Help - Search - Members - Calendar
Full Version: Double dithering?
Hydrogenaudio Forums > Hydrogenaudio Forum > General Audio
mp3chan
I'm using MPC plugin for winamp and shibatch equalizer. There is an option to dither the output on both of that. Is it good to turn on both of them. If only one is better, where should I activate it? in eq or in mpc plugin?
yq
Check http://groups.yahoo.com/group/eac/message/6445, part about dither. It seems that dither should be applied only once.
Garf
If it should only be applied once, the last plugin in the chain should be your choice.

Edit: I am a stupid moron, ignore me.
Frank Klemm
QUOTE(Garf @ Feb 15 2003 - 10:08 PM)
If it should only be applied once, the last plugin in the chain should be your choice.

This answer is completely wrong for WinAMP 2.x .
I hope Garf can explain why.
Garf
QUOTE(Frank Klemm @ Feb 15 2003 - 10:36 PM)
This answer is completely wrong for WinAMP 2.x .
I hope Garf can explain why.

You are right - sorry.

WinAMP 2.x does

compressed data -> decoder (32 bit float) -> 16 bit -> DSP -> output

You want the dithering on the 32 bit float to 16 bit conversion, so at least in the decoder plugin. I'm not sure about dithering multiple times in this setup, logically it would be needed for each plugin that does 32 bit -> 16 bit conversion internally, but I do not know how noise adds up. I think noise would be actually minimized if we dithered each time, is this correct?

For FB2K my advice holds, because it does

compressed data -> decoder (32 bit float) -> 32 bit -> DSP -> 32 to 16 bit conversion+(dither) ->output
mp3chan
Does mpc plugin decode to 32bit float? I think it only decodes to 16 bit.

compressed data -> mpc decoder (16 bit) -> DSP -> output

So where should I add dithering process?
Garf
QUOTE
Does mpc plugin decode to 32bit float? I think it only decodes to 16 bit.

compressed data -> mpc decoder (16 bit) ...


I said

compressed data -> decoder (32 bit float) -> 16 bit

same thing...data is converted to 16 bit before being sent off to DSP.

QUOTE
So where should I add dithering process?


I think everywhere.
mp3chan
Hmmm. It would be better if mpc doesn't convert the output to 16 bit like what MAD plugin does.
Garf
QUOTE(mp3chan @ Feb 16 2003 - 12:08 AM)
Hmmm. It would be better if mpc doesn't convert the output to 16 bit like what MAD plugin does.

This is not MPC's problem, it's a limitation of WinAMP's plugin architecture, which converts everything fed to the plugins to 16 bits. MAD has the sample problem with plugins.
mp3chan
QUOTE(Garf @ Feb 15 2003 - 10:22 PM)
This is not MPC's problem, it's a limitation of WinAMP's plugin architecture, which converts everything fed to the plugins to 16 bits. MAD has the sample problem with plugins.

I typed a wrong sentence. I mean it would be better if there is mpc plugin that produce 32 bit output like MAD.

Mad has the sample problem? I never experience that problem. What's the problem? I think the output of MAD is really 32 bit, because my sound card that doesn't support 32 and 24 bit natively cannot play any song if I set MAD plugin to 24 or 32 bit. I tried on my friend's soundcard (SB extigy) and it works. Furthermore If I set MAD to 32 bit, shibatch eq cannot process the sound data. Perhaps Shibatch only supports up to 24 bit.
Garf
Sigh, *fourth* attempt at explaining.

Whenever there is a DSP plugin in the chain that doesn't support 32 or 24 bits (almost all of them), everything needs to be downsampled to 16 bit. This is independent of whether you're using MPC or MAD, it's a limitation of WinAmp 2.x.
mp3chan
I'm sorry Garf, but I really don't understand your explanation well. I hope that I have understood your explanation well now. Regarding to your explanation, even MAD plugin also limited to 16 bit too in winamp 2.x though it's configured to 32 bit resolution.

compressed data -> MAD decoder (32 bit float) -> 16 bit -> DSP -> output

Is this correct? One more sorry if you have to do fifth attempt... unsure.gif :'(
KikeG
The right thing to do in this case would be to apply dither twice, one at mpc decoding and another at Supereq output. The only drawback is that you will have slightly increased background noise due to this "extra" dither, but I don't think it is noticeable under usual listening conditions. Many consumer cards have higher noise at the output that the produced due to this extra dither.
Garf
QUOTE
I'm sorry Garf, but I really don't understand your explanation well.


Probably because it's a rather complicated mess and I'm not always very good at explaining things.

QUOTE
Regarding to your explanation, even MAD plugin also limited to 16 bit too in winamp 2.x though it's configured to 32 bit resolution.

compressed data -> MAD decoder (32 bit float) -> 16 bit -> DSP -> output

Is this correct? One more sorry if you have to do fifth attempt... unsure.gif  :'(


Yes. If your DSP plugins support 24 or/and 32 bit input, the conversion is not needed, but most do not, so in practise you always end up with 16 bits.
mp3chan
QUOTE(Garf @ Feb 16 2003 - 08:46 PM)
QUOTE

Regarding to your explanation, even MAD plugin also limited to 16 bit too in winamp 2.x though it's configured to 32 bit resolution.

compressed data -> MAD decoder (32 bit float) -> 16 bit -> DSP -> output

Is this correct? One more sorry if you have to do fifth attempt... unsure.gif  :'(


Yes. If your DSP plugins support 24 or/and 32 bit input, the conversion is not needed, but most do not, so in practise you always end up with 16 bits.

I think Shibatch eq can process input up to 24 bit. So if MPC plugin doesn't convert to 16 bit, it should be

compressed data -> MPC decoder (32 bit float) -> 24 bit -> DSP -> output

What do you think?
mp3chan
QUOTE(KikeG @ Feb 16 2003 - 08:01 PM)
The right thing to do in this case would be to apply dither twice, one at mpc decoding and another at Supereq output. The only drawback is that you will have slightly increased background noise due to this "extra" dither, but I don't think it is noticeable under usual listening conditions. Many consumer cards have higher noise at the output that the produced due to this extra dither.

Yeah. You're right. But since I use digital output of my soundcard and sent to my Cambridge Megaworks 250D that support 24bit/96kHz and has SNR 99dB (at least it's mentioned at the box).

If I dither it once, the maximum background noise is -90dB, but if I dither it twice it's possible to get noise about -88dB. Hmmm... not good I think, since I want to maximize the quality of the input to my speaker.
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.
Invision Power Board © 2001-2008 Invision Power Services, Inc.