Podcast aggregator

Create podcasts from Youtube, Odysee and Soundcloud. Also supports creating a master feed by aggregating multiple sources. Integrates with S3 compatible object storage - no podcast hosting software required.

Use cases:

• Create a podcast for a Youtube or Soundcloud channel
• Create a podcast for a curated list of Youtube videos or Soundcloud assets
• Create a master feed that aggregates audio from multiple sources

Configuration

Create a podcast.config.js (see example.config.js) to set podcast metadata and configure the upload to S3. Ensure your bucket allows public access (or configure Cloudfront).

Create a .env file with the following environment variables (or set them manually when starting the process):

• AWS_ACCESS_KEY_ID
• AWS_SECRET_ACCESS_KEY
• HTTP_PROXY (optional, beta) A proxy can be used to circumvent geo restrictions. Try using one if you're getting 403 responses from the youtube API
• NODE_OPTIONS (optional) If you're on a machine with limited memory, consider setting NODE_OPTIONS to --max-old-space-size=500 whereas 500 is a bit lower than your limit in MB. This forces the garbage collector before you run out of memory

Usage

npm start

If you want to pass a different config file than the default:

npm start -- --config=./other.config.js

Your feed will be uploaded to your AWS S3 bucket as an XML file. If you don't intend to publish your feed publicly, pocketcasts supports submitting private feeds (also a good way to test your configuration).

Good to know

• A machine with disk is recommended as audio files get cached locally
• Does not retain a full history of all previously processed episodes. New episodes in the configured feeds will eventually replace new ones
• The Youtube API can have restrictions, use a proxy to work around it
• Youtube downloads can be flaky in general (rate limits, etc). It's running sufficiently stable, but don't expect this to run without any hiccups all the time
• Soundcloud is incredibly restrictive and the API is not public. A developer account is required to get a personal clientId
• To monitor a background cronjob, consider setting up a dead mans switch https://nosnch.in
• This is an early version that works for me (TM), contributions are welcome