Description

MODEL . I made the same model three different times. Do I care? Nope. Trained by me. No credit needed. Thumbnail art also by me. Inference audio for the demos was made with SynthV, using the ANRI Lite and Teto Lite voicebanks respectively. EXTRA NOTES/OBSERVATIONS/WHATEVER: This version of the model is *slightly* better with high notes than the previous versions, but it still kinda sounds like nails on a chalkboard if it goes too high. The index is kinda wonky, so I recommend using a different index file, or turning the index feature ratio down to somewhere between 0.35 and 0.55.