In the scheme of generating virtual voices, there are many things one can do – I am more interested in seeing if anyone has tried to create anything using multispeaker embeddings and then choosing a space in the matrix, or played with WaveRNN and accents.
