While creating UTAU voicebanks, I observed that human voices maintain timbre during pitch shifts, whereas synthetic ones distort. Inspired by topology, I hypothesized an acoustic topology invariance in vocals: when pitch (F0) changes, harmonics and formants self-adapt to preserve timbral "shape".
Aldof101/WaveTopo
Folders and files
| Name | Name | Last commit date | ||
|---|---|---|---|---|