Decoding Order Matters in Autoregressive Speech Synthesis
Quick summary
arXiv:2601.08450v2 Announce Type: replace-cross Abstract: Autoregressive speech synthesis often adopts a left-to-right order, yet generation order is a modelling choice. We investigate decoding order through masked diffusion framework, which progressively unmasks positions and allows arbitrary decoding orders during training and inference. By interpolating between identity and random permutations, we show that randomness in decoding order affects speech quality. We further compare fixed strategies, such as \texttt{l2r} and \texttt{r2l} with adaptive ones, such as Top-$K$, finding that fixed-or
Key takeaways
- arXiv:2601.08450v2 Announce Type: replace-cross Abstract: Autoregressive speech synthesis often adopts a left-to-right order, yet generation order is a modelling choice.
- We investigate decoding order through masked diffusion framework, which progressively unmasks positions and allows arbitrary decoding orders during training and inference.
- By interpolating between identity and random permutations, we show that randomness in decoding order affects speech quality.
Why it matters
“Decoding Order Matters in Autoregressive Speech Synthesis” illustrates how changes in the AI ecosystem can affect products, workflows and user expectations together. Its lasting significance depends on measurable adoption, cost and safety outcomes.
Member comments