一作ACL 2026Findings· Findings of ACL
D i TR educio: A Training-Free Acceleration for D i T -Based TTS via Progressive Calibration
Yanru Huo · Ziyue Jiang · Zuoli Tang · Qingyang Hong · Zhou Zhao
摘要
While Diffusion Transformers (DiT) have advanced non-autoregressive (NAR) speech synthesis, their high computational demands remain an obvious limitation. Existing DiT-based text-to-speech (TTS) model acceleration approaches predominantly focus on reducing sampling steps through distillation tech...