#11
|
|||
|
|||
The more you use Roon the better Roon Radio gets at learning the kind of music you like. When I first started using Roon Radio it felt like their picks were random and weren’t a good match for my taste in music. After a while, however, Roon Radio as learned what I like and does a much better job of choosing songs I like. If I had to guess, I’d say Roon Radio was selecting songs I liked only about 25% of the time initially. Now, it’s closer to 90%.
|
#12
|
|||
|
|||
PS: it helps if you use the thumbs up/down feature when you like or don’t like a Roon Radio selection.
|
#13
|
||||
|
||||
It’s the algorithm I don’t like. Inevitable, but one of my ways to resist digitization (not a Luddite - I just like the personal music connection).
|
#14
|
|||
|
|||
Quote:
+1 |
#15
|
|||
|
|||
Let me put this another way.
The algorithm learns your musical taste based on your feedback (thumbs up/down and in the case of a thumbs down, the reason why you gave it a thumbs down). It also uses the feedback from other users who are similar to you. From that, it chooses selections that match your feedback. You may not like the way the algorithm works or the relatively slow learning process but the logic behind the algorithm is sound. As mentioned earlier, it takes time (weeks/months) and effort on the part of users to provide the thumbs up/down feedback. Another approach to building an algorithm is to scan your existing music library and use that information to learn your preferences. However, that method creates potential issues with data/network security. |
#16
|
||||
|
||||
I just like an fm DJ picking out music (or me discovering it via this forum, magazines, etc). I work with algorithms all day and don’t want them in my music collection. This is, admittedly, inherent and somewhat illogical bias on my part.
|
Thread Tools | |
Display Modes | |
|
|
Audio Aficionado Sponsors | |