Home of the Squeezebox™ & Transporter® network music players.
Results 1 to 10 of 1186

Threaded View

  1. #29
    Senior Member philchillbill's Avatar
    Join Date
    Jan 2019
    Location
    The Netherlands
    Posts
    938

    [Announce] 'MediaServer' *certified* Alexa skill for LMS

    Name:  listFav.png
Views: 163
Size:  21.9 KB


    Version 5.0 of MediaServer is now live and automatically deployed when you use the skill. The Amazon app will not list it as updated as there were no changes to the voice model — however, there are many, many changes to the code:

    * First off, most spoken responses are quite a bit shorter. For example, if you ask to play something that targets the assumed player, the response no longer mentions "... on the bla bla player" since that's pretty implicit. Playback/stream responses now usually start with a verb, and superfluous information is skipped (e.g. if an album name includes the artist's name — Bob Seger's Greatest Hits — there's not much point in saying that it's "... by Bob Seger".

    * It can take anything from 300ms to 3 seconds to start up the /stream.mp3 network client in LMS. That is a blocking delay because the playlist cannot be queued until the player is created. The spoken response from Alexa (which also blocks your audio from starting because it has to finish before the music starts) is now sent asynchronously so that you hear the speech while the player is still being created in parallel. This makes streaming much snappier.

    * The ListFavorites command(s) now show the favorite number and the associated dotted-ID (in addition to the favorite-name) on Echos with screens. In the screenshot above, the favorite number is in orange, the dotted notation is in blue, while unplayable items such as folders are dimmed with a red strikethru over the notation.

    * There's a new sorting rule for favorites (still based on the default skin), whereby nested folders are exhaustively extracted before popping back up a level to the next item there. This provides consistency up to 1000 items per nested folder and makes the numbering 100% predictable. When counting favorites manually, as soon as you see a folder, go into it and keep counting everything there (and below!) before moving back upwards in the hierarchy. It'll be obvious on your Echo's screen.

    * Asking to play a song by name now first queues the oldest match in your library with the shortest name, which is usually the original. Say you have 10 versions of 'Valerie' by Steve Winwood, where 5 are called just 'Valerie', 2 are called 'Valerie (Radio Edit)', one is called 'Valerie (Extended Mix)', and 2 are called 'Valerie (Live)'. The original 'Valerie' from Talking Back to the Night will play first. Next, one each of the Live, Extended and Radio Edit will be randomly selected and those 3 will be auto-shuffled. Meaning that 4 tracks will be queued. So the version you are probably after will always play first rather than a Live or Remix variant. Player shuffle being active will override this as the original-first-and-then-Fisher-Yates-shuffle is actually done by the skill and not LMS (it queues an array of track_id's).

    * Play/Stream song-index-from-album now only queues the track in question, rather than first loading the whole album it's from and jumping to that track. This results in artwork on Echos while streaming because that makes sense in the context of a single track.

    * Play/stream playlist commands have the same asynchronous speech as mentioned above for network clients, as they can also take noticeably longer before LMS processes them.

    * Playback commands that match many artists (e.g. play something by 'Bob' while forgetting to add 'Dylan' or 'Seger') will now show all the matched artist names on your Echo's screen to help you narrow things down.

    * Album/Song mismatches between requested/actual artist will now display results with coverart on your screen to help guide your choice. If you ask to play the album 'Abacab' by 'The Beatles' you'll see the retrieved cover art and metadata in addition to hearing that the matched-artist is actually Genesis. This is deliberately shown with a smaller image than the now-playing screen to avoid confusion, but becomes now-playing size if you choose to accept the correction.

    * Now-playing queries for online services Spotify, Deezer, Qobuz and Tidal are now supported.

    * The background image for the Now-Playing screen has been changed to provide better legibility of textual metadata against the backdrop.

    * Fuzzy matching of artists has been noticeably improved due to making even more round-trips to LMS if needed.

    * Many commands make more use of the screen than they used to. Also, text is more responsive to the size and shape/aspect-ratio of your Echo's display.

    * Any artwork displayed is now set to best_fit rather than best_fill so non-square images won't look weird (thanks Staresy).

    Enjoy!
    Last edited by philchillbill; 2021-11-27 at 14:23.

Tags for this Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •