Alexa reply: "Playing music from provider XXX is not supported"

[warren-bank]

Hi. This project looks great!

My issue is probably/hopefully pilot error.. I'm new to Alexa and skills.

Everything is deployed and appears to be working properly:

• Firebase database..
  • with pre-populated data:

    * collection = "alexa"
    - document = "responses"
      * fields:
        - (array of strings) launch       = ["Resuming"]
        - (array of strings) now_playing  = ["You are tuned in to"]
        - (array of strings) next_playing = ["Next up is", "next,", "up next,"]
        - (array of strings) stop_playing = ["Catch you later!", "Ciao!"]

  • with updated access rules:

    rules_version = '2';
    service cloud.firestore {
    match /databases/{database}/documents {
      match /{document=**} {
        allow read, write: if true;
      }
    }
    }

• Vercel web app..
  • cloned from this commit
  • with read/write access to Firebase

• Alexa skills..

  • code:
  • uses a modified copy of this zip file
  • changes:
    • file: lambda/index.js
    • line: 6
      const BASE_URL = "https://alexaradio.vercel.app";
    • change: use base URL for my clone of the Vercel web app

  • notes:

    Code:
    * click:  "Import Code"
      choose: "AlexaRadioSkill.zip"
    * click:  "Deploy"
    
    Build > Interfaces:
    * enable checkbox: AudioPlayer interface
    * click: "Save Interfaces"
    
    Build > Interaction Model > Intents:
    * enable:
        AMAZON.CancelIntent
        AMAZON.HelpIntent
        AMAZON.StopIntent
        AMAZON.NavigateHomeIntent
        AMAZON.FallbackIntent
        AMAZON.PauseIntent
        AMAZON.ResumeIntent
    * add custom Intent:
      - name = PlayMediaIntent
      - slots:
        * name = SEARCH_TERM
          type = AMAZON.RadioChannel
    * click: "Save Model"
    
    Build > Interaction Model > Intents > PlayMediaIntent:
    * sample utterances:
        "to play {SEARCH_TERM}"
        "for {SEARCH_TERM}"
        "play {SEARCH_TERM}"
        "play some {SEARCH_TERM}"
        "read me {SEARCH_TERM}"
    * click: "Save Model"
    
    Build > Invokations > Skill Invocation Name:
    * Skill Invocation Name = "warrens juke box"
    * click: "Save Model"
    * click: "Build Model"

notes:

• the status of the skill is: In Dev
  • Code: Promote to live
    was never clicked
  • I only want this skill available to Alexa devices logged into my Amazon account

issue:

• when I say:
  Alexa, ask Warrens Juke Box to play [name of channel]

• the response is:
  playing music from provider Warrens Juke Box is not supported

• ---

What was my mistake?.. What installation step did I miss?

---

Thanks again for all this hard work! I'm looking forward to being able to configure my own .m3u playlists.. without needing to depend on any 3rd-party service.

Besides music, this is going to work great for accessing free audiobooks from Librivox.. which appear to:

• host all files on archive.org
• follow a convention where each chapter is a separate .mp3 with an .m3u playlist

[warren-bank]

update:

• I figured out the problem.. it's an error in your lambda code
• lambda/index.js

  let track = await getRemoteData("?search="+slotValue)
  // code will access fields:
  track.speech
  track.channel
  track.name

• the JSON response is structured:

  {
    track: {channel: "", name: ""},
    speech: ""
  }

• the corrected code would look more like:

  const data = await getRemoteData("?search="+slotValue)
  const track = data.track
  // code will access fields:
  data.speech
  track.channel
  track.name

[warren-bank]

wouldn't be a bad idea to:

• add a new branch: lambda
• unzip the content of AlexaRadioSkill.zip into this new branch

that way:

• there's more transparency into what a user will be installing and running
• you can track changes
• you can accept PRs

[balsimpson]

Great idea. Will keep the lambda function in the same branch as an unzipped JS file.

[balsimpson]

    const data = await getRemoteData("?search="+slotValue)
    const track = data.track
    // code will access fields:
    data.speech
    track.channel
    track.name

have not worked on this code for a few months. will take a look in the weekend.

[warren-bank]

no worries..

I just forked a copy.. and will play around with it tonight.

Feel free to merge.. or cherry pick any changes you might like.

[warren-bank]

this commit fixes that problem..

now, when I run a test in the Alexa Simulator:

• it successfully processes the request, and answers correctly..
  

  the JSON response looks perfect..

  ```js { "body": { "version": "1.0", "response": { "outputSpeech": { "type": "SSML", "ssml": "You are tuned in to Jazz" }, "directives": [ { "type": "AudioPlayer.Play", "playBehavior": "REPLACE_ALL", "audioItem": { "stream": { "token": "Music::Jazz", "url": "https://cdn.pixabay.com/download/audio/2023/02/24/audio_3daa1e8650.mp3", "offsetInMilliseconds": 0 } } } ], "type": "_DEFAULT_RESPONSE" }, "sessionAttributes": {}, "userAgent": "ask-node/2.12.1 Node/v12.22.12" } } ```
• a toast displays the notification:

  Unsupported Directive
  AudioPlayer is currently an unsupported namespace. Check the device log for more information.

update: when I test on an Echo Dot (5th gen).. no change: playing music from provider Warrens Juke Box is not supported

update: the discussion in this issue says that the toast notification is expected behavior, because audio won't play in the browser.. but that doesn't explain why it won't run on the Echo Dot. TBD..

[warren-bank]

wow.. just wow

unless I'm missing something.. developing for Alexa is a nightmare; I'm been (metaphorically) banging my head against a wall all day.. trial and error.. no debugger.. even reading logs is slow..

worst of all, nothing that should be working.. is working.

well, all of the built-in intents are working.. can "open" the app and issue "pause", "resume", "stop", "next", etc..

but, most importantly.. the custom intent is never passed to the skill. using utterances that are verbatim to what's in the model.. no intent ever hits the "canHandle" test phase. instead, even though I always prefix with "Alexa ask XXX".. Alexa always sends me elsewhere: Google results, Amazon music, Pandora, ...

without the ability to specify a keyword.. to include in the API request that searches for a matching channel or station.. the app always launches on the same channel/station/offset where it left off when stopped.. without any ability to change channel.

it's still better than nothing.. I could live with a single channel (ie: playlist) that can be managed through a web UI. but.. it would be so much better.. if

I've done a ton of refactoring.. and importing the skill directly from the git repo works much better than a zip file.. but, none of it fixes the issue that's driving me nuts.. the damn thing just refuses to work

if I can't fix it.. I'll probably just cause a channel to loop back to the 1st station when it reaches the end of its playlist. Implementing "previous" is also (probably) low hanging fruit.

note to self..

• investigate how "offset" works when stopping and resuming a "station" that is an m3u playlist.. both ones with internal duration tags, and ones without; this feature may need to be conditionally disabled (ie: for m3u, store "offset" value 0 in firebase for MRU entry).
• when an m3u is playing..
  • is there an API to:
  • query whether the current "station" is a playlist
  • query current position within the playlist
  • navigate previous/next within this playlist
  • if so, then "next" and "previous" intents could:
  • determine whether the action should be applied to the current m3u playlist
  • if not.. then change to a different station (within the current channel)

[balsimpson]

I completely understand your frustrations. debugging is so damn difficult, sometime i've abandoned projects midway and started fresh, because i just couldn't figure what the error was. if you are up to it, one of these days we could do a share/pair programming session and try and sort it out.

[warren-bank]

the thing that is driving me most nuts.. is that the search query (custom intent with slot) works in the simulator.. the music doesn't play, but the API request runs and the spoken response is appropriate, but on a real device, (logs indicate that) it never runs the skill

..after all this, I'm still getting:
Playing music from provider XXX is not supported

[HammyHavoc]

@warren-bank, did you ever get anywhere with this?

[warren-bank]

@HammyHavoc , I don't think I did.. but I honestly can't even remember. I spent about a month doing a deep dive into writing Alexa skills (a little over a year ago).. and then never touched it again. Memory is a blank. Sorry.

[HammyHavoc]

@warren-bank, no prob! :- ) It does seem like writing Alexa skills is weirdly obtuse for what should otherwise be quite an intuitive and straight-forward process. Laborious and time-consuming versus the likes of fiddling with FPGAs and DSP lol.

Hope you're well!