[BETA] Ear Wax - NOW AVAILABLE - Testers Wanted

[ginjaninja 616]

2 hours ago, MarvinB said:

@ginjaninja

Hi, I havent heard anything from you lately, so I am checking in to see how testing is going and to update you on a few changes..

1) I have finally figured out how to shuffle a playlist from the beginning consistently.

2) and random playback is even more random before.

 

Give the above changes a listen 

Above all, send me your one off/edge cases, MAYBE just MAYBE, they can be addressed

Hi @MarvinB

That (improved random functionality) feels much better to me.

For me the next most important thing is easing the ability to play tracks.

I think the skill is onto something asking the user to re speak if not heard properly, it gives me hope of something more but so far that  just keeps the skill alive (which is promising and useful in itself). I wonder if the desired results can gradually be homed in on?

The chance of getting the skillname, trackname and artist name and keywords all heard properly is ~10%.
if the skill responded to track keyword / trackname with "which artist?/which Album?" and the user could specify "Artist None" (to remove artist filter) or "Artist Elton john" or "Album Honky Château" that might greatly improve recognition.

You said Alexa has to return something to user within X seconds, Could a holding song or message be returned whilst the skill did more heuristic matching when not sure and then interupt the holding session.

"Ask Alexa play X by Artist Y"

"No confident match found, searching....." to give the skill a bit more time?

 

If the skill could prioritise tracks where the artist is a member of album artists then you are more likely to get the best default version.

 

Other than that my next priority on wishlist would be  anything that can be done to improve recognition accuracy first time success. Make us of Musicbrainz search hints for odd sounding/spelt names https://musicbrainz.org/artist/7c20f558-d664-42bd-b6f1-81acf9dbb72d/aliases.

Could popular artist names and tracknames feed into an accuracy improving engine?

probably well past Alexa's capability/architecture but you did ask...:-)