MSU-Bench is a new four-tier, speaker-centric benchmark for spoken language understanding in multi-talker conversations, and early results show audio language models decline as tier complexity increases.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
eess.AS 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
MSU-Bench: Towards Understanding the Conversational Multi-talker Scenarios
MSU-Bench is a new four-tier, speaker-centric benchmark for spoken language understanding in multi-talker conversations, and early results show audio language models decline as tier complexity increases.