Implement Navidrome Audiomuse AI plugin to Symfonium

[TheKingPin]

Feature description:

@Neptunehub has just released the plugin for Navidrome. It is still in the early stages and I have not tested it myself yet.

This feature request is for the integration of the plugin features into Symfonium, similar to how the jellyfin plugin already works.

Problem solved:

Makes Audiomuse available for Navidrome in Symfonium. Something, many of us have been waiting for.

Brought benefits:

Smart queue track plays for Navidrome, and, also others if the plugin in future opens up other API features from Audiomuse, getsimilarartists etc…

Other application solutions:

I dont think there are any at the moment.

Additional description and context:

See this post please for further details on the plugin:

Option to use the server-side implementation for creating instant-mix playlists

Hi all, I’m back to say that the AudioMuse-AI plugin for Navidrome is ready: It implement the similarity on Song (InstantMix button) and Artist (Radio button) both on getSimilarSongs2 api (I think is easier if you install navidrome and do a try if you want to look the type of call). It is actually supported on Navidrome develop container but should be supported in stable in days. Will be nice to have the support for navidrome on Symfonium too.

** Im just making the feature request and take no credibility for the actual development of the plugin. Hopefully NeptuneHub can assist with any questions you might have.

Screenshots / Mockup:

[TheKingPin]

@neptunehub Here is a dedicated feature request so we can track the process

[Tolriq]

They just replace a random endpoint This is not a proper usable API for anything like smart flows.

Symfonium already use getsimilarsongs for radiomixes that is working on a lot more servers than getsimilarsongs2 that was randomly chosen here.

As soon as the proper OpenSubsonic API is validated and implemented by a server it will be supported.

[deluan]

Just to be clear, the AudioMuse plugin works for both getSimilarSongs and getSimilarSongs2.

So if you are supporting getSimilarSongs already, it should work, right?

[Tolriq]

If they enable the option in the provider settings then it will work for radio mixes yes.

But not for the more polished smart queues and smart flows.

[TheKingPin]

@Tolriq

But not for the more polished smart queues and smart flows.

And I think I speak for more than just myself. That smart queues is what we really desire

[Chandandharana]

The navidrome plugin still has to implement the endpoints used for the smartflows and queues that the jellyfin plugin provides, so that ball isn’t in the symfonium’s park yet. I for one am looking forward to when the plugin achieves parity so that I can get away from jellyfin’s awkward music handling

[TheKingPin]

Agreed.

@Chandandharana

get away from jellyfin’s awkward music handling

I would add Jellyfins inferior way of handling music.

I reverted back to Navidrome after the plugin release.. Mind you, I still have a Jellyfin / Audiomuse container parked on my server if I want some smart queues.

[deluan]

The ball is actually in OpenSubsonic’s park. Symfonium talks to Navidrome via OpenSubsonic API. Unitl the API is not defined, there’s no point in implementing more AM integration, if clients won’t be able to use it.

[TheKingPin]

And for those interested. OpenSubsonic feature request can be tracked here:

[API Request] AudioMuse-AI integration · opensubsonic/open-subsonic-api ·...

[Tolriq]

You have not answered there about that part We solve the getSimilarSong stuff but there’s no point in me building the specs if no server is willing to implement and validate what it want to support.

The API is relatively clear as already in Plex and Jellyfin. But I don’t know how you intend to build that on your side.

On Jellyfin side, there’s actually no Jellyfin API for that, there’s a plugin API where I can request if audiomuse is installed and the plugin add endpoints inside Jellyfin that I can call.

I think a proper sonicApi is better API for all other servers, but this is eventually a solution to just add a plugin endpoint that expose plugins installed in servers since you now have the concept of plugins (but don’t know the scope of what they can do)

[deluan]

@Tolriq

You have not answered there about that part We solve the getSimilarSong stuff but there’s no point in me building the specs if no server is willing to implement and validate what it want to support.

Catch-22: There’s is no point on me planning the implementation if I don’t know the spec

@Tolriq

The API is relatively clear as already in Plex and Jellyfin. But I don’t know how you intend to build that on your side.

Not sure what do you mean by “how you intend to build”. Once we have the spec, I’ll implement it.
If you want to know the details, it is simple: By default the endpoints will return not_implemented or empty responses, as I don’t plan to add this functionality directly in Navidrome. I’ll introduce the plugin extension point for AudioMuse-AI (and others) to provide the service.

@Tolriq

I think a proper sonicApi is better API for all other servers

I completely agree with that. I plan to allow plugins to add new endpoints as well, but I think this should be specified in the OpenSubsonic API

[Tolriq]

@deluan

Catch-22: There’s is no point on me planning the implementation if I don’t know the spec

Well maybe [API Request] AudioMuse-AI integration · opensubsonic/open-subsonic-api · Discussion #172 · GitHub was not clear enough that I’m waiting for confirmation of servers willing to implement to not work in vain

Same as transcoding where I waited for LMS to work on it and polish the proposal toward something that works best for servers.

I have the API client side from other servers and it’s mostly fully described in the OP post, that does not means that’s what server prefer to implement.

This is even more true if you intend to delegate to plugins and so may require additional data or return less data in the answers. Like size constraints in dialog between plugins and navidrome and plugins or whatever could be

@deluan

By default the endpoints will return not_implemented or empty responses, as I don’t plan to add this functionality directly in Navidrome. I’ll introduce the plugin extension point for AudioMuse-AI (and others) to provide the service.

Would probably be better to not publish the OS extension if you have no plugin to handle it. That avoid network calls for no reasons.

[106wasted]

If it means anything, when I find a song and hit ‘Radio Mix’, it seems to query my Audiomuse-AI plugin in Navidrome, so it looks like there is some functionality right now.

[TheKingPin]

It has been working for quite some time. But as mentioned before:

@TheKingPin

And I think I speak for more than just myself. That smart queues is what we really desire

These queues are way more accurate. With current radio mix, more often than not you will find some tracks just dont belong, for example it can go from a metal track to something totally different.

[beggi]

Hey @Tolriq and @deluan I saw that @neptunehub PR was merged

adding similarTracks and findPath new api for AudioMuse-AI sonic simiarity (#219)

What are the next step to get the features on navidrome and Symfonium?

Really looking forward to it:)

[TheKingPin]

I have also been watching that PR like a hawk. Super excited to see it being merged. I think the day it goes live on OpenSubsonic I will let out an almighty internal cheer…. Actually on second though it will probably be very much audible. Thanks to everyone involved for their hard work and dedication.

[beggi]

I gess as I understand it OpenSubsonic only defines the official API, how I understand it is that when the API changes are live, the servers like navidrome still have to implement it. Is that right @deluan ?

[TheKingPin]

Navidrome already implement the opensubsonic api. Thats how we connect to it. The PR is to extend the api with access to new Audiomuse endpoints. Navidrome and symfonium probably just need to add the new endpoints.

[beggi]

@deluan and @neptunehub and @Tolriq

Is there any update on this topic?