VALL-E 2: Neural Codec Language Models are Human Parity Zero-Shot Text to Speech Synthesizers

Speech and Audio Models report from Microsoft with 9 connected researchers in the LLMpeople atlas.