turbo.bin
ggml-large-v3.bin

Run

./voice_linux

On startup, the app opens a dialog where you choose:

CPU or GPU mode
Whisper model from ./models/

Default hotkey: Ctrl+Alt+V

Press once to start recording
Press again to stop recording, transcribe, and type into the focused window

If you see [stt] [BLANK_AUDIO], recording worked but audio energy was near silent. Check your input source with:

pactl list short sources

Then set audio_device= in config.ini.

Versioning workflow

This project uses semantic git tags (vMAJOR.MINOR.PATCH) plus header macros in src/version.h.

Use increment_and_push.sh to increment version, update VOICE_LINUX_VERSION, commit, tag, and push:

# patch bump (default)
./increment_and_push.sh "Fix VAD trigger behavior"

# minor bump
./increment_and_push.sh -m "Add new GUI debug indicators"

# major bump
./increment_and_push.sh -M "Breaking config change"

# release mode (also runs bash ./build.sh and creates tarball)
./increment_and_push.sh -r -m "Release minor update"

Notes

This implementation targets X11 (not native Wayland).
In template-based AppVMs, system packages may not persist unless installed in the template.

2.1 KiB Raw Permalink Blame History