Decoding Order Matters in Autoregressive Speech Synthesis
It is shown that randomness in decoding order affects speech quality, and by interpolating between identity and random permutations, it is found that fixed-order decoding is suboptimal, while adaptive decoding yields better performance.