There isn't a whole lot of information on steganography in the modern age, from what I can tell. A google search for steganography usually brings up a list of very impractical methods (e.g. tattooing a message on a guy's head then waiting for hair to grow over it). The only modern (and practical) method I've seen involves manipulating the least significant bits of an image. This method appears everywhere when you search for steganography. Are there any other practical methods of steganography?
The International Journal of Soft Computing and Engineering has published many articles on the subject.
There are plenty of tools available for hiding data in compressed images and audio (e.g., Steghide, JPHide and Outguess). However, it is often possible to detect when they have been used (e.g., using neural networks).
I devised a (presumably non-conventional) scheme which re-formats an input text stream like emails a little bit such that the number of words in a line of the output file mod 2 gives the stego bit in the line. The stego bit rate is thus admittedly rather low, but it could under circumstances (e.g. when the realm of discourse is highly limited such that a few numerical codes from a code book used by the communication partners suffice to form the stego message to be transmitted) be useful IMHO. See http://s13.zetaboards.com/Crypto/topic/6939954/1/
I have another (presumably non-conventional) scheme for stego with texts as cover, which is based on a (secret) partitioning of the alphabet into 3 sets that is agreed upon by the communication partners. If a given stego bit is 0, the sender attempts to write a sentence that begins with a charater in the first set. If that bit is 1, he attempts to write a sentence that begins with a character in the 2nd set. In case he has difficulty with these two rules, he writes a sentence that begins with a character in the 3rd set. The receiver can easily recover the stego bits from the message of the sender accordingly. See http://s13.zetaboards.com/Crypto/topic/7338098/1/