#10 How about 32-bit, maybe even more?


Well, the whole point of this project is that the higher
resolution decoding increases quality, correct? Unless
there's something more complicated to it than that, why
not also support higher quality decoding? First,
supposedly even the 24-bit is better quality when
dithered downward to 16, so I imagine 32-bit wouldn't be
so bad dithered downward to 24. Second, there are a
few soundcards that can play back audio at 32-bits
(such as mine) which would likely allow some people to
get even better quality yet. It occured to me that if it
weren't better quality when dithered to 24-bit, then it
might still work well if the decoder automatically used the
appropriate amount based on which setting was chosen.
Eg, if the user chooses 24, decode at 24, if the user
chooses 32, decode at 32. Really I imagine the main
thing holding such a thing back is the extra CPU power
that would take. Most people these days have better
PCs than my mere athlon 1800+, and I believe mine could
easily handle that with quite a lot of room to spare.

I know this isn't terribly important, but it seems to me like
it couldn't exactly hurt anything to try and might actually
improve things.


  • Nazo

    Nazo - 2003-12-11

    Logged In: YES

    Oh, I forgot to say. With my Turtle Beach Santa Cruz, I have
    tested via sound editors such as soundforge, goldwave, etc
    and discovered that I can play back at up to 64-bit IEEE float
    at 96KHz. The in_mad winamp plugin is able to play 24-bit as
    well as 32-bit via the default waveout or directsound plugins,
    despite the fact that it says on the website that it can't be
    done. I doubt this is the only card that can do that (perhaps
    the Audigy 2?) and I imagine there are people who would love
    to take advantage of such things.

  • Nobody/Anonymous

    Logged In: NO

    > How about 32-bit, maybe even more?

    32bit quantization has no advantage over 24bit quantization
    (at the same sampling rate). It only takes up 25% more space.

    Word sizes greater than 24bit are only meaningful at low
    sampling rates (take the "edges" off the quantization).

    > the whole point of this project is that the higher
    > resolution decoding increases quality, correct?

    no. well, only up to a point. Professional studio recording
    occurs with 20bit or 24bit samples, which at a 96kHz
    sampling rate is as good as possible ("lossless"), unless
    you have the hearing of a bat (you can hear the beat of a
    butterfly's wings at 100 yards)


  • Nazo

    Nazo - 2004-02-06

    Logged In: YES

    Ok. I wasn't really sure. I'm no audiophile. I hear
    definite differences using the MAD decoder, but beyond
    that I can't really say, so it just occured to me to wonder
    if it would be beneficial or not.

    I do wonder though. You said it might be slightly
    beneficial at low sampling rates. Maybe an option then
    to enable it if sample rates fall below a certain amount or
    something if people wanted it.


Log in to post a comment.

Get latest updates about Open Source Projects, Conferences and News.

Sign up for the SourceForge newsletter:

JavaScript is required for this form.

No, thanks