1 paper · 1 filter
Zekai Liu, Zhilin Wang, Xuzheng He +2
Text-to-music systems produce increasingly convincing audio, yet evaluation reveals little about whether the result matches user intent. A global text-audio relevance score can ove…