With so few parameters, I imagine a voice fine-tune might be readily tractable.