| As a carrier medium for concealed messages abc is pretty poor. | The tunes are too small, and contain too little redundant information.
Yeah; I think you're right. This was, really, why I made the rather silly suggestion a few messages ago. It's also the basis of my suggestion that HTML might be a better medium. If you look at the original intent with HTML, it wouldn't be too good, either, since the intent was a sparse, unobtrusive markup that doesn't much affect readability. But now we see a lot of machine-generated HTML that is mostly markup and is unreadable by mere humans. Hiding extra junk in such a junk pile is pretty easy. If we want to do this with plain-text music, we'll probably have to wait for the XML-based encodings. These will probably end up a lot like the current crop of "garbage" HTML documents, entirely machine generated and unreadable, with a thousand or so bytes used per note. In such an encoding, it will be easy to embed a spurious tags that don't effect the musical content. It will just look like the unreadable junk that you find in commercial HTML. And the files will be sufficiently large that you can add a lot of extra stuff without anyone becoming suspicious. | abc is simply too good at doing what it was designed for to be used for | anything else. This is probably true of most compact, efficient encodings. I've seen the observation that JPEG isn't a very reliable format for steganography. The explanation is that JPEG is designed to make it easy to "simplify" a picture by shrinking the scale and/or lowering the resolution. This is done (conceptually) by selectively ironing out low-order bits in adjacent pixels that are similar. The actual algorithm is complex, but very efficient for a computer. This tends to remove any data that may be encoded in low-order bits. One of the reasons that some web sites use JPEG a lot is that they need only store a single high-res image, and smaller (especially "thumbnail") images can be efficiently generated on the fly. If there is network congestion, some routers (especially those that cache) can now recognize JPEG images and automatically decrease their resolution by some N%, which most end users won't notice until N gets large. There is so much software on the Net now that does this that there can be serious problems getting steganographic data through with JPEG. We've already seen this with ABC. There have been tunes posted that came from a MIDI-to-ABC translator, and the MIDI was from a recording that gave lengths like 345/987 or 1252/4706. This is horrible, of course, though it's how real musicians play. The natural reaction is to reduce such lengths to simple fractions. This is just lowering the resolution, and most MIDI software can do it automatically. By doing this, you eliminate data hidden in the low-order bits of the lengths. Musicians do this all the time, often knowingly, with the idea that anyone who knows a style will know how to play the correct note lengths. Thus, it's common to write hornpipes and shottishes with even notes, for readability, with the expectation that anyone who recognizes these rhythmic terms will know how to play them. I wonder if there any known cases of musicians encoding messages in the fine details of how they play? This is done with song lyrics all the time, of course, mostly by using metaphor. But I don't think I've read of it being done with the music itself. To subscribe/unsubscribe, point your browser to: http://www.tullochgorm.com/lists.html
