Singingfish.com's spider sees the original website and declares it open season on the sound files. The extractor then blindly hits those sound files without checking the robots.txt file of the hosting domain.
I sent them an email. In the meantime, I am banning their extractor:
ADDED: Got a reply to my email. They do not seem concerned that their extractor ignores robots.txt. Instead, they said they would add my domain to their exclusion list and run a script to remove our files from their db. Not what I was hoping for...