Search papers, labs, and topics across Lattice.
1
0
3
Current audio-language models are culturally tone-deaf, failing to grasp nuances of Persian poetry and music despite strong performance on transcribed text.