[API Request] AudioMuse-AI integration

[NeptuneHub]

Type of change

API extension

Proposal description

AudioMuse-AI is a back-end with minimal front-end that introduce Sonic Analysis OpenSource and free.

It is a contenirized application that dealing with Jellyfin, Navidrome and other subsonic like MediaServer API do:

1. Analyze songs: extract from the audio the main caracteristic
2. Clustering for automatic playlist creationg among the entire collection
3. Sonci fingerprint: it analyze the recent and most listened song, search for sonic similar song, and automatically create a Sonic fingerprint playlist
4. Song Similarity (also called Instant Mix): you input a song, and it return an entire list of sonic similar song
5. Song Path: you give two song and it create a sonic path of song between them

If the first 3 are more "server side functionality", the last two will be very helpful if exposed as API from the Mediaserver to the final client because are very interactive with the final user.

Backward compatibility impact

I think this are totally new endpoint. Till now I never find Sonci Analysis implemented on Sonic Mediaserver.

Backward compatibility

• No backward compatibility impact.

API details

Below some specific example of API call and response. Installing AudioMuse-AI you can also use the interactive swagger at:

• /apidocs/

Additional example of test can be also done by Installing AudioMuse-AI and just trying the integrated front-end OR by looking at how [audiomuse-ai-plugin]((https://github.com/NeptuneHub/audiomuse-ai-plugin) remap them also.

1 - /api/analysis/start [POST]

it start the analysis task. It can be run also without any mandatory parameter to use the defualt. The task_id returned is useful to cancel the task

Response:

{
    "status": "queued",
    "task_id": "76be1d12-274f-4d30-9ea8-37c6f5e69f7e",
    "task_type": "main_analysis"
}

2 - /api/clustering/start [POST]

it start the clustering task. It can be run also without any mandatory parameter to use the defualt. The task_id returned is useful to cancel the task

Response:

{
    "status": "queued",
    "task_id": "c11c828a-3b29-4f57-94aa-528111c58aa7",
    "task_type": "main_clustering"
}

2.5 - active_tasks/Cancel APi for Analysis and Cluster

/api/active_tasks [GET]
Give you a status update and a log for the task active. Is needed for Analysis and Clustering because they are async.

{
    "details": {
        "albums_skipped": 0,
        "albums_to_process": 52,
        "log": [
            "[2025-08-17 16:15:47] Launched: 43. Completed: 0/43. Active: 43. Skipped: 0/3210.",
            "[2025-08-17 16:15:47] Launched: 44. Completed: 0/44. Active: 44. Skipped: 0/3210.",
            "[2025-08-17 16:15:47] Launched: 45. Completed: 0/45. Active: 45. Skipped: 0/3210.",
            "[2025-08-17 16:15:47] Launched: 46. Completed: 0/46. Active: 46. Skipped: 0/3210.",
            "[2025-08-17 16:15:48] Launched: 47. Completed: 0/47. Active: 47. Skipped: 0/3210.",
            "[2025-08-17 16:15:48] Launched: 48. Completed: 0/48. Active: 48. Skipped: 0/3210.",
            "[2025-08-17 16:15:48] Launched: 49. Completed: 0/49. Active: 49. Skipped: 0/3210.",
            "[2025-08-17 16:15:53] Launched: 50. Completed: 0/50. Active: 50. Skipped: 0/3210.",
            "[2025-08-17 16:15:54] Launched: 51. Completed: 0/51. Active: 51. Skipped: 0/3210.",
            "[2025-08-17 16:15:55] Launched: 52. Completed: 0/52. Active: 52. Skipped: 0/3210."
        ],
        "log_storage_info": "Log in DB truncated to last 10 entries. Original length: 54.",
        "status_message": "Launched: 52. Completed: 0/52. Active: 52. Skipped: 0/3210."
    },
    "parent_task_id": null,
    "progress": 6,
    "running_time_seconds": 19.109625101089478,
    "status": "PROGRESS",
    "sub_type_identifier": null,
    "task_id": "76be1d12-274f-4d30-9ea8-37c6f5e69f7e",
    "task_type": "main_analysis"
}

/api/cancel/76be1d12-274f-4d30-9ea8-37c6f5e69f7e [POST]

It used for Analysis and Clustering task that are async to stop them

Response

{
    "cancelled_jobs_count": 5,
    "message": "Task 76be1d12-274f-4d30-9ea8-37c6f5e69f7e and its children cancellation initiated. 5 total jobs affected.",
    "task_id": "76be1d12-274f-4d30-9ea8-37c6f5e69f7e"
}

3 - /api/sonic_fingerprint [GET]

It create a list of sonic similar song to the more frequent played music. It actually need to know user and password/api token of the user.

/api/sonic_fingerprint/generate?n=200&navidrome_user=user&navidrome_password=password

Response:

[
    {
        "author": "PLACEHOLDER",
        "distance": 0.0,
        "item_id": "4e9099f9d5f828e01be7eca8dfed035b",
        "title": "PLACEHOLDER"
    },
    {
        "author": "PLACEHOLDER",
        "distance": 0.0,
        "item_id": "6e3f3ffd06a3a675b15ffda425c808c2",
        "title": "PLACEHOLDER"
    },
...
]

4 - /api/similar_tracks [GET]

It create a list of similar song of one song it. you can give n as number of similar song.

/api/similar_tracks?item_id=edbc325978d35662ec7420e78fad15cd&n=10&eliminate_duplicates=true

Response:

[
    {
        "author": "<AUTHOR>",
        "distance": 62.968772888183594,
        "item_id": "5df257555090cf5c6a578dc5fe7f1eb2",
        "title": "<TITLE>"
    },
    {
        "author": "<AUTHOR>",
        "distance": 79.81998443603516,
        "item_id": "2c342b1231439e530fbced867ac79d23",
        "title": "<TITLE>"
    },
    {
        "author": "<AUTHOR>",
        "distance": 81.96104431152344,
        "item_id": "0dd8e01b260c2fb04b472dafd12220c1",
        "title": "<TITLE>"
    },
    {
        "author": "<AUTHOR>",
        "distance": 84.10198974609375,
        "item_id": "f0821e180fc3b1b1656ed7eb6b71a800",
        "title": "<TITLE>"
    },
    {
        "author": "<AUTHOR>",
        "distance": 84.95748138427734,
        "item_id": "839373efcdb132f1ff5a6d8a1d487448",
        "title": "<TITLE>"
    },
    {
        "author": "<AUTHOR>",
        "distance": 86.48504638671875,
        "item_id": "e3553561cb756d3a94a8f8871b1928bc",
        "title": "<TITLE>"
    },
    {
        "author": "<AUTHOR>",
        "distance": 86.49391174316406,
        "item_id": "7df0f2e503285607f1f7ed3098c7917a",
        "title": "<TITLE>"
    },
    {
        "author": "<AUTHOR>",
        "distance": 86.72184753417969,
        "item_id": "068b17be9216a947d5722bc1de84a404",
        "title": "<TITLE>"
    },
    {
        "author": "<AUTHOR>",
        "distance": 88.31651306152344,
        "item_id": "eaf348350797eeb50d078456b6c5a9a7",
        "title": "<TITLE>"
    },
    {
        "author": "<AUTHOR>",
        "distance": 89.19606018066406,
        "item_id": "a9d5f706f6a74cdd8b9a1641e8c20f7c",
        "title": "<TITLE>"
    }
]

5 - /api/find_path [GET]

It return a path between two song id. You can write the max_steps or not. If not the default is 25.

http://audiomuse.192.168.3.131.nip.io:8087/api/find_path?start_song_id=f260106767eb26f4cd411f883e9fccd2&end_song_id=a40d380625abb801e33884ee5ebc66da&max_steps=10

Response:

{
    "path": [
        {
            "author": "<AUTHOR>",
            "energy": 0.2501956,
            "item_id": "f260106767eb26f4cd411f883e9fccd2",
            "key": "E",
            "mood_vector": "rock:0.419,indie:0.262,alternative:0.194,indie rock:0.145,alternative rock:0.130",
            "other_features": "danceable:0.49,aggressive:0.83,happy:0.47,party:0.21,relaxed:0.12,sad:0.64",
            "scale": "minor",
            "tempo": 110.29412,
            "title": "<TITLE>"
        },
        {
            "author": "<AUTHOR>",
            "energy": 0.19047475,
            "item_id": "ba6e6deb42b383e7a237b376cac0d979",
            "key": "A",
            "mood_vector": "rock:0.412,indie:0.252,alternative:0.195,punk:0.150,indie rock:0.141",
            "other_features": "danceable:0.54,aggressive:0.68,happy:0.67,party:0.44,relaxed:0.13,sad:0.39",
            "scale": "minor",
            "tempo": 133.92857,
            "title": "<TITLE>"
        },
        {
            "author": "<AUTHOR>",
            "energy": 0.21160908,
            "item_id": "f32c793d45291fb69da5ee99d18935f4",
            "key": "A",
            "mood_vector": "rock:0.433,indie:0.238,alternative:0.196,indie rock:0.147,alternative rock:0.133",
            "other_features": "danceable:0.68,aggressive:0.86,happy:0.68,party:0.40,relaxed:0.11,sad:0.36",
            "scale": "minor",
            "tempo": 133.92857,
            "title": "<TITLE>"
        },
        {
            "author": "<AUTHOR>",
            "energy": 0.23702899,
            "item_id": "3aa90697b5f9751b650e22acf2939d98",
            "key": "C#",
            "mood_vector": "rock:0.398,indie:0.223,alternative:0.173,punk:0.130,indie rock:0.120",
            "other_features": "danceable:0.73,aggressive:0.82,happy:0.67,party:0.39,relaxed:0.09,sad:0.34",
            "scale": "minor",
            "tempo": 144.23077,
            "title": "<TITLE>"
        },
        {
            "author": "<AUTHOR>",
            "energy": 0.26209036,
            "item_id": "fcd3d76e2ffbf380cf180530ba20fc95",
            "key": "C#",
            "mood_vector": "rock:0.360,indie:0.177,alternative:0.152,punk:0.108,alternative rock:0.100",
            "other_features": "danceable:0.43,aggressive:0.74,happy:0.55,party:0.14,relaxed:0.18,sad:0.51",
            "scale": "minor",
            "tempo": 117.1875,
            "title": "<TITLE>"
        },
        {
            "author": "<AUTHOR>",
            "energy": 0.24837823,
            "item_id": "14a26ee7f4da38abb2c43b3580fa2fda",
            "key": "F#",
            "mood_vector": "rock:0.430,alternative:0.165,indie:0.161,alternative rock:0.124,hard rock:0.108",
            "other_features": "danceable:0.44,aggressive:0.75,happy:0.56,party:0.25,relaxed:0.20,sad:0.58",
            "scale": "minor",
            "tempo": 110.29412,
            "title": "<TITLE>"
        },
        {
            "author": "<AUTHOR>",
            "energy": 0.24007338,
            "item_id": "192d5b03abdb9c687387e38b1c4e8a6a",
            "key": "D",
            "mood_vector": "rock:0.457,alternative:0.142,hard rock:0.130,indie:0.127,alternative rock:0.110",
            "other_features": "danceable:0.37,aggressive:0.90,happy:0.67,party:0.29,relaxed:0.04,sad:0.36",
            "scale": "minor",
            "tempo": 133.92857,
            "title": "<TITLE>"
        },
        {
            "author": "<AUTHOR>",
            "energy": 0.20871356,
            "item_id": "575954d3fe7bb9da0650184a6138bb33",
            "key": "D#",
            "mood_vector": "rock:0.438,hard rock:0.168,alternative:0.134,punk:0.114,indie:0.106",
            "other_features": "danceable:0.36,aggressive:0.92,happy:0.66,party:0.37,relaxed:0.06,sad:0.32",
            "scale": "minor",
            "tempo": 125.0,
            "title": "<TITLE>"
        },
        {
            "author": "<AUTHOR>",
            "energy": 0.23377606,
            "item_id": "99417829924d70ee809e92e2b24a20c9",
            "key": "F#",
            "mood_vector": "rock:0.406,punk:0.145,alternative:0.141,indie:0.115,alternative rock:0.109",
            "other_features": "danceable:0.27,aggressive:0.85,happy:0.62,party:0.13,relaxed:0.10,sad:0.48",
            "scale": "minor",
            "tempo": 104.166664,
            "title": "<TITLE>"
        },
        {
            "author": "<AUTHOR>",
            "energy": 0.1704033,
            "item_id": "a40d380625abb801e33884ee5ebc66da",
            "key": "C#",
            "mood_vector": "rock:0.365,punk:0.207,alternative:0.116,indie:0.088,hard rock:0.080",
            "other_features": "danceable:0.41,aggressive:0.92,happy:0.45,party:0.11,relaxed:0.06,sad:0.19",
            "scale": "minor",
            "tempo": 104.166664,
            "title": "<TITLE>"
        }
    ],
    "total_distance": 70.79050350189209
}

Security impacts

The only security impact that I see is the Sonic Fingerprint that pass user and password in query string. If this functionality is seen as a security issue can be skipped or alternative you can give suggestion for a better implementation. But I would like to avoid that this single functionality become a blocker for the entire API implementation.

Potential issues

I don't see any potential issue, but if you see I'm open to work on it to solve. Thanks.

Alternative solutions

Actually AudioMuse have his minimal front-end. But this is more for testing or initial adoption. For a better user experiance will be better if this API are exposed from the Subsonic API server, and then implemented directly in the front-end that play the music.

[NeptuneHub]

I also attached the swagger file: AUDIOMUSE_API.JSON that you can look on the swagger online editor

There is multiple not-needed API just focus on the one described above.

[epoupon]

LMS already has a built-in recommendation engine.
Currently, it only relies on tags, so that's the only source used to calculate similarities between songs. This is quite limiting but still works reasonably well if everything is properly tagged.

Previously, it relied on low-level data provided by AcousticBrainz (a collaborative project based on extractors from the Essentia library, which unfortunately was discontinued a few years ago). Using this data, a SOM (Self-Organizing Map) was built to group songs based on acoustic similarity.

The recommendation engine then used this map to find the closest songs starting from a given one (and it was also possible to find a more or less short path between two songs).

All of this is to say that it’s a topic I’d like to revisit for LMS, namely, managing the extraction of low-level features from songs, in order to rebuild that SOM and reconnect the recommendation engine to it.

I see that the API offered by this AudioMuse solution seems to propose something quite similar. For OpenSubsonic, we’d just need to retain the high-level concepts and let each server handle its own implementation. And here, it’s indeed the server itself that should provide the data (even if internally it relies on a third-party service like AudioMuse-AI)

[NeptuneHub]

Audiomuse-AI work with Librosa and Tensorflow. Not only get tag, energy and tempo, but also extract for each song a 200 size vector based on Musicnn embbeding. Is from this vector that the most useful functionality works.

It started by using Essentia-Tensorflow but then switched to Librosa AND Tensorflow because it work also on ARM and is more updated.

Anyway I agree, if we agree on the API signature then each server could then implement it how they want. Off course I would like is with AudioMuse-AI.

[Tolriq]

For Open Source (OS)

Open Subsonic ;)

IMO the necessary public endpoints are more or less the same as Plex.

1. get similar sonic results for a single song,album or artists (with a limit in difference value and count)
2. get a sonic path between 2 songs to enable nice automatic djs between random songs.

[epoupon]

Open Subsonic ;)

Ahah, I asked an AI to check my english and I didn't notice it replaced OS by this x)

get similar sonic results for a single song,album or artists (with a limit in difference value and count)

About that, what is the unit of this difference limit? I noticed it is really difficult to have a meaningful scale when computing distances between songs.

[Tolriq]

This is usually server dependent and scaled to a 0..1 range.

[NeptuneHub]

In case of Audiomuse-AI for similarity:

• we give an API that search for song similarity. Than it can be used form the media server also to compose Album or Artist similarity. For example on the Jellyfin Plugin we do that by getting some “song seed” from album or Artist and then search for similar to that (by the end you can analyze a song. An album or an artist are only a “collection” of song);
• The AudioMuse-AI API don’t work on maximum distance, instead on number of song. In this way is more predictable the result.

[Tolriq]

There's should be a maximum distance. If I only have a single classical music track and ask for similar. If there's none this is OK. It should not return AC/DC tracks just because we asked for 10 tracks.

[NeptuneHub]

So are we going to add at least this new API for sonic similarity and song path to open subsonic API? Do you need some extra information from me?

I would really like to look at Music Server like LMS to implement it by talking with AudioMuse-AI.

[Tolriq]

Public API changes takes times and it's still the holidays for most :)

[NeptuneHub]

Understood, thanks ! I was just worried that something for me was still needed and you all was waiting for me :)

[Th3K1ngP1n]

Revisiting this request to ask the devs if this integration is planned for near future? I really want to get back to using Navidrome with OpenSubsonic, but, this killer feature is still pending. ;)

[epoupon]

I am interested in working on this (so to prepare a PR to expose the endpoints), but I can't tell when

[Th3K1ngP1n]

@epoupon Exciting to hear that. I think many in the community are all ears on this integration.

[NeptuneHub]

Thanks eupopon for this sharing! I avoided to re-ask to don’t be too pushy, but very glad to read this. If need any kind of support to better integrate AudioMuse-AI please let me know!

[MoriarT3a]

Some time has passed again and I'm eagerly awaiting a way to connect Navidrome with AudioMuse-AI. In my opinion, this is exactly the right step to move this forward.

Are there any current updates on the development status?

[Th3K1ngP1n]

I am also eagerly waiting but dont want to push to hard for development 😃

[MoriarT3a]

Yes, I didn't want that either. :-) On the other hand, I always think if only 5 people speak up about what 100 people want, that gives a false impression of the "urgency."
Although urgency also sounds quite pushy again....
I hope you understand. ;-)

[epoupon]

As a reminder, it is planned to add two endpoints to the OpenSubsonic API (maybe more later):

• Retrieve similar sonic results for a single song, album, or artist (the legacy Subsonic API only exposes similar artists, originally tied to Last.fm recommendations).
• Retrieve a sonic path between two songs.

These endpoints are intended for client use, for example to allow a client to leverage the server’s recommendation engine to generate dynamic playlists or similar features.
The server may use any mechanism it wants to build recommendations; audiomuse-ai would be one possible option among others.

[NeptuneHub]

Hi All,
I just want to share that Navidrome just implemented getSimilarSongs2on song, so you can click on a song and select Instant Mix in their develop container version (and soon also in the latest stable!).

Thanks to this and to Navidrome plugin I was able to implement the AudioMsue-AI Navidrome plugin here:

• https://github.com/NeptuneHub/AudioMuse-AI-NV-plugin

I'm also looking forward for @epoupon implementation on LMS!

[MoriarT3a]

Oh, wow. That was to obvious. :-D
Yes the advantage of audiomuse-ai is, to have a real simliar track and not a random track from a simliar Artist, what can result to first heavy-metal power play to second acoustic version of an lovesong from a heavy metal band....

Okay, okay. I understand. 😇

Then we need a own endpoint for audiomuse-ai like discribed. 😅

[NeptuneHub]

For me ANY solutions is ok as soon as it is a solution.
In August 2025 I asked to agree on a specific API to get similar song and the only move forward was from Navidrome, that just extended getsimliarsongs2 to accept as input not only artist but also song.

So why just agree that this API can accept both Artist and Song ID and then return similar song ? At least is a solution.

[Tolriq]

Navidrome that is full part of OpenSubsonic and never answered here ;)

Your current implementation, works for Navidrome and whoever want to use it. It is not something OpenSubsonic related.

The API validation for such features is tied to servers approval more than clients, LMS still had not time to work on that, Navidrome have skipped the post and no reaction from other servers.

You already know that I support AudioMuse AI when it's possible to do that in a proper way. This is the proper way.

[NeptuneHub]

I totally appreciate the effort that every one of you is doing, and I want to get the opportunity to say thanks to everyone of you for the time that you're dedicating to this topic.

I want to highlight that if I'm writing here is exactly to get a standard agreement between the different Music Server, so that then become also easy the integration on the different frontend.

My final goals is standardization, because I want Sonic Analysis free, open and accessible to much user possible.

For the specific case of Navidrome I took this opportunity to develop the AudioMuse-AI plugin and override the API that it is already using. If this is the best option or there is better one, I don't know. I only try to share the information that I have hope this could help to move forward :)

If there is something else that I can do to help, please let me know. Thanks!

[deluan]

There are some misunderstanding in this thread regarding OG getSimilarSongs(2) and what I've done in Navidrome.

The behaviour of allowing any ids in getSimilarSongs(2) was implemented in Navidrome way before OpenSubsonic was created, in October 2020. This was done because, based on OG Subsonic docs, getSimilarSongs accepts any ids. Even in OpenSubsonic it is documented like this: id is "The artist, album or song ID."

As in Navidrome both endpoints share the same implementation, Navidrome always allowed song or album ids in getSimilarSongs2, but was never documented.

I recently implemented "Instant Mix" in Navidrome's UI, that leverages the fact that ND's getSimilarSongs2 accepts songIds. And as part of the plugin system, I added extension points to allow plugins to override the response of these endpoints by different IDs. But this was an internal implementation, the focus was the Navidrome UI, not Subsonic clients. For that I'm still following this thread.

As I said before in other proposals, I'm glad to implement whatever the group gets consensus on, just don't have time to participate actively in all these discussions.

[dweymouth]

OK, attempting to build consensus here before writing up an extension PR. How about adding an OS extension for getSimilarSongs2 that adds a new songId parameter? Since the spec says the existing id param only accepts artist IDs

[Tolriq]

I have not fully thought about it, but I would have done it as search3 with different top arrays and limits for it.

That covers all needs and can do all in one for display purpose too.

But yes no polymorphism ever :)

[paulijar]

I would also add support for songId and albumId to getSimilarSongs2 but agree that this is completely separate discussion from AudioMuse. The extended getSimilarSongs2 is something which servers already using Last.fm could implement very easily, while the new AudioMuse-like API would actually need integration to AudioMuse (or similar) and a lot more work. From my own perspective I can say that my server could support the extended getSimilarSongs2 almost immediately upon the possible API spec change while the new more powerful API could maybe be implemented in the distant future, at best.

Of course the method getSimilarSongs2 is limited but that's also its strength: This means it can be implemented with the services provided by Last.fm (or similar). An API with more control and options cannot.

[Tolriq]

getSimilarsongs can't returns albums hence the new endpoint discussed above.

[dweymouth]

No one is saying that an extension for getSimilarSongs2 would remove the need for additional endpoints to support smart queues anchored by two tracks, or supporting a maxDistance param, or requesting similar albums. This would just be to correct a deficiency in the original API and bring getSimilarSongs2 up to parity with its predecessor, which according to the spec should be viewed as deprecated for clients/servers that organize with ID3 tags rather than folder structure.

[Tolriq]

I don't talk about sonic here, just about an extension for an old endpoint.

The correct question would be as it was done before, what's the status of the current servers in OS. If all servers who implement the endpoint support the other ids then we do a clarification.

I really do not think that an extension on an old endpoint that would be better used as an extension to cover that and other related needs is more logical in the scope of a public API. The goal is not to have clients checking 3 extensions to cover the similar (not sonic) calls.

Pretty sure we can already go clarification BTW ;)

Edit: And in the end this is completely unrelated to this post ;)

[dweymouth]

@opensubsonic/servers As Tolriq points out, if all current member servers support track and album Ids as arguments to getSimilarSongs2 (as is supported by getSimilarSongs according to the original spec), OR will commit to doing so, then we can just formalize a clarification rather than an extension to formalize this behavior (at the time when the latest release of all member servers support this). If we can formalize this, it allows clients that don't use getSimilarSongs to get access to song mixes powered by Last.fm, or whatever similarity backend the server uses.

Status for getSimilarSongs2

Server	Supports artistID (required)	Supports songID	Supports albumID	Notes
Navidrome	✅	✅	✅	Since the original release
LMS	✅	✅	✅	ref
Gonic	✅	❌	❌	ref
Ampache
Astiga	✅	✅	✅
Nextcloud Music	✅	✅	✅
Supysonic	❌ ?	❌	❌	doesn't support endpoint at all?
QM-Music

[Tolriq]

What I mean what they do from the ID. If all servers internally do the same query for both endpoints then everyone can just use the getSimilarSongs that works on all servers already.

[dweymouth]

But the spec says getSimilarSongs should not be used if organizing by ID3 tags. The alternative clarification could to be to formally say that OpenSubsonic servers support calling getSimilarSongs even if the client and/or server organize primarily by ID3 tags.

[Tolriq]

When I added support at that time and checked all servers had the getSimilarSongs working with ID3 stuff as expected else I would not have added this.
Some servers did not implement the end point that's why it's being a setting on Symfonium side.

[dweymouth]

OK, sounds like we are good to go with a clarification stating that "For OpenSubsonic servers, getSimilarSongs can be used even if organizing by ID3 tags". Then clients can use the openSubsonic=true flag in the server responses to choose which endpoint to call (if they care about doing that).

[gravelld]

Astiga supports albums, artists and songs, expressed as IDs, as inputs to getSimilarSongs and getSimilarSongs2. So, three ticks.

[NeptuneHub]

I don’t understand if finally an agreement was reached and if yes what I need to do to be supported from the different music server and client on top.

Multiple people ask for me the support of Symfonium on top of Navidrome and my understanding was that he is waiting for an agreement here and probably same action from my side.

[NeptuneHub]

On my side the last 2, song similarity and song path are the one that add more value if integrated directly in a frontend. This because they have much value going to alter the song queue more than creating a playlist.

Analysis, Clustering and Sonic Fingerprint can be used from the integrated audiomuse-ai frontend and/or scheduled as a cron job.

But yes the last 2 are the more impotent from my point of view. Song similarities I was able to implement in the Navidrome plugin, for the song path I wasn’t able.

[epoupon]

I think we already said that only the last 2 endpoints of the original proposal are meaningful for the OS API. The others are more about controlling internal implementation details, and we don’t want the OS API to be tied to a particular solution.

For LMS, I am currently working on restoring the acoustic similarity work, but it will take a while before it becomes operational (and it does not use AudioMuse-AI, it is just another implementation)

[deluan]

I think we already said that only the last 2 endpoints of the original proposal are meaningful for the OS API

Ops, missed that (TL;DR 😅). In that case, @NeptuneHub, go ahead and create the PR. Once we have the API definitions, I can implement them in Navidrome and create the extension points for your plugin to implement it.

[NeptuneHub]

PR where? is there any example that I can follow? :)
(I mean a PR here in the opensubsonic repository I suppose, but is there an already merged one that I can get as example to do it in the right shape?)

[deluan]

Yes, PR to this repository, to document the endpoints. Lots of examples in the Pull requests section.

[NeptuneHub]

PR opened for similarTracks and findPath#219