In paper WavMark: Watermarking for Audio Generation we proposed the WavMark model, which enables encoding 32 bits of information into 1-second audio. In this tool, we take the first 16 bits as a fixed ...