TY  - RPRT
TI  - An Ultra-Low Latency, End-to-End Streaming Speech Synthesis Architecture via Block-Wise Generation and Depth-Wise Codec Decoding
AU  - Tianhui Su
AU  - Tien-Ping Tan
AU  - Salima Mdhaffar
AU  - Yannick Estève
AU  - Aghilas Sini
PY  - 2026
UR  - https://arxiv.org/abs/2604.12438
ID  - 2604.12438
ER  -