You can use MusicScope to capture…
https://www.xivero.com/musicscope/
There’s a discussion on HRA…
It’s got to be possible to define some effective measure of bandwidth… Possibly the highest frequency at which 95% of the energy in the signal is stored. You point about false ultrasonic components is also important, I wonder how to deal with that.
I have Musicscope, but I haven’t used it to capture a stream, did not know that’s possible. What I have used is TIDAL or Audirvana → Soundflower → AudioHijack.
I tried directly to AudioHijack but had some trouble with that, possibly user error.
1 Like
I’m wiling to say that you could use a bandwidth definition that is defined as:
BF=min(Freq for 95% of energy, Freq where dropoff goes to -100db, Freq where noise profile for DSD kicks in)
Then you could bucketize that a bit just so that you have some sort of DR-like buckets like F22 (equivalent to redbook), F24 (equivalent to 48KHz sampling), F44 (equiv to 1xDSD, 88KHz sampling), etc…
Nah, not that simple. Much recorded music by nature has very little ultrasonic content, so your 95 percent figure of merit might produce spurious results on many recordings.
And how would your bandwidth methodology assess this previously noted MQA example?
AJ
… the day the music died 
it’s not about ‘wavelenths’.
Imagine if digital audio was more clever implemented, with 20 - 20 khz audible frequency, and an independent sample rate not tied to frequency? And then some clever lossless compression on top of that. In a way, MQA is an effort to achieve this (but not sure they succeeded very well).
Not sure this is possible, but coming from the IT world it seems to strange with audio files that pack so much information that is not audible, like it was made for dogs or something 
Fine, so it needs work. Come up with something!
This graph does not show any music content above the audio due to sharp cut-off of the digital filter. It is probably made from a 44.1k master. The noise start to rise above 22.05k is ultrasonic noise and aliasing caused by MQA leaky filter.
A better example is below, showing PCM 192k that extends above 48k bandwidth.
There’s some ultrasonic music contents above 20k bandwidth.
MQA support is now live.
4 Likes
Wow! Very nice. Can’t wait to try it out later when I get home. Thank you so much!
Just to clarify, if using a non-MQA Streamer like an Auralic Aries, Roon will do the first MQA unfold just like the Tidal desktop app?
Yes it will.
1 Like
The Aries, or any other streamer for that matter, would essentially be completely transparent to the process. Roon will see the DAC via the Aries. If the DAC is a fully-decoding MQA DAC, and no DSP is applied in Roon, then Roon will stream the data undecoded. If you apply any form of DSP in Roon, Roon will:
1- Unfold the data to 2x the original data’s sample rate
2- Preserve the MQA rendering information on the side
3- Apply the DSP to the unfolded (but not rendered) stream
4- Reattach the MQA rendering info to the stream
5- Send that to the DAC via the Aries
If the DAC is MQA compatible, either full compatibility of just renderer, it will render the stream using the MQA-specified filters. The DAC will not (need to) do the first unfold since that already happened in Roon.
BTW… Do we know for a fact that any full-decoding DAC will indeed work as a renderer if a first-unfold is performed in software? I don’t think this is obvious.
2 Likes
The Explorer 2 isn’t at the moment but will when (if) a firmware fix is released.
Yes, the Pro-Ject S2 DAC shows MQB on it’s screen when the DAC is set to MQA renderer mode.
The same way it’s been working when Tidal Desktop App and A+ have been doing the 1st unfold.
Yes I realize the mistake in my original post. I should have stated that I have an Aries and a non-MQA DAC, the Chord Dave. So I will now get the benefit of the first unfold from Roon.
Many MQA USB DAC do. Lumin does when used with Roon. Does your DAC support it?
For those that do not support this feature, Roon Labs have said they’re chasing the manufacturers to get this fixed.
I will check when I have a sec. Easy to tell as it would tell me the filter being used is M1.
Actually prob doesn’t matter as the E2 never turns off the MQA filter regardless!