Okay, somewhat technical question here and I know just enough to be dangerous, so please forgive if what I’m asking makes no sense.
In HQ Player, there’s a setting for “Bits”, which (as I understand it) represents the actual resolution capability of the DAC. For R2R DACs, this is typically somewhere around 20 bits, even though our MSB units can accept files up to 32 bits.
Does anyone know the correct HQP setting based on the true resolution of our DACs?
HQ player needs the DAC resolution for doing either dither or noise shaping. It is assuming that it can replace the internal dither or noise shaping in a DAC box by knowing the true resolution of the internal DACs. This will not be advantageous for any of our current DACs. Instead our DACs will be optimally dithered no matter what the input sample rate or bit depth is. HQ player does not need to add additional dither to compensate for the DACs true resolution. HQ player should be configured to give the DACs the highest resolution possible over the delivery channel you are using. This would be 24 bits for most inputs (S/PDIF, AES/EBU, USB, Pro USB) or 32 bits for the Renderer (or perhaps a server with 32 bit capability over ProISL or I2S, though none currently exist). Since HQ player is not directly interacting with the DAC modules, the onboard processing will take the input stream and apply the correct dither for the modules while also raising the sample rate from whatever you set in HQ player to about 3Mhz (2.8224Mhz or 3.072Mhz).
Do make sure that the HQ player setting does not exceed the digital channel you are using. If you set it to 32 bits and the DAC display reads 24 that will not be good for quality. In this case HQ player should be set to 24 bits.
Thanks for the quick and in-depth reply.
I’d like to make sure I understand your comment:
"or perhaps a server with 32 bit capability over ProISL or I2S, though none currently exist). "
If I have a 32 bit source file played Roon > HQ Player > NAA (Sonore Rendu) > Pro USB/Pro ISL, will I get 32 bit resolution in the DAC?
No, the ProUSB is limited to 24 bit by design. When it comes to 32 bit audio playback not every playback software does it correctly. There is a difference between 32 bit fixed point and 32 bit floating point files. Due to inconsistency is servers, software, and distributed music we made the decision to limit it to 24 bit.
Thanks Jonathan.
So to be clear, if I want to play a 32 bit file “natively”, I can only do so via the Renderer? Which also means I’m limited to 4X DSD?
That is correct. It may be that in the near future a server maker will implement ProISL at 32bits, but as of now only the Renderer will playback 32bits correctly (we control the playback in the renderer). We found too many software implementations of 32bit playback over USB to have serious problems. As you pointed out the Renderer only supports up to 4xDSD.
There are no “native” 32 bit files. They will all be up-sampled or processed to 32bits.
Thanks Dustin. Yes, by native 32 bit I meant something that I had stored as a 32 bit file, such as something remastered using PGGB.
Hi Dustin.
I use Nucleus to drive the Cascade, I always have all the processing turned off. If there is a high res version of a stream and a 44/16 version of a stream I always use the 44/16 version because I assume the high res version upscaled. When I buy a stream, for example from Native DSD, I alway buy the version the stream was recorded at, not one of the upscaled versions.
Is this a good strategy for Cascade?
Thanks,
Dan
It seems to me there’s a challenge here if using a Roon endpoint feeding the ProUSB/ISL.
With Roon, any type of DSP (even just Convolution filters for room correction, without upsampling or any other DSP processing) results in a 32 bit output. This is fine with the V2 Renderer which passes 32 bits.
Using a Sonore Rendu as a Roon endpoint feeding the ProUSB/ISL combination, the signal will be “downgraded” (for lack of a better term) by the MSB hardware to 24 bit which to Dustin’s point, reduces quality.
With HQ Player, the output can be set to 24 bit so even with Convolution filters or other processing, the output bit depth can be aligned with the capability of the ProUSB/ISL modules.
If using PGGB to create upsampled versions of original files, it seems those files should be created as 24 bit if using ProUSB/ISL vs 32 bit if using the V2 Renderer.
So to get the best playback quality using ProUSB/ISL as the input, the incoming file should not exceed 24 bits which precludes the use of DSP if Roon is the player. It also suggests specific management of HQP settings or PGGB processing to ensure 24 bit streams.
Do I understand this correctly?
If you are using Qobuz there is often a high-rez version which is the studio master (original digital recording). If you are using the Cascade it is usually very plain which is the original, it will be much less muddy and congested than all of the other versions regardless of sample rate or bit depth. Here are a few rule of thumbs to find it among several options. I usually que up all versions and after listening to only about 10 seconds of each (even if it is just a few notes) you should be able to determine which is the original master. Then to maximize the quality I give the phase invert button a few presses while listening to something in the midrange to determine the correct phase for that recording. For the music I enjoy, there is about a 40% chance that inverted phase is correct.
If there is a 44.1Khz 24bit version or a 48Khz 24bit version, these are always the original studio master regardless of any other versions available.
If there is only a 44.1Khz 16bit and a 88.2Khz 24bit version, the 88.2Khz version is almost always the studio master.
If there is a 176.4Khz version, it has a much higher than 50% chance of being the studio master if there is no 88.2Khz version.
If there are only 96Khz and 192Khz versions, there is a roughly equal chance either could be the master.
If there is a 44.1Khz 16bit version, a 96Khz 24bit and a 192Khz 24bit version, any of them could be the studio master with roughly equal chance.
Just turning on the DSP block in Roon (even with no significant processing selected) completely destroys the sound quality of the recording even through the renderer at 32bits. It will make me quickly exit a listening room if it is turned on for any length of time. To me PGGB makes all recordings sound the same, not bad per se. Its not an effect that I personally enjoy.
Thanks Justin,
Dan
Thanks Dustin. Are you confirming that my understanding is correct?
Best,
Bob
Hi Dustin
Is the phase button implemented on the Cascade? I tried what you suggested, pressing the phase button to see if a music track sounded better or not, but did not hear any change.
I have a number of pink noise test tracks, some in phase and some out of phase. The sound difference between them in and out of phase tracks, of course, dramatic.
An in phase track appears to come right from the right in the middle of the image, between the speakers. An out of phase track the appears to come from two directions, one from each speaker, and the middle of the image disappears.
When I press the phase button on the remote, a crossed out zero icon shows up on the Cascade display but there is no change at all to the sound coming from the speakers for these test tracks.
I also tried a number of music tracks and got the same result, the phase button did not appear to make any difference in what came out of the speakers. Could hear a very slight click, most of the time, when I pressed the phase button, so it seems like the Cascade is doing something, but it doesn’t seem to affect the phase relationship between the tracks.
Dan
Hello Dan,
If you see this icon it means the absolute phase is inverted. When doing tests in our listening room what I noticed when toggling phase invert was sometimes different from what Dustin noticed. For example, in one song was able to pick up the difference with the piano but Dustin noticed it with a different instrument/voice. Once you learn to hear the difference you are able to focus on it and tell on a new recording quickly. What I did to learn was to focus on one instrument and toggle it. To me, it would make some instruments just sound more believable.
In regards to your phase test files, it sounds like you are talking about is one channel is out of phase. When this happens there won’t be an image and the difference will be HUGE. If both channels are playing the same noise there will be a void when you are centered as the two channels are cancelling each other out. It sounds horrible and as you walk around your room there will be voids/dead spots. Small changes in your head position will make things jump around the room. This is the effect of one channel out of phase.
Hi Johnathan,
Ah… Now I know what to listen for. The button flips the phase of both speakers! I misunderstood what it was doing.
Thanks
Dan
Thanks for this information. So, to my understanding, the internal processing of DD happens at DSDx1 rates (in 44.1 or 48khz family). Is this correct? And if a file at a higher sample rate be fed to DD, I assume it would be do samples to DSDx1.