Sound quality for Rossini using lossless Rock streaming

I am using Roon ROCK (v1.5 build 363) on a NUC8 i7 (16GB RAM/SSD) to stream to my Roon endpoint (RAAT) dCS Rossini DAC via a Netgear GS108 switch. Music from a Synology NAS and Tidal. The NUC BIOS is tweaked (turbo off, speedstep off) and Roon reports a Lossless signal path (Fixed Volume, no DSP).

Compared to using the dCS streamer and UPnP renderer the sound quality is not the same. There is some treble harshness, a little less bass, sligthly less width and depth of soundstage and a small reduction of resolution. The differences are similar independent of sampling format or streaming source NAS/Tidal. To avoid exaggeration, Roon sounds good, just not as good as dCS on its own.

I have read the suggested Roon instructions, forums etc but currently I am not sure on how to proceed. I was under the impression it is possible to get similar sound quality using Roon (without resorting to DSP, upsampling or hardware optimizations like fanless, LPS, audiophile switches and so on)?

Hi Mats, have you tried just disabling DSP for this endpoint in Roon? (Master switch, top left in DSP settings)
That usually gives a perceptible increase in SQ, at least to my ears.
That said, i use an Aries G1 which also performs a notch better with Lightning DS, than as a Roon Ready endpoint. Perhaps it’s up to the implementation of RAAT in these apparatus…

Hi Mikael,
thanks for the reply, DSP is disabled, which I believe is the only way to get a lossless path.
Also, I forgot to mention disabling a few more things in BIOS, like bluetooth and wifi.

Three clarifications from you?

1- By dCS streamer/UPnP you mean using the dCS app itself to access files in your NAS, correct?
It’s an interesting comparison that I have not made as I don’t use a NAS in that kind of configuration.

2- You’re not changing the hardware setup at all - ie your Rossini is connected to the switch and nothing else on the computer/network side in all cases? Specifically you don’t have a USB connection to the computer correct?

3- Your DSP setup in Roon is completely disabled, correct?

In the case above the sound difference probably comes from the difference in implementation of RAAT (binary provided by Roon) vs the renderer (probably an open source component, probably optimized for the network card implementation). My understanding is both of these are done in the network component of the Rossini, not the FPGA, so by the time you reach the actual DAC process the path is the same.

Frankly it could be an unavoidable difference, but @AMP would be the one to chip in here.

I have only used the Rossini via RAAT over the network, with the network connection coming from the mac mini directly (not a trivial setup, I didn’t want the clutter of a switch). I truly love the liquidity of the sound. But I admit not having done comparisons, and I would not dismiss your findings with an argument such as bits are bits.

In the current version of the network firmware RAAT and UPnP share the same signal path. The only difference is in the code that receives data from the network and passes it along. There are some major downsides to this approach from a technical standpoint, but there should be no difference in audio performance – both sources are playing from a memory buffer at that point and for the same file they should be shifting the same bits.

This will be completely different in a couple of weeks so sit tight and wait for the next update. There may or may not be an audible change, but there’s definitely a major technical change and, therefore, a significant change in overall performance.

That’s as far as I’m going to dip into this discussion. This one comes up occasionally, and after performing tightly controlled tests the end result has always been that there’s no real difference. In most cases the question has come up after someone reads something online that says that RAAT sounds worse than XYZ and that little bit of bias goes a long way. Remove the bias and the difference evaporates.

2 Likes

Thanks for your comment.

  1. Yes, I am using the dCS Rossini iPad app.
  2. No changes to hardware. No USB, only power and Ethernet connected on Rossini (same for NUC).
  3. No DSP whatsoever

Thanks for the explanation Andrew, looking forward to the update.

Not trying to start an argument here, but my confirmation bias was in Roon’s favor after having read the Roon article i TAS a month ago. I have repeated the test with ample time between sessions, reversing the order etc, but at the end of the day it is just my opinion. My only goal is to understand what I have missed, fix it and live happily with Roon.