What's the concensus on software upsampling at head-fi?

Apr 23, 2006 at 2:39 AM Thread Starter Post #1 of 15

Cyrilix

500+ Head-Fier
Joined
May 31, 2005
Posts
715
Likes
10
Software upsampling....something like using the foobar2k SSRC resampler set to a higher sampling rate like 96K, for example.

What kind of sound differences do people note?
 
Apr 23, 2006 at 2:49 AM Post #2 of 15
Quote:

Originally Posted by Cyrilix
Software upsampling....something like using the foobar2k SSRC resampler set to a higher sampling rate like 96K, for example.

What kind of sound differences do people note?



I used to use it with my Audiophile 24/96 card and noticed a sense of increased smoothness/spaciousness to the music. Effect was subtle, but definitely noticeable.... more "analog like" sound, particularly in the highs. Dunno if it's just that the DAC on the card likes 24/96 better or what.
 
Apr 23, 2006 at 2:54 AM Post #3 of 15
I'm wondering if upsampling will reduce the fidelity of your audio in a significant way.

I remember people were talking about the Audigy 2 ZS and resampling problems.
 
Apr 23, 2006 at 4:28 AM Post #4 of 15
There is no agreement just like with burn-in. I personally don't upsample because I want the unaltered original sound.
 
Apr 23, 2006 at 5:02 AM Post #5 of 15
The only instance where I actually believe upsampling can make an audible difference is in certain cases where sound cards have built in upsamplers which are inferior in quality to Foobar's software upsampler. By upsampling the audio in Foobar you would effectively be bypassing the upsampler on the sound card and hopefully end up with better sound.
 
Apr 23, 2006 at 6:41 AM Post #6 of 15
Hmm, I see...it's just that lots of DACs are OS DACs. Do they not do the same thing?
 
Apr 23, 2006 at 9:19 AM Post #7 of 15
I only notice the sound difference with very few songs (of my collection). The majority of my stuff don't notice but with a few stuff (say every 100 of my 10 000 songs) i tested out anyone could notice the difference in sound. Because of this, i don't really resample my stuff since it kills my cpu. However if i'm not running anything then i consider it.
 
Apr 23, 2006 at 1:58 PM Post #9 of 15
I think it's theoretically possible to improve >0 dBFS signals if you upsample to an even multiple of the original sampling rate while at the same time increasing the bit depth.

I don't suspect converting 44.1 to 96 or any other fractional multiple is ever going to be a good idea, and even going to an even multiple is going to be bad if you don't change the bit depth, because your DAC might have some headroom, and the software obviously doesn't.

By increase the bit depth I mean, for example, dumping 16 bits into 24 without rescaling to fill 24. Some music - especially hotly mastered pop - is going to overflow 16 bits when you resample it.

I don't know if any players do that kind of bit depth change, but would be interested in listening to one...
 
Apr 23, 2006 at 2:24 PM Post #10 of 15
I can hear a different between 44 & 16 vs. upsampled (wide mode) using my Lavry DA10. I would say preference is the difference. I attempt to keep my Lavry DA10 set to Crystal lock on 44.1 kHz. Sound so pure to me...But I would not contend that this is doctrine. I would compare the Lavry DA10 wide to the Benchmark normal setting, while Crystal lock gets is just right.

Right now I am feeding my Lavry DA10 via my HagUSB Coaxial output.
 
Apr 24, 2006 at 4:02 AM Post #11 of 15
For AV710, 96KHz upsampling definitely changes the sound - the soundstage gets better defined, sound gets "closer" ina way, dynamic range is there, but at the same time it's more harsh and brittle. I like it sometimes, I dislike it most of the time.

You could oversample if you don't like what you get.
 
Apr 24, 2006 at 4:48 AM Post #12 of 15
I think I may be mixing up upsampling and oversampling. To me, they both mean to alter the sample rate to something higher.
 
Apr 24, 2006 at 5:28 PM Post #13 of 15
upsampling, or resampling, originated out of the recording studio for DAT equipment (48 Khz) to match Redbook CD equipment (44.1 Khz). Upsampling your sound files does not make your music more accurate or increase any form of dynamic range. Most of the time, upsampling actually LOWERS your dynamic range, and can add a ringing effect in the high frequencies. Dedicated hardware resamplers usually have a high dynamic range (130-140 db), but they can still be prone to high frequency ringing as well. The SSRC resampler used in foobar lowers your dynamic range by 1-2db; test it in RMAA to find out.

People may prefer the sound of the upsampling DSP because it smears the sound, similar to how an SET or Tube amp smear the sound through odd order harmonic distortion. There is nothing wrong with prefering the sound of an upsampler, as long as you understand what it's actually doing. Hope this info helps
smily_headphones1.gif
 
Apr 24, 2006 at 5:51 PM Post #14 of 15
I use 192KHz resampling on Winamp so I can bypass the hardware resampling on my Audigy 2. When using the upsampling, I think I hear the sound just open up a bit more. Everything seems to just become slightly better defined. I have no idea if this is a placebo, with me trying to hear a difference with the upsampling. I haven't done any blind tests so I'm really not 100% positive on this, but I think I can hear the difference.

Aditya
 
Apr 24, 2006 at 7:45 PM Post #15 of 15
Quote:

Originally Posted by a1rocketpilot
I use 192KHz resampling on Winamp so I can bypass the hardware resampling on my Audigy 2. When using the upsampling, I think I hear the sound just open up a bit more. Everything seems to just become slightly better defined. I have no idea if this is a placebo, with me trying to hear a difference with the upsampling. I haven't done any blind tests so I'm really not 100% positive on this, but I think I can hear the difference.

Aditya




Creative cards (except for X-fi, audigy 2zs pcmcia, and Audigy 2 value), and certain AC97' cards, always resample 44.1khz data to 48khz. Usually, their resampling algorithms are poor. So in your case, you are using a better resampling algorithm than what creative uses in hardware. This would result in better sounding music. If you are upsampling, it is usually best to upsample by simple integers. So, if you are upsampling 44.1 data, it is easier for the resampling algorith to change that 44.1 data to 88.2 khz than to change 44.1khz data to 192khz or 96khz. The problem is that not all soundcards support 88khz.
 

Users who are viewing this thread

Back
Top