lmiguelgato/audiowmark — reverse-engineered prompt

Build me a simple audio watermarking tool. I want to give it a WAV file and a 128 bit hex message, then get back a new WAV file that sounds basically the same but has that message hidden inside it.

It should also be able to read a watermarked audio file later and print the hidden message without needing the original file. The watermark should be reasonably tough, so it can usually survive normal MP3 or OGG conversion at decent quality. Add support for a private key file so only someone with the key can add or read the watermark. Also let me choose the watermark strength, with a sensible default.

Please make the commands easy, like add, get, and gen key. For reading, show the decoded message, confidence style scores, and optionally save machine readable JSON. Include basic examples and tests with sample audio so I can verify that adding and reading a watermark works.