ADIFF: Explaining audio difference using natural language Soham Deshmukh, Shuo Han, Rita Singh, Bhiksha Raj

Two new datasets were created; a prefix-tuning baseline and ADIFF, which uses a cross-projection module and position captioning, were compared; ADIFF showed significant improvements via objective and human evaluation.

Great opportunity to work with amazing set of people!

Hi @jonathanleroux.bsky.social, could you please add me to the list as well? Thank you in advance!