Thanks for reminding me about Gone to Earth. It was the soundtrack to my uni days and one of those serendipitous miracles that don't seem to crop up much anymore. Maybe I'm just too out of the loop. I wish I could get Suno to create that kind of chill, ambient stuff. There's probably some way of training it, but I think I'll wait till it improves to the point of actually doing what I tell it.
The reason Suno can't do that is because there isn't a large enough data set to accurately reflect the style. There's just that double album and nothing else like it exists. It's the same with Mahavishnu Orchestra stuff or stuff that sounds like Yes or King Crimson. There's just not enough for it work up a passable variant.
It's really good at doing 1960s-1970s Funk and R&B because the dataset there is huge.
Thanks for reminding me about Gone to Earth. It was the soundtrack to my uni days and one of those serendipitous miracles that don't seem to crop up much anymore. Maybe I'm just too out of the loop. I wish I could get Suno to create that kind of chill, ambient stuff. There's probably some way of training it, but I think I'll wait till it improves to the point of actually doing what I tell it.
The reason Suno can't do that is because there isn't a large enough data set to accurately reflect the style. There's just that double album and nothing else like it exists. It's the same with Mahavishnu Orchestra stuff or stuff that sounds like Yes or King Crimson. There's just not enough for it work up a passable variant.
It's really good at doing 1960s-1970s Funk and R&B because the dataset there is huge.
I guess it will get there one day.
I suspect not. There's just not enough good suff around to use as a data set.