This repository contains home server version of Almond, the end user programmable assistant. It is a single-user version, suitable for running on low-power devices and smart speakers.
Almond is part of Open Thing Platform, a research project led by prof. Monica Lam, from Stanford University. You can find more information at https://thingpedia.stanford.edu/about.
The recommended way to run almond-server is through podman, a replacement for docker that allows the container to run as your regular user (and thus access PulseAudio from your normal session).
To run, use the command:
podman run -p 3000:3000 --uidmap keep-id \
-v /dev/shm:/dev/shm \
-v $XDG_RUNTIME_DIR/pulse:/run/pulse \
-v ${XDG_CONFIG_HOME:-$HOME/.config}/almond-server:/var/lib/almond-server \
stanfordoval/almond-serverYou can now navigate to 127.0.0.1:3000 to access Almond, or use your voice with the hotword "Almond".
Voice support is only available on Linux. On Mac or Windows, you can use the following docker command:
docker run -p 3000:3000 \
-v $HOME/.almond-server:/var/lib/almond-server \
stanfordoval/almond-server:latest-portableChange the -v line to a different path to save the almond-server configuration files in a different directory than $HOME/.almond-server.
To develop almond-server, you should clone this repository, then install the dependencies with:
yarn
This will only install the minimal set of dependencies, and will not install any voice support. To enable voice, you must also run (Linux only):
dnf -y install pulseaudio-libs-devel mimic-devel python3-numpy python3-scipy portaudio-devel libcanberra-devel
pip3 install 'tensorflow<2.0.0' 'git+https://github.com/stanford-oval/mycroft-precise'
then run yarn again to pick up the new dependencies.
(The above command is for the Fedora Linux distribution. Other distribution have similar commands.)
After installing the dependencies locally, the server can be started using yarn start.