Deej-AI VS Snapcast

For automatic playlists that make at least some sense I used https://github.com/teticio/Deej-AI in the past, is not human-level good but the results are enjoyable.

I recently found this Deej-Ai project on github (https://github.com/teticio/Deej-AI), and I think it would be perfect for this use case. The AI can either be Self-hosted (as mentioned in the github page) but there is also a publicly accessible site (https://deej-ai.online/playlist). On this post I'm going to focus on a possible implementation using the site.

This is FAR from a click or two but I have wanted to do something similar. I haven't had the time to dig into this yet but I came across https://github.com/teticio/Deej-AI which is a way to train models based on your personal music library (among other sources). Ideally from this you'd be able to create Jellyfin playlists but using the Jellyfin API: https://api.jellyfin.org/#tag/Playlists

https://github.com/badaix/snapcast!

Works perfectly on pis scattered around the house.

Snapcast: The cheapest and most modular option is deploying small form factor linux clients (raspberry pis, or any host all) with an audio interface (3.5mm, RCA, or otherwise), and connect the audio output interface to your choice of speaker. Snapcast is the open-source multi-room brains. It works well in my experience, but it is tricky to get setup properly to start with. Spotify Connect support is possible so you can launch your Spotify listening session from the Spotify app. Snapcast also integrates into Home Assistant for ease of adding/removing clients from the active listening sessions (yes, plural. You can have multiple listening sessions from multiple audio input sources at once).

- snapcast - this server/client software will broadcast synchronized audio from any source to any snapcast client. The server and clients can all run on the same computer. Individual clients are assigned to their own usb soundcard. Source and volume for each client is controlled through the webpage and/or homeassistant.

I have a music library on my home server that I use mopidy to play via the iris plugin integrated into my home assistant UI. It plays over Snapcast which streams over the network to multiple devices in the home with independent volume control. I can fire up the Snapcast client in my phone to get it going there as well, which does work over vpn if I'm away, though I generally just fire up the files from my phones SD card for out-of-home listening. I recently started using whipper on Linux to extract audio from craigslist cds.

https://mopidy.com/

https://mopidy.com/ext/iris/

https://github.com/badaix/snapcast

https://github.com/whipper-team/whipper

You want SnapCast. You'd run snapserver on your Linux box and snapclient on your Pi's. Snapserver has support for Airplay (via shairport-sync) and Spotify (via librespot). I recommend using MPD for your music library, as I don't think PlexAmp can output audio in a way that's useful for snapserver.

Instead of splitting audio from the jack output you can create a Little network with a router and have more rpi streaming music over It using snapcastsnapcast .

If you're into the idea of an open-source client/server architecture, then there's Snapcast. If you have a few Raspberry Pis (one per output source) or any other box that can run Linux and has an audio out interface, then this is cheap and easy with an active Home Assistant integration too.

I'm setting up a whole home audio set up and looking for suggestions in the design. Currently, I run Snapcast on Raspberry pis connected to various soundbars or amps. It's working ok, but has issues (Mostly on some of the 2.4GHz WiFi Pis that can't keep up with the bandwidth).

For the multi-room audio, check out snapcast. If you like it, we could try to convince the Symfonium dev to add the ability to cast music directly to snapcast, which I would love to have!