Simple question really.

What would give me the best sound quality with the Soundblaster Audigy 2. AFAIK it is a 24/96 soundcard?

At the moment I have output data format set to 16 bit. Would 24 bit or even 32 bit make any difference? The files I am playing are 16 bit 44.1khz Monkey's audio.

Thanks
John

Turn to 24 bits and use KernelStreaming as output (if You are using kX drivers try DirectOutput2) You can try using dighthering, but there is small diffrence in sound quality when using dightering and 24 bit.

"Small difference" is an exaggeration. At 24bits there is no perceivable difference between dithered and undithered output and dithering is a pure waste of resources.

I suggest using 24 bits without dithering, and resampling to 48 kHz.

I am Using the Audigy 2 ZS platinum pro, wich should be the same architecture as the "normal" audigy 2. My Settings are 32 Bit output, resample to 48000, no dithering, KS. 32 bits because the Kernelstreaming-config-window in Fb2k tells me that the prefered output mode should be 32 bit when using Kernelstreaming (and it really works!!) I don't know if this is just a bug (because its definetely a 24 bit card (which works intern with 16 bit)).

While on this topic, does anyone know what to use on teh Nvidia SoundStorm APU? The tech-pdf states it supports in- and output of 8, 16 and 32 bits formats (and i'm using SPDIF only), and ofcourse 48kHz. I tried 24 and 32 bits with KernelStreaming, and 24bits didnt work, but 32 bits works perfectly. Also, should i use SSRC to resample to 48kHz ?

Ok thanks for the replys. As always support is top notch from HA.

I'm now using Kernal Streaming, output data format set to 24bits, and finally resampling to 48000Hz.

There are two confusions to this though. Am I resampling to 48000Hz simply so the Audigy doesn't do it? If so, what benefit does this give? And there is a note on the KS options saying that 32 bit is optimal (which meischder points out). So I'm not sure whether 24 bits or 32 bits is needed. 24 bits sounds optimal for the Audigy and 32 bits sounds optimal for Foobar.

Cheers
John

If S/PDIF output is 48kHz then you should use 48kHz resampler. If SS supports 32bit output, use it also.


meischder: I think same settings as Yaztromo will be best for you too.

Ok. Thanks.. But when i go in the Kernel Streaming options, it says "NVIDIA® NForce™ Audio - Optimal resolution: 16 bits per sample".

Still it works perfectly on 32-bits fixed. 24-bits gives me a big error message. And according to the technical docs (which can be viewed here: http://www.nvidia.com/object/LO_20020712_6735.html - read on the bottom: Appendix) it supports 8, 16 and 32 bits input and output over S/PDIF.. There's also 2 different SoundStorm APU's. The basic one (Nvidia NForce APU), and the more advanced one (Nvidia SoundStorm APU, which i have)..

A bit confused here..

This is what soundcard drivers report and isn't necessarily accurate.
And Audigy2 works internally with 24-bit, as long as you use p16v chip.

The maximum word length the S/PDIF interface supports is 24bits per sample.

If you use 32bit output in fb2k it will be downsampled to 24bit afterwards.

I also notice that using resmapling to 48000hz in both slow mode and normal mode gives clipping warnings in fb2k console. Although I'm guessing that when the audigy 2 resamples to 48000hz that clippings also occur?

Thanks
John

Not with my SB Live! Are you using Replaygain? If you are, try to lower in Playback preferences Preamp value in "Files with RG info". If you don't use Replaygain, try lowerin "Files without RG info" value. Something like -1dB or -2dB must be enough i think.
EDIT: Clipping should not came from resampling. With Audigy HW resampling it produces distortion. What other DSP's are you using?

bidz: Do you have any equipment with what you could see what format S/PDIF output data is? If its 16bit when you use 32bit output, then you should use 16bit with dithering maybay. If outputformat is 24bit, then 32bit output mode from foobar is better.

I'm currently using AC3 encoding on my S/PDIF (SoundStorm APU does AC3 encoding in realtime), and passes this on to my receiver, which decodes it, so i can mix the channels (rear, middle, front and sub) as i see fit in my SoundStorm configuration. I guess that would be 24-bit/48kHz ? strange, since i get a error message from fb2k when trying 24-bit, but 32-bit works fine..

No other DSP's at all. Replaygain, volume are both off. Only active one I see is Resampler. The files are monkey's audio, which don't clip when resampling is off.

EDIT: Tested numerous MP3's which have been max-no clip gained and no clipping occurs. Only seems to clip with Monkey's Audio files.

I think I am missing something...

I have an SB Audigy 2 and have Foobar2000 set to 24bit output. If 24bit output is what my sound card supports, what good does setting Foobar2000 to 32bit output do? Also, I thought resampling problems on SB got fixed on the Audigy and Audigy 2 series?

I'm sorry, I'm missing something big, here.

If my CDs, MP3s and Vorbis files are all 16-bit, then what perceivable difference would there be between 24- and 16-bit output? I can see that there would be a difference with my AC3 files, but I don't have too many of those.

Edit: I should note that I haven't actually checked -- I am using a SB Audigy, which does not support 24-bit output, thus I can't check for myself. I'm just curious about how or why there would be a difference if, hypothetically, I could have output in 24-bit mode.

Do you think Kernel Streaming output really better then DirectSound v.2 for Audigy 2 ?
I'm using drivers from Creative version: 5.12.01.0402-1.82.0020