Embedded signals and water marking...

Has anyone out there played much with encoding or decoding embedded signals in audio files? Watermarking is a common concept i'm aware of. I'm looking to find some ways of embedding some small bytes of data into an audio signal then recover the data. would like the data to be upto 1 second of precision accurate and be able to survive most audio encoding and conversion. This will typically be used to signal events or actions to take place. It would also be nice if we could embed metadata like lyrics directly into the audio as it plays.

I know water marking is a common practice, but not only do we need to embed data we need some, playback accuracy on the data like: at 00:50, signal 'Event 1"

Right now i'm interested in what people know about these concepts...

Does BASS have anything to offer on these topics?