TY  - RPRT
TI  - How Reliable Are Predicted MOS for Reproducing Human System-Level Preferences in Speech Enhancement?
AU  - Nahomi Kusunoki
AU  - Tsubasa Ochiai
AU  - Naohiro Tawara
AU  - Marc Delcroix
AU  - Naoyuki Kamo
AU  - Tetsuji Ogawa
AU  - Shoko Araki
PY  - 2026
UR  - https://arxiv.org/abs/2609.39032
ID  - 2609.39032
ER  -