Mango Madness
18th September 2002, 06:21
I don't know much about anything other than how to make perfect xvid/ogm dvd backups. But in thinking about how stuff is compressed, adaptive or variable methods usually produce better results. It's my assumption from the research i've done that Vorbis and mp3 use a single high pass filter and low pass filter and the encoder itself encodes everything inbetween. I know that's probably a simplified view. I was thinking about when the encoder sees a frame of music instead of cutting the stuff above and below the thresholds it looks at the bits and then applied an adaptive filter that better responds to the audio present. Like if you have a scene of cannons you'll need to preserve more bass than say an opera which would require a higher high pass threshold. It would add complexity to the whole process because it would probably require multiple passes to obtain correct information and execution. It's easy to say let's cut everything below this and everything above that then run the encoder to take whatever is left and produce the final file. The difficulty i see is teaching the algorithm to produce superior results. YOu'd need some kind of scoring system or something, i dunno. But again, this is just a thought I had and I'm not sure if its implemented, in the works, or whatever. So I just thought i'd throw it out there.