salleq/glados-docker

*** DISCLAIMER ***

I DON'T KNOW WHAT I'M DOING THIS IS MY FIRST TIME TRYING TO CREATE A DOCKER IMAGE

*** END OF DISCLAIMER ***

This Dockerfile should build a Ubuntu 20.04 based image with all the needed prerequirements for getting https://github.com/nerdaxic/glados-voice-assistant to work. Consider this work in progress (as is the app this docker image is for).

AVX2 is required from the CPU. Ubuntu 20.04 because the developer is using it too🤷‍♂️

Usage: clone the glados-voice-assistant and glados-tts repos to your home folder.

git clone https://github.com/nerdaxic/glados-voice-assistant
cd glados-voice-assistant
git submodule init
git submodule update

Then clone this repo to your home folder.

git clone https://github.com/salleq/glados-docker

rename the settings.env.sample in the GLaDOS repo to settings.env and modify to your needs. This seems to use whichever sound card and microphone you've chosen in Pulseaudio (?) Set at least the TTS_SAMPLE_FOLDER to point to /glados-voice-assistant/audio/tts Also create ~/glados-voice-assistant/audio/tts folder. Then:

./build.sh

check host_runner.sh so that folders are intact. If your used ID is something else than 1000 you may need to edit more stuff. This doesn't work if your user doesn't have docker access (if you need to 'sudo' to run the next script) Also your user ID and GID should be 1000, if it's something different then modify host_runner.sh accordingly.

You also have to edit gladosTTS.py in the folder: add this on line 19(ish):

from subprocess import call

a few lines below that modify def playFile like this:

def playFile(filename):
        call(["aplay", filename])

finally, find the comment #speak from the file and and replace this:

ps('output.wav')

with this

call(["aplay", "./output.wav"])

save the file. Then:

./host_runner.sh

then you can run the python script (inside the container) with:

cd ~/glados-voice-assistant
python3 /glados.py

I've had audio working, as long as it's selected from GUI (I don't know if it uses pulseaudio or alsa). I don't know how to change that from console.