That was my move--You stole my move!
Well, I thought I had a few tricks to myself, but I'm not the only one working on ways to keep youtube.com from compressing a sample so much to destroy it (link). This is one of the problems I was having while working on my StegoFS/watermarking media project--how to make things survive conversion.
For example: Here we have a video clip that was uploaded and downloaded to youtube 4 times. Notice how the audio quality (distortion, with its own wavelike crescendo/decrescendo) goes from this to this. The FLV bounces around at 370KB depending on how many times you feed it back in to youtube's interface.
So, with the audio, adding a sine wave tricks youtube into leaving the audio alone, but the sine wave is still removed from the final compressed upload. This is one way to be sure that any data encoded as audio gets through unmolested (from my point of view, my stego has better fu than it used to).
Now, what can we do with video . . . Still working on a few tricks there myself, hopefully I'll get the remaining, nagging, lurking, festering details rooted out this week so they'll be reading for my DefCon 16 presentation.
For example: Here we have a video clip that was uploaded and downloaded to youtube 4 times. Notice how the audio quality (distortion, with its own wavelike crescendo/decrescendo) goes from this to this. The FLV bounces around at 370KB depending on how many times you feed it back in to youtube's interface.
So, with the audio, adding a sine wave tricks youtube into leaving the audio alone, but the sine wave is still removed from the final compressed upload. This is one way to be sure that any data encoded as audio gets through unmolested (from my point of view, my stego has better fu than it used to).
Now, what can we do with video . . . Still working on a few tricks there myself, hopefully I'll get the remaining, nagging, lurking, festering details rooted out this week so they'll be reading for my DefCon 16 presentation.