Upsampling, Can there be too much?


I've owned the Chord Mscaler for a year and loved it, but recently added two new components that have built in upsampling: The Aurender W20SE, and the Jay's Audio CDT3-MK3. I find the Mscaler works well with the Aurender's built in upsampling, but not the Jay's.

 

Conclusion: not upsampling the Jay's, and standard redbook 16-bit 44Khz to the Mscaler gives incredible 24-bit 705Khz to the Hugo TT2 DAC for finest sound.

 

With multiple upsamplers in a chain has anyone gotten static, popping, smearing, or any kind of distortion from too much upsampling?

128x128brandonhifi

I like upsampling music playback, NOS sounded a bit rough to my ears, but some people swear by it as some holy grail DAC mode. Definitely subjective opinion here.

Agreed, unquestionably subjective. Some listeners do not like up/over sampling due to algorithms/mathematical use and subsequent manipulated-recreated or reconstruction of the signal. So definitely alternative approaches preferred by different listeners. Thankfully options of choice exist.

Charles

@brandonhifi, the first thing that I’d like to ask is what made you get an Aurender W20SE music server? Very nice! My go to audio gear guy in the Twin Cities since 1976 has been Bill Soderholm who owns Stereoland in Bloomington, MN. I bought my first audiophile system from him when I was a sophomore in college, which was a Marantz 100w per channel stereo receiver, 4 Bose 901 speakers, and a Thorens turntable. He’s got some really amazing stuff at his newly expanded showroom. 
 

   My suggestion when using a TT2 and an M Scaler is to let Rob Watts magic do it’s work and to not use any of the upsampling capabilities of any other powered boxes that you might have in your listening chain. I’ve found this approach to give the most accurate and non-fatiguing music reproduction. It’s also rewarding if you bypass the less then great Amanero USB input of the M Scaler in favor of using a single BNC cable instead. Here’s part of my setup. Music Server  > AQ Diamond USB > Audiowise SRC.DX > single WAVE Storm BNC > Hugo M Scaler > dual WAVE Storm BNCs > Chord Hugo TT2. 

@erik_squires  has it correct, I've experienced this. @adasdad  looks like he's found the answer for basic chain with similar setup.

 

I no longer pay attention to over sampling up sampling, whatever, 16/44 is just fine, and with your Aurender the transport and streaming should be running neck and neck. I gave up cd transports years ago, and I had a pretty nice transport.

Upsampling adds nothing but distortions.

 

Well, I wouldn’t go this far either. There are measurable and IMHO audible differences in how the frequency response changes with most DACs as the sampling frequency is increased to 88 kHz and above. Frequency response measurements are perhaps the thing listeners are most sensitive to. For this reason I have my Roon set to upsample below 88kHz and otherwise give me the original signal above that.

The biggest difference I’ve ever heard though is with how poorly older DACs handle Redbook playback vs. many new models. Back then it made 100% sense to upsample. Now, not so much.  I think it is from that early time that audiophiles are still stuck on the myths of high resolution formats.

FWIW We use professional DAW software to create .wav files of 384 KHz 32 Bit Floating Point, and our Chord DACs sound better than ever replaying these on ROON with absolutely no DSP of any kind.  Higher resolution source files are better than any DSP we have ever tried over the last 5+ years.    Why 384?  There is laboratory data showing that the digital noise floor dissipates around 356 (see DXD specs), and 32 Bit Floating Point everything sound superior in every way to 24 bit and below; way superior [Try It You'll Like It]

If up-sampling is all you got:

Our experience has revealed that the DAC/Streamer up sampling to DSD 2.8/5.6 affects the most pronounced change in auditory information.