It’s someone who hasn’t seen the dozens of features added by free firmware for many years on the SC5000 clearly
2 Likes
That show no sign of stopping as far as we can tell.
2 Likes
Just as a FYI for everyone. The heave compute part of the stems is in the front end analysis. While it would be possible for the analysis to be done on the deck, if it did, it would be painfully slow since more then likely there would not be native support for the instruction set the software would want to use. But it still could be done. You can run a LLM off a CPU, but the inferencing is just really slow. With that being said, the solution would probably be that all the analysis would need to be done before it got to the deck. The real question is, can the decks support the compute needed to read the files container and process each layer individually. I can’t say how much it could or could not handle, but they could allways do a limited solution if they don’t have the resources to do process all the layers individually at once. In a limited compute situation, they could allways just say hey, we can only process two layers with one layer being just the vocals and the second being everything else, however the cost of doing so is that you loose the ability to use the second layer on the SC5000 deck(since they only have the compute resources to handle two tracks at once). Its not entirely possible, however, it might be take a complete re-write of the sc5000 firmware due to existing dependencies and thus might be too costly for the company to implement.