Hi @brian
Thanks for the clarification. I am very impressed that you are using your own algorithm. Even for NLU and NLP nothing like what is already available on the market at AWS, Google Cloud Platform or Azure? Very nice!
I understand by what you are saying that the choice of the algorithm is technical, not business. And thanks for clarifying the point on the cut of the TIDAL subscription.
So that means because it is your own model and that you own the technology, we could see in the future this algorithm applied to local library, right?
I am asking this question because maybe I am confused but I have “a feeling” that the Radio function lacks some “randomness”. I have experienced that today again with 1.6 and I am glad you ran some tests with a “local library”. I just wonder why your tests were inconclusive as many AI models have now features to avoid or mitigate such possible bias. If you willing to share, would you like to tell us more about it?
I can see the value of Roon specifically with the screen in mind of “This artist–>collaborated with that one—> and that one is inspired by that other one” so here is the basis of the algorithm, no?
I know this is hypothetical and not real life but I wonder what is your opinion on this “lack of randomness” in Roon.