affiliation not provided to SSRN
Watermark, Speech Generation, Low-Rank Adaptation, Vocoder, Codec
Fake audio detection, dataset, noise condition.