4 ms·Masked Audio Generation Using a Single Non-Autoregressive Transformer1 points by mvoodarla 3y ago