Yeah I've been steadily uploading what I've made from V3 (which I think I've only got about 3 songs left to upload from V3) but lately I've began working with V3.5 as it gives some good results if done right. I've noticed that if you do extend from the main clip more than once it seems to mess up but if you get two clips, the main clip and the extend clip, and hit get full song then continue off of that and do the same method after every extend, it seems to help it not mess up.
@@twistedperceptions My issue was that it seems to repeat my outros twice and then cuts off in the middle of it. (I guess it's trying to give me a full 4 minutes.) I've just taken to reediting it in Audacity. Sometimes you just have to experiment.
Gotcha. Yeah I generally end up setting the extend from time to where I feel it would do best from and try to get outros. But honestly since 3.5 released I've not had much problems with it ending the song for me most of the time. May just be luck that I've been getting it that way to be honest but it'd be nice if it is actually working the way it was intended. If this was actually intended to do that. lol And yeah experimenting with the models is the best we can do as it helps the AI learn and get better with each new version. n_n
Have you tried v 3.5? I like the fact that you can get longer songs, but it seems to mess up when you extend. At least it does for me.
Yeah I've been steadily uploading what I've made from V3 (which I think I've only got about 3 songs left to upload from V3) but lately I've began working with V3.5 as it gives some good results if done right. I've noticed that if you do extend from the main clip more than once it seems to mess up but if you get two clips, the main clip and the extend clip, and hit get full song then continue off of that and do the same method after every extend, it seems to help it not mess up.
@@twistedperceptions My issue was that it seems to repeat my outros twice and then cuts off in the middle of it. (I guess it's trying to give me a full 4 minutes.) I've just taken to reediting it in Audacity. Sometimes you just have to experiment.
Gotcha. Yeah I generally end up setting the extend from time to where I feel it would do best from and try to get outros. But honestly since 3.5 released I've not had much problems with it ending the song for me most of the time. May just be luck that I've been getting it that way to be honest but it'd be nice if it is actually working the way it was intended. If this was actually intended to do that. lol And yeah experimenting with the models is the best we can do as it helps the AI learn and get better with each new version. n_n