The Alexa Smart Screen SDK (by Amazon) provides a simple way to build screen based Alexa devices (Echo Show like things). As the build steps are tricky and tedious, this Docker wraps it all up with Alpine Linux to make it much easier for someone to try.
The repository contains a RUN script which will run the Docker image correctly configured. The essential docker run command is as follows:
docker run --name alexa
--mount type=bind,source=$PWD/run/database,target=/alexa/database
--mount type=bind,source=$PWD/run/config,target=/alexa/config
--mount type=bind,source=$PWD/run/GUI,target=/alexa/GUI
--mount type=bind,source=/dev/snd,target=/dev/snd
--network host
--env HOSTIP="..."
--privileged \
alexa /alexa/run.sh
There are various binds which are important for successful operation:
- /alexa/database
Binds to an empty directory where various SQLite databases are created. Without this, everytime the application is restarted, it will have to reauthenticate with the Alexa Voice Service.
- /alexa/config
Binds to a directory holding your configuation. If these configuration files don't exists, they will be created on first-run. You can then edit them.
- /alexa/GUI
Binds to an empty directy where the application will place HTML and JS files. These files are used to display the smart screen in a browser.
- /dev/snd
Binds to the sound device on the host, provides microphone and speaker access for Alexa.
Tested using host networking. The server attaches to all networks by default - simple but not terribly secure. The actual host ip is passed via the HOSTIP environment and is used to modify the Javascript loaded by the web browser.
Sound is provided using the hosts /dev/snd devices, and is configured in the container's /etc/asound.conf file. By default, sound is played and captured using hardware device 1. This can be changed by binding an alternative configuration to this filename.
...