Gencaster

Gencaster is a non-linear audio streaming framework for real-time radiophonic experiences and live music. It consists of multiple services which represent all the necessary building blocks.

The audio streams have a low latency (about 150ms) and can be listened to in any modern browser, and can dynamically render audio content based on a given story graph that can react to user input such as name, time, GPS position, or even microphone streaming after permissions have been granted.

caster-sound is a service that handles all streaming and audio rendering functionality, using SuperCollider to generate audio and Janus to distribute audio to listeners via WebRTC
caster-back is a web backend to manage the streams of caster-sound, written in Django
caster-front is a web frontend that allows users to listen to the streams of caster-sound, written in Vue
caster-editor is a web-editor in which the actions of a stream, called a story graph, can be created, edited and tweaked using Python or SuperCollider, and is also written in Vue

For further information please visit:

Documentation

Gencaster uses Sphinx for documentation and can be accessed online via GitHub pages at gencaster.github.io/gencaster. The sources for the documentation are located in the ./docs directory and can be build locally by executing

make docs

Development

In order to have consistent styles and a good history Gencaster uses pre-commit. After installation of pre-commit and cloning the repository the necessary scripts can be set up via pre-commit install.

Secrets

The secrets are located in a .secrets.env file and are not checked in with the repository, so it is necessary to create one via

touch .secrets.env

An empty file is sufficient, the docs cover the optional variables.

Local development

To start a local instance of Gencaster with all its services using Docker simply use

make docker-local

There are additional flags to control which services are spawned

flag	comment
`no-editor=1`	Starts without the editor
`no-frontend=1`	Starts without the frontend
`no-sound=1`	Starts without the sound server

Example: make no-editor=1 no-frontend=1 docker-local starts without editor and frontend.

This allows e.g. to have access to the Gencaster backend but use the host environment to develop the editor and frontend as development with NodeJS within Docker is complicated due to the shared node_modules directory.

The default development credentials for the backend http://127.0.0.1:8081/admin are located in vars.env and are

key	value
username	`admin`
passworrd	`admin`

Due to shared cookies it is necessary to access the frontend and the editor via 127.0.0.1 instead of localhost!

Network

As WebRTC only works within a SSL environment we use a nginx reverse proxy to forward the port 8089 to the local port 8088 which is the http version of the Janus server. By doing this we can let nginx handle the SSL context and not need to embed this into Janus.

Copyright

License

AGPL-3.0

Gencaster / gencaster

readme