TY  - RPRT
TI  - SPEARBench: A Benchmark for Naturalness Evaluation in Streaming Speech-to-Speech Language Models
AU  - Thomas Thebaud
AU  - Yuzhe Wang
AU  - Hao Zhang
AU  - Sathvik Manikantan Napa Ugandhar
AU  - Ashish Hallur
AU  - Georgi Tinchev
AU  - Venkatesh Ravichandran
AU  - Laureano Moro-Velazquez
PY  - 2026
UR  - https://arxiv.org/abs/2607.05365
ID  - 2607.05365
ER  -