General Improvements update!

Soo... As i said in my previous post, i am trying to train and fix a new ai that can be fully guided. Since that was taking a little while to do I decided to split the update. Here are some new features:

• Better soundfont!
• Semi-customizable instruments (note this wont save when outputting to midi file)
• (better?) large model (by default)
• "Randomness" slider
• Probably some random fixes i forgot about too