A Midsummer Night’s Dream - Compare Simultaneous DSD64 and DSD256 Session Recordings

This is my third reply, which the previous disappear. A +0.34dB Signalyst remodulated Grimm track has been generated and uploaded to our Amazon server. It will appear with the others as soon as it can be imported.

(This "Leave Comment" program might have a problem.

I can leave a comment when I am on "Log in" state. However, I can't brows my comments unless I get "Log out".

When I am on "Log out" state, I can look at my posts. However, on this occasion, no attachments I added during my valid posting process are shown.)

If you are interested in quantization noise profiles of four sources, please refer screenshots available here.

https://www.dropbox.com/sh/ip2a2ucjvzgpbly/AAAOi93o0d6Fi7DceLW7ocqha?oref=e&n=20907960

Hi, I would like to pose some questions regarding the recording itself:

1-On one of the picture´s label is said that the 256 DSD stream feeding from the Horus to one of the Pyramix DAWS was via the red ethernet cable. So, it was handled by Ravenna into the Merging recommended PCIe network card. On the Grimm side, I presume that the DSD 64 stream went into the second Pyramix DAW through a Mykerinos card with a SDIF daughter card, taking the feed from the Grimm SDIF output. Correct?

2-Was the base clock provided using the converters in internal mode or an external clock was used to feed both converters, thus running them in slave mode?

3-Despite the clock method employed, what are your thoughts on clocking stability between streaming and clocking via an ip protocol such as Ravenna and a SDIF output and separate clock feed to the Mykerinos card from the Grimm? Any issues?

4-Finally, would it be interesting if one day you could make another test, using two Horus and their mic pres straight from the mics on stage and recording at 64 and 256 simultaneously to two Pyramix via Ravenna. Just a suggestion of course, by now, thanks for the experience so far.

Hi Paulo

1- Correct. Both the Horus and Grimm analog inputs were fed from the same analog source in parallel. Except for that, they were completely independent recording systems.

2- The Horus system ran off its internal clock while the Grimm off an external master clock generator, which I believe was also a Grimm.

3- While I've used both, I don't have sufficient experience to have noticed any issues. Each of those architectures are from different hardware generations, and each perform well in their own usage. When using multiple Hours', one has to be declared master Clock with the others slaved to it over the network. There's been several years of history of that configuration with no reported issues, including stability.

4- If I understand your point to being a setup like the current combined Grimm and Horus recording the analog mix, but with two Horus at different DSD bit rates, that may soon be possible. Please recognize that while all microphones are tracked directly from the stage for safety purposes, the real recording is the analog output of the mixing desk, and that which is presented here. The 5.0 multichannel is these same two stereo outputs, plus the center positioned and rear ambience omni microphones as added channels. Those three additional microphones are the only "straight from the mics on stage" used, and their outputs are pre-amplified by separate Rens Heijnis mic pres. Both the Grimm and Horus were fed at line level.

Thanks for your questions Paulo.

The 0.34dB level increased file pf the Grimm 64fs is now available as Track 5. Unfortunately the site labeling has yet to catch up, and says n/a, but clkicking the 64fs button for Track 5 will download this named file:

justlisten-JLBFOMendelssohnsession-JLBFOMendelssohnsession+stereo-05_Mendelssohn_BFO_Scherzo_grimm_signalyst_dot34dB_2ch64

Let us know how that works, and if that makes a difference.

I could get the file without any problem. Thank you!

I wonder why Miska adopted DSF format instead of DFF(DSDIFF) format. (All the original sources are of DFF format.)

In my case, a previous version of Korg Audiogate was used for a conversion from DSF to DFF.

I played the files last night. For some reason, A+ would not take the Grimm or Horus256 files, so I could only compare the Horus64 and 128. I could hear more air, the sense of being there with the 128 than the 64, similar to a previous poster, but with my system, it sounded fairly easy to hear the difference. I downloaded the Grimm and Horus256 again. They came out exactly the same size as the first time, but I'll try to load them again today. I am using my Exasound E28, so I should be able to hear all four files.

Larry

Thanks for setting this up. I'm using an exaSound e22, which handles DSD up to 256. I told JRiver to shuffle the four tracks, and listened to the list a couple of times. All four sounded fantastic, but I was able to identify them correctly. The Grimm sounded more like a recording, perhaps a little veiled, while the Horus 256 sounded more like live music. The 256 version had a lot more spatial cues than the others. My wife also heard a clear difference and said the 256 version sounded best (actually she just said "this one", because I hadn't told her what was different or identified them in any way).

A number of listeners have made the observation about the Horus 64fs having different qualities than expected. It needs to be emphasized again that the Horus labeled 64fs and 128fs tracks are 2nd generation remodulated products of the original Horus 256fs session file, and any judgement made using them is in a large part a judgement of the qualities of the Pyramix DSD Converter software app.

To provide an additional example, I've converted the original Horus 256fs file to 64fs using the Signalyst processor. It will be listed on the Just Listen Mendelssohn album page as Track 6.

Let us know what you think.

Thanks for setting this up. I'm using an exaSound e22, which handles DSD up to 256. I told JRiver to shuffle the four tracks, and listened to the list a couple of times. All four sounded fantastic, but I was able to identify them correctly. The Grimm sounded more like a recording, perhaps a little veiled, while the Horus 256 sounded more like live music. The 256 version had a lot more spatial cues than the others. My wife also heard a clear difference and said the 256 version sounded best (actually she just said "this one", because I hadn't told her what was different or identified them in any way).

I heard it the same way: the Grimm sounds like a very nice recording - the Horus (esp. at DSD256) sounds almost live.

Could not listen to the Horus 258, but the rest OK.

Main difference between the Grimm 64x and the Horus128x was the Grimms lack of 'air' or hall ambience. The sound was slightly more congested and slightly less dynamic. The 128 track sounded as if one were there, while the 64 x tracks for both versions lacked that air of the hi res track.

Equipment: Aurender X100 --> Accuphase DC-37 etc -- Larsen 8.

I compared Grimm 64fs file (with +0.34 boost) to Horus 256fs on my system - LKS Audio double es9018 dac, Sanders Magtech, Magnepan MiniMaggies.

Grimm file was excellent, but the sound stage was more compressed. Horus file was more dynamic and presented a more 3d sound stage where instruments were more discernible, especially in full orchestral passages.

I then tried to upsample Grimm 64fs file to 256fs with HQPlayer (poly-sinc/dsd7) and I had the impression that the difference between the two files was only slight and not so discernible anymore.

Thanks to make the files available.

Hi Paulo

 

4- If I understand your point to being a setup like the current combined Grimm and Horus recording the analog mix, but with two Horus at different DSD bit rates, that may soon be possible. Please recognize that while all microphones are tracked directly from the stage for safety purposes, the real recording is the analog output of the mixing desk, and that which is presented here. The 5.0 multichannel is these same two stereo outputs, plus the center positioned and rear ambience omni microphones as added channels. Those three additional microphones are the only "straight from the mics on stage" used, and their outputs are pre-amplified by separate Rens Heijnis mic pres. Both the Grimm and Horus were fed at line level.

 

Thanks for your questions Paulo.

Hi Tailspn,

thanks for the detailed explanation.

Yes, my proposed setup for a future experience would be two Horus units (or Hapi with the same AD cards), fitted with the internal mic pres which would take the signals from the stage mics directly into the AD card on one of the converters. Then the mic signals from that converter would feed the second via direct outs or Ravenna, depending on your Horus card configuration. Meaning no external mic splitters either. So no mixing console or external pres involved. You could leave the converters next to the stage to shorten mic cables and extend the ethernet cables to the DAW`s in the control room.

Then both Horus would feed the two Pyramix DAW's via Ravenna, one at 64fs (or 128fs) and another at 256fs, eventually using one of the recommended ethernet switchers or with direct connection to the DAWS.

The idea for this would be to compare the recording at diferent sampling rates only, using the same protocol for the DSD stream into the DAW, thus removing the converter comparison and tone from the mixing desk or external pres.

Well, just an idea, if one day you feel like it. Thanks.

Hi bibo01,

Your observation is most interesting. To me it suggests that the file sound quality differences have less to do with the recording DSD bit rate, and more to do with the bit rate of the playout file, enjoying some/most/all of the SQ benefits of 256fs by upconverting the bit rate of a 64fs native DSD file to 256fs (I avoid the term upsampling, for there are no samples in DSD, only the density of 1's and 0's bits proportional to the signal level).

Aside from the individual A/D converter design influences, mostly in the analog stage(s), there's no reason for there to be any SQ difference within the audio band of interest between 64fs and 256fs DSD. Paulo's above suggestion of using two Horus, instead of a Horus and Grimm at the two different DSD bit rates would eliminate that difference.

It may well be the sound quality differences between DSD 64fs and 256fs are how the entire playback system reacts to the different ultrasonic noise content in each, rather than recording bit rate differences. Testikoff's spectral deltas graph (above) and his observations also support this idea.

I was able to download the new version 5 successfully and played it last night comparing to the Horus 128 file. (I haven't been able to play the original Grimm or the 256 files - downloaded them twice, but they won't drop into my A+ playlist.) I could easily hear the difference, with the new Grimm sounding slightly artificial - like a very light coating of caramel were over the sound, which the 128 was more like you were really at the event.

Thanks to all for having these comparisons.

Is anyone using A+ (I'm still on the latest pre 2.0 version) for these comparisons and successfully loading versions 1 and 4?

Larry

I have a Big 7 Lampi which only goes up to DSD128, so I have listened to all the files except the 256 one.

The first order difference I hear is that all the Horus files sound better than the Grimm ones. They are clearer, more dynamic, I am more "there" and they draw me into the music more. Although the words may be different, I think this is consistent with many previous comments. When I say "first order", I mean these differences are significant and consistent. Enough to considerably increase my enjoyment.

A second order difference is that I think I prefer the DSD128 to the DSD64, but I struggle to put words to this and it could easily be expectation bias.

To be honest, I find this result to be rather disappointing! I have well over 100 Channel Classic recordings and, although they are all very good, I find myself thinking that I might enjoy them even more if they had all been recorded using a Horus rather than the Grimm, and (to a lesser extent) at 128 rather than 64.

Cheers

Philip

My previous msg hasn't appeared! Can't be bothered to retype it all! So briefly:

* Horus 64 substantially better than Grimm- clearer, more "there" and better dynamics.

* 128 may be a bit better than 64 but could be expectation bias.

Hope this comment "sticks".

To me, the 256fs file sounds more wide-band (less congestion in the mid-band, more extension especially in the treble), with greater transparency and low-level resolution (airier, detailed, also hear more hall acoustics cues, if not necessarily more in-time/in-phase), impulses sound drier (as if faster), bass is tauter (albeit with less weight), imaging is very stable.

But I also hear less body (less realism in a specific sense: that there's really someone - a human being, there) to e.g. the conductor's voice, less color and character to winds and strings, less warmth, texture and physicality, a lesser tactile quality overall.

The 64fs file has an almost tube-like warmth and colorfulness to it that I find appealing, and despite the comparative lack of air and treble extension, the depth/width relations sound just as realistic. Energetically, the sound of instruments emanate from where they're placed in the soundstage top to bottom, instead of through a comparative eye of the needle with a halo around it (admittedly exaggerating to get the point across). The harmonic content doesn't necessarily sound more in-time/in-phase, but it adds up to more being audible, adding to the overall impact. As a result, the playing of the orchestra seems to have greater urgency (if not thrust). Sound appears built from the bottom up.

The 256fs file, even though "better" in what to me are ultimately more "rational" aspects, sounds more ethereal, emasculated, there's less texture to sound and imaging. It's tempting to say that because one hears greater detail that the sound is "better", but it's not in a holistic way: wood sound less "woody"; the voice sounds less human (more of a head voice, too); etc. All this does add to the "winged" quality of Mendelssohn Dream without the music necessarily breathing more: in its overall impact, sound seems built from the top down.

In a nutshell the 64fs file seems to err on the side of (too much) warmth and (dis-?)coloration, versus the 256fs file's too little. A gut versus brain decision: the 256fs may file sound more like what I'm used to hearing in the concert hall, whereas the 64fs files, in some respects at least, sounds more beautiful than real.

But I wonder if the sonic imprint I'm hearing has more to do with the respective ADC rather than the sampling frequency? I'd expected the 256fs file to sound less like a digital recording, instead it's the 64fs file that, on the whole, makes me feel as if I were listening to a nice old Kenneth E. Wilkinson "Decca tree" recording (alluding to my favorite recording of the Midsummer Night's Dream) - which I realize may not be a positive assessment in everyone's book.

Long, rambling, sorry…

Greetings from Switzerland, David.

First of all a question to Jared.When will the full album be released?

Regarding the files I am still on a superslow wifi connection where I have only been able to download the two that interested me most in this comparison. The Grimm level corrected 64 and the Horus 128 which is the highest sampling my portable Hugo plays in DSD.

Like most posters here I too hear a difference between the two to the advantage of the 128.

I have basically the same impression as with the album from A Far Cry where especially in the Beethoven "Gedankensang" I hear slightly more ambience and air and definition/resolution from the 128 versus the 64 binaural file.

In that case with both files emanating from the Horus if I remember correctly.

I know some here will argue that Hugo is not the best DSD Dac to use in such a comparison.

I think it converts DSD internally to 32/384 before outputting as DSD.But it is the one I use when travelling and in both cases it enables me to hear the differences I mention.

Unlike the many tech-savy posters here I don't really have a clue as to why the 128 files sound slightly more resolved in both cases . I just hear that via my Hugo and HD 800 headphones they do.

Is it as simple as less noise /better sound or what?

When I get back to Singapore in early May I will try to compare all rates via a DAC that also plays DSD 256.

Meanwhile I would buy the 128 over the 64 of this album if available.

I also have a suggestion to make future comparisons even more interesting . Why not also record with the Horus in DXD parallel to the different DSD rates?

That would be the "ultimate shoot-out" I suspect.

Then again, Rob Watts told me recently that he is also working on his own ADC which might become the ultimate GRIMM/HORUS/ADC contender.

What if he delivers the most resolving ADC ever?

And combined with a DAC with 164000 taps?

Would the Horus still rule or not?

Hi tailspin,

I tend to agree with you. I have not found a system yet that does not react to higher bit rate of the playout file. Would it be possible to DSD record the signal coming out of the DAC and A-B it with the original? How is it possible to measure the ultrasonic influence on the rest of the playback system?

Tried something not specifically asked for: converted both files to 24/176.4 PCM using DSDMaster, and compared them on my home system. The respective sonic imprint or character of the Grimm 64fs versus the Horus 256fs files I wrote about at length above remains virtually the same after conversion. This makes me wonder if the specific "sound" of the respective ADC makes more of a difference on the net result than the sampling rate?

Greetings from Switzerland, David.

Tried something not specifically asked for: converted both files to 24/176.4 PCM using DSDMaster, and compared them on my home system. The respective sonic imprint or character of the the Grimm 64fs versus the Horus 256fs files I wrote about at length above remains virtually the same after conversion. This makes me wonder if the specific "sound" of the respective ADC makes more of a difference on the net result than the sampling rate?

 

Greetings from Switzerland, David.

Exactly. Sampling rate is a distinction (and probably has a different effect on different hardware chains), but seems to me each system should be viewed end-to-end, including resistance to "noise" at the input, internal electronics, quality / implementation of the ADC, etc. It could very well be a given converter might produce "better" output at DSD64 than another at DSD256.

That said, I do like the sound of the Horus better, in general. The Grimm sounds very warm and rounded ("like buttah"), but the Horus "opens up" the space in the hall in a way that is more lifelike, at least to my ears.

Gang, thanks for all these responses so far. Realize that some of your results are also due to the strengths/weaknesses of your DAC at those bit rates....so please, as I asked in the article, tell us your DAC playback system when responding.

This Mendelssohn recording will be released publicly next year.

Gang, thanks for all these responses so far. Realize that some of your results are also due to the strengths/weaknesses of your DAC at those bit rates....so please, as I asked in the article, tell us your DAC playback system when responding.

 

This Mendelssohn recording will be released publicly next year.

I originally listened to the samples on my iFi Micro iDSD, which handles all the sample rates. Second round using my LH Labs Pulse xFi, which currently supports up to DSD128.

Player each time was HQPlayer, set for "DirectSDM", i.e., no conversion / resampling, etc.

Unless I'm fooling myself (not very easy to do a blind test on these), I heard the same from each DAC ("That said, I do like the sound of the Horus better, in general. The Grimm sounds very warm and rounded ("like buttah"), but the Horus "opens up" the space in the hall in a way that is more lifelike, at least to my ears.").

Just seeing this.. Have HAPI with AD8P/DA8P cards. Will test it out this evening