Does anyone else want or need this?

[MarvinB 3]

Hello,

I’ve spent years managing my own media servers and home setup, and after Plex changed direction and Alexa support became unreliable, I moved fully into Emby and started building a better voice-control solution for my own library.

I’ve been building a custom Alexa skill for my personal Emby music library because I got tired of the limitations with standard Alexa music control and the lack of good voice support for private libraries.

What I wanted was simple in theory:

“Alexa, ask Ear Wax to play The Police”

and have Alexa play music directly from my Emby library on Echo devices — artists, albums, playlists, genres, shuffle, continuous playback, and “what is playing” support.

In reality, it turned into a much larger project involving custom intents, queue management, playback recovery, persistent state, and Emby API integration.

At this point, the core system works reliably:

• artist / album / playlist / genre playback

• queue refill and continuous playback

• playback recovery

• “what is playing”

• stable two-step voice commands through Alexa

The only remaining annoyance is reliable one-sentence invocation versus the current two-step flow.

Before I spend time turning this into a proper beta with account linking and user setup, I wanted to ask:

Would other Emby users actually want something like this?

Specifically:

• Echo device playback from your own Emby music library

• voice control without relying on Amazon Music

• support for artists, albums, playlists, genres, and random mixes

Would you use it, or is this just a personal obsession that escaped containment?

 

Thank you,

 

Marvin

[ginjaninja 606]

On 16/05/2026 at 13:47, MarvinB said:

Hello,

I’ve spent years managing my own media servers and home setup, and after Plex changed direction and Alexa support became unreliable, I moved fully into Emby and started building a better voice-control solution for my own library.

I’ve been building a custom Alexa skill for my personal Emby music library because I got tired of the limitations with standard Alexa music control and the lack of good voice support for private libraries.

What I wanted was simple in theory:

“Alexa, ask Ear Wax to play The Police”

and have Alexa play music directly from my Emby library on Echo devices — artists, albums, playlists, genres, shuffle, continuous playback, and “what is playing” support.

In reality, it turned into a much larger project involving custom intents, queue management, playback recovery, persistent state, and Emby API integration.

At this point, the core system works reliably:

• artist / album / playlist / genre playback

• queue refill and continuous playback

• playback recovery

• “what is playing”

• stable two-step voice commands through Alexa

The only remaining annoyance is reliable one-sentence invocation versus the current two-step flow.

Before I spend time turning this into a proper beta with account linking and user setup, I wanted to ask:

Would other Emby users actually want something like this?

Specifically:

• Echo device playback from your own Emby music library

• voice control without relying on Amazon Music

• support for artists, albums, playlists, genres, and random mixes

Would you use it, or is this just a personal obsession that escaped containment?

 

Thank you,

 

Marvin

hi @MarvinBi would be very interested in an alternative Alexa skill to encourage progress by way of friendly competition; I would particularly welcome

• a 'what is playing" function,
• more reliable skill recognition and command invocation.

 

 

[MarvinB 3]

Hi @ginjaninja;

Thank you for the encouragement, I'm at the 95% completion for my beta. I'm still working on the single sentence invocation, that really bothers me. I've reached out to Alexa development support team for a possible solution. If I don't hear back from them, I am going to go with my gut and take a calculated guess.

So far, no bugs, plays what I ask for.

The what's playing function is solid, reports current track title and the number in the queue and the size of the queue.. ex; You are listening to {song title} from the album {album name} from your selection {Rock}, tune ### of ### is playing. Playback recovered from ### stream errors, ### retries, ### skipped tunes.  

Ear Wax attempts to stream audio to Alexa speakers as is. If the stream encounters an error, it will transcode the file to mp3 320 and stream it. if that fails, it will skip that file and stream the next file/tune in the queue. Most of my files are Flac and MP3. The commands are straight forward, after you hear  the Ear Wax reply what would you like to hear?  You would say something like; Play artist {artist name}, Play album{album name}, Play some music... etc.. Play some music is a random mix.

I am interested in a few beta testers to test this on their own emby server

[ginjaninja 606]

I would be happy to test your plugin. I find the current Alexa skill only reliable enough as a technical demonstration rather than a daily driver which is frictionless to use.

 

In case it helps, here are a few issues the current skill has found over the years.

• Special characters causing playback fails (now fixed)
• large playlists not playable (now fixed  - limited to 1st 50 tracks with shuffle support) - would be great if 50 limit was overcome.
• No way to filter [out/in] audiobooks from search results, i dont think the Emby schema/api can tell the difference and i dont think the search engine supports search by library so this may not be possible. Although i think the items endpoint supports 'path' so that might be a way to limit music searches to music libraries (and avoid audio books for people who have these contents in separate paths)
• unnecessary verbage in using the skill via a two stage skill/command, every time i use the current skill it talks a whole paragraph to me. The 10th time ive used the skill i know i can ask it 'to play a few songs' and i dont need telling every time.
• Not preferring tracks, in search results, where both the artist is a match [must] and album artist is a match [should], the api does support 'artisttype' so might be possible. reason: A song from a studio album would be preferred over one from a compilation. The search api doesn't natively support this intelligence in 1 call but i think its doable via multiple.
• Inability to keep the skill active/at forefront, so issuing subsequent commands is easier (possibly a limitation of alexa ecosystem more than anything else)
• Seamless specification of output device / remembering preferred output device per device. What device do you want to play from (mostly fixed now with 'change player' command).
• [Unreported] Echo device reporting 'media not supported'. With audio files (which are supported) and video files (search results should be limited to what the device supports imo). somethings up here with the current skill which i haven't got  a handle on.
• More support for logging and general troubleshooting of connectivity issues, where is the breakdown...what was heard? What api calls were made? connectivity to server?

Suggestions for "what is playing" type functionality. (both initial playback announcement and specific invocation)

• Artist (when present) metadata should be announced in preference to albumartist for a track IMO. reason Albumartist can be a lot less relevant on a compilation but artist is always [more] relevant.
• if the playback request is a playlist (+album?Book?) ie content the user is likely 'expecting'  then there is no need to announce what is playing for 1st track of a playlist as a matter of course: reason i know what i asked for. Indeed as a whole announcing what is playing (as a result of play command) is perhaps only necessary if what is returned might not be expected. eg the result of a track search. Certainly if the skills isnt confident the requestor is getting what was expected then its useful to announce as a sensecheck against computer fails..
• My primary use of ad hoc "what is playing" would be to know Track Name, Artist name, Album name. reason ive asked for some random unexpected stuff and i dont know what is currently playing. Or i am playing a compilation album but i dont recognise the track/artist. I wonder wether the required results of "what is playing" is different depending on what has been requested and the nature of what is play [in relation to the request] eg whether the content is from a studio album or not. For an album [requested], both album artist [when different] and artist might be nice, track #x of #y might be more useful.

 

looking forward to seeing what you produce.

[AgostinoMedia 4]

I would be interested. Personally what I miss is "Random Album Radio" which I would use with Plex. Also, not sure if this is the right place to mention it, but I wish Emby's shuffle was more random. Right now the way I play music is "tell M.B. Home to play some music" and it shuffles the songs in my library. but just not random enough to my liking.