Great stuff, thanks for showing us this.
I took a liberty here for my own learning and let me know if I am out
of line.
I took your wave and did a mp3 conversion the way I normally would do
it. I would greatly appreciate feedback from all on if my method is
based on wrong assumptions and if the result is good or bad.
I took the wave file of 3.3 meg and saw your mp3 at less than 10% or
around 200 kb.
I first did a filter based on a fft mic rumble filter cutting the
freqs below 110hz. I did this by the feeling that I wanted mp3 to
maximize its compression in this area so it would give me less
compression where I cared more about.
I converted the sound to mono as there was not a huge amount of
stereo dependent sound here and this would save me 50%.
I noted that the highest gain was still around -20 db. Because I
don't want mp3 to treat the -20 like the -30 I amplified the sound 14
db.
I then applied a constant depth mp3 conversion based on 32khz
bandwidth as I didn't value the md sound on this recording between 15
khz and 20 khz so therefore wanted mp3 to maximize its compression up
there. I set the final mp3 at 96 kbps mono.
my example for learning purposes only is temporarily at:
http://home.comcast.net/~richpeet/alt017.mp3
My result is roughly the same size as your mp3.
Which is better and is my thought stream a proper approach?
Rich Peet
> By the way : This is the very first time I hear such a BIG
difference
> between a mp3 and the original wav (08-017.mp3 and 08-O17.wav). I
find the
> loss of quality really surprising.
>
>
> [Non-text portions of this message have been removed]
________________________________________________________________________
________________________________________________________________________
|