Testing and Feedback - Jellyfin Skiil

This skill plays music from your Jellyfin server. It is a fork of the emby skill.

How to install jellyfin-skill

How to setup and test jellyfin-skill

  • Configure the Skill Settings in home.mycroft.ai
  • You can generate a API key from the settings in your jellyfin server
  • “play song Poison” will use the common play framework
  • “play song Poison from Jellyfin” will be played using the jellyfin-skill
  • You can play artists and ablums too
  • “play artist blackmore’s night”
  • “Play album Alive”

Where feedback on jellyfin-skill should be directed

General feedback and questons can be placed in this thread.
For feature requests and issues; please use github https://github.com/tuxfoo/jellyfin-skill

1 Like

The interesting thing would be to be able to tap into the Live TV M3U Tuner.

Hi there @tuxfoo
I have managed to get my Mycroft to connect with Jellyfin server, it is able to find files and tells me they’re “playing”, but no music plays.
I checked in the Mycroft CLI and this is definitely the case.

not sure what’s up with it and I don’t really understand enough of how it works

@tuxfoo
I wonder on which jellyfin version this was tested.

Set up a new server (with latest jellyfin 10.7.7) and while it connects it can’t request media (API Doc) under
/Search/Hints?SearchTerm=...

request = 400 Bad Request


nevertheless i think it would be beneficial if instead of a (rudimentary) client we take advantage of a Python Client coded from a member of the Jellyfin team
(i’ve seen he uses a whole other api call to search for media)

… and maybe this way we can expand on the features provided

Did you set up an API key? I vaguely recall if you don’t that’s the error that comes up. Create an API key from a browser => Jellyfin URL => Home => Advanced => API Keys. Then Click the ‘+’ sign to add a new key. Copy that key to your Jellyfin skill in Account.

I’m doing a technical doc on a music playing stack, but the first draft is not quite ready. Watch this space …

-Mike M

Still running the emby skill but this is (besides renaming and a few updates) essentially the same. There the api key had to be added manually, but that is not the problem.

But are you saying you run the latest jellyfin (server) with this working?


Yet, regardless i see bigger problems coming up as the codebase of emby and jellyfin diverges. Sooner rather than later the shift should be made, especially since the linked api-client originates from a jellyfin dev who has oversight about backwards compatibility and recent developments. Otherwise we’re playing catchup.

i’ve setup a dev branch and will see what is needed

1 Like

Yes, I have one reference box that will play music on command using jellyfin, with some restrictions:

"Play <artist|album|song>" works most of the time, but curiously "Play song|track <song>" and "Play artist <artist>" do not work. 

I need to dive down in the code, but too many other things are on the TODO list right now.

What I like about Jellyfin over Emby is that it does not require a password like Emby (of course you could argue the API key is a password), and it seems to be more free and open.

-Mike M

the emphasis is on latest (stable) here. (10.7.7 in my case)
Just tried it with jellyfin skill but the same, as expected.

Got it working, yet still the biggest hurdle (in my book) is when - and this is mostly the case - the server not working with fuzzy matching. I guess its less a problem in english speaking countries but somehow i can’t force myself to hear german music :stuck_out_tongue:

Things like Elephant (White Stripes) becomes something like “elefant” and will miss in an album search.

Thanks for the feedback guys.

I have not used Mycroft in a while, I still use Jellyfin a lot.
I just ordered Respeaker 4 mic array hat for raspberry pi so that might motivate me to do some more work on the Jellyfin skill once it arrives.

With the integration of jellyfins semiofficial python client api (originates from KODI) i will separate the skill from Ricky Phewitts’ work (Emby-skill) as i explore the possibilities that goes beyond “voice integration”. Just to inform about the pathway.

:bulb: Get item by letter and then fuzzy matching with mycroft should work

The artist search is a bit tricky.

To give an example:
I have 2 nirvana albums, none is found (API) with mediaType MusicArtist and only one is found with AlbumArtist. Have to dig into that.

On top, the code is searching the wrong id (not the artist id) so you wouldn’t end up with a correct list anyway.

I guess the same applies to a song search. Album and playlist searches are straight forward in comparison.

@SGee - thanks for the reply. … “The artist search is a bit tricky” … so we can fix that … eventually…

-Mike M

Fix is in the making … The artist id has to be filtered out erlier in the search process.
I rewrite the search process to be more articulative/communicative about the options.

Say you’re searching an artist. if not specified further you will get different types back. Mycroft will ask if different types are found, so the user may choose.

album search is written already with a choice from which position/title you want the album to play from

@SGee - Excellent! Let me know when you update the code - I’ll give it a try.

-Mike

@SGee - any progress on this?