Open haomjc opened 2 years ago
Not currently, the goal of this model is to physically simulate sound, but AI singing is very difficult to achieve by simulation because it is very difficult to simulate the vibration of the vocal cords and the resonance of the head cavity. In contrast, a purely data-driven approach is more suitable. such as Wavenet.
Can this model simulate human sound? Human voice is complicated but interesting, AI drawing is popular right now but AI singing is not developed so far.