[musickid]

Already lost.

 

[Rob Watts]

No I am afraid once timing errors are introduced then it's impossible to recover. In the case of DSD it's down to the delay through the modulator - small amplitude transients (like a step waveform, or Heaviside function) have a longer delay than large signal step waveforms. And it's impossible to repair this kind of damage.

 

[Peter Hyatt]

I recently bought some 50GB of music - I tend to go on a spree every 4 months or so - and I bought a lot of Max Richter, half of which was 96 kHz. Sure the HD versions are generally better, but not enormously so. And the issue, I am convinced, is not the increased bandwidth - I can't hear above 15k anyway - but the precision of the transient reconstruction. And from this POV, doubling the sample rate only quarters the innate error - or two bits worth. And 96k sounds nothing like as good as a native 768k recording; and for sure 192k alone against an M scaled RBCD, the RBCD would easily win from a SQ POV.

For sure, Davina will answer these questions (bandwidth limiting, aliasing issues and how well M scaling works against the original). But we are going to have to wait some time for these recordings and tests! A huge number of fundamental questions will get answered with this project.

Mari Samuelson?

Which Richter releases did you get?

 

[Rob Watts]

Here is the cover art:











The first (44.1 24b) and last one (96k) I guess is the best of the bunch.

I also got this which has one Richter track too, and is 96k:



Also to strongly consider are these box sets:













As these box sets are typically 4GB of RBCD standard and only £9.75 for each box set. The SQ is excellent too, with (to my limited musical expertise) fine musicianship.

I have bought a lot of Brilliant Classics box sets, and they have never disappointed. I also got some CD box sets too, which are a bit more expensive and a pain to rip but worth the effort:







I also got 3 Naxos CD's of James Whitbourn, of which the best is:

 

[bhobba]

But what do I know? We can't know anything for certain until we perform rigorous SQ tests. And that gets me nicely back to Davina, when we can actually do these tests, and will know this for certain.

Having heard your M-Scaler and TT2 direct connected to speakers, the reaction of myself and other audiophiles, I think you know more than designers of other DAC's. But exactly what was heard will need to wait until I do my write up. It's 4.30 in the morning here in Aus so I don't want to write it up now - probably sometime tomorrow - but there is one issue I want to clarify.

Yes it's about reconstructing the bandwidth limited signal. And I maintain that that is all one needs to do; my view is that properly bandwidth limiting to 20kHz would be perfect SQ wise. But what do I know? We can't know anything for certain until we perform rigorous SQ tests. And that gets me nicely back to Davina, when we can actually do these tests, and will know this for certain.

I think I may have been misunderstanding what you have been writing. I thought you were saying that your very high quality up/down sample rate converters preserve timing much better but you are still limited by the timing in the down-sampled material. By increasing the down-sampled frequency the timing that is available to preserve is better, The point I have been trying to get across is very few recordings have any content other than basically noise above 48khz so going any higher than 96k sampling in what is being recorded will not give better timing information. Of course it is different to the input - but since its only ultrasonic noise that has been removed it makes no audible difference to timing information or anything else. However I think I misunderstood you. You think 20k is all that needs to be down-sampled to, and it will still maintain proper timing. I cant understand this because timing information is definitely there above 20k and it is above noise.

Can you perhaps clarify? I have found 96k sounded better through the M-Sample and TT2 (but not by a lot). Admittedly we only tried a few - some Dianna Krall, Nat King Cole and the Rat Pack.

Thanks
Bill

 

[csglinux]

Hi Rob,

Could you recommend just one or two good-quality tracks you'd recommend as demos for the M Scaler? Ideally, would you have recommendations for tracks you yourself use when listening to the sound-stage depth with those changes at -gazillion dBFS?

 

[STR-1]

(@Rob Watts) Rob, will you be at CanJam London this year?

 

[Rob Watts]

Having heard your M-Scaler and TT2 direct connected to speakers, the reaction of myself and other audiophiles, I think you know more than designers of other DAC's. But exactly what was heard will need to wait until I do my write up. It's 4.30 in the morning here in Aus so I don't want to write it up now - probably sometime tomorrow - but there is one issue I want to clarify.



I think I may have been misunderstanding what you have been writing. I thought you were saying that your very high quality up/down sample rate converters preserve timing much better but you are still limited by the timing in the down-sampled material. By increasing the down-sampled frequency the timing that is available to preserve is better, The point I have been trying to get across is very few recordings have any content other than basically noise above 48khz so going any higher than 96k sampling in what is being recorded will not give better timing information. Of course it is different to the input - but since its only ultrasonic noise that has been removed it makes no audible difference to timing information or anything else. However I think I misunderstood you. You think 20k is all that needs to be down-sampled to, and it will still maintain proper timing. I cant understand this because timing information is definitely there above 20k and it is above noise.

Can you perhaps clarify? I have found 96k sounded better through the M-Sample and TT2 (but not by a lot). Admittedly we only tried a few - some Dianna Krall, Nat King Cole and the Rat Pack.

Thanks
Bill

Yes a 96k file will preserve timing better than a 48k; but I guess I ought to talk about what I mean by preserve timing. So to go from 96k to 48k we have to bandwidth limit so that energy at 24k is zero; then we can decimate (throw away half the samples) to give us 48k. There are a number of issues with this - the first one being the bandwidth limiting and decimation.

So I said that we need to ensure that 24k the signal is 0; but modern ADCs are poor in this regard, as they use half band filters because they are cheap; this means the level at 24k is at -6dB, not an infinite attenuation. So we will get aliasing, and this will be signal components that are at the wrong frequency and around 20-24 kHz. These errors are very audible because in themselves they will change the timing of transients, and this is audible, even though the aliasing components around 20-20kHz is not by themselves audible. So a 96k recording will sound better than the 48k version, simply because of the poor quality of decimation filters. But this problem is definitely fixable - Davina project has 220dB rejection filters, so it won't suffer from this issue at all.

Another issue is that a 96k master file is often converted to 44.1k. This involves sample rate conversion (SRC) and because this is non-integer conversion, will always create more THD and noise - it's impossible to do it without adding distortion. Moreover, the noise added is slew rate related, so we get noise floor modulation dependent upon the high frequency content.

But my first comment was that a 96k file would preserve timing better than 48k - and by preserve timing I am not talking about the absolute shape or timing of transients. So bandwidth limiting to 24k (for 48k sample rate) from 48k (for 96k sample rate or FS) will changing the timing of transients, as of course the slopes will be less steep. But of itself I am not worried at all about this, as it is a linear effect; the delays are consistent, and not varying. The problems of timing are when the delays change with signal, and this is the aspect that destroys sound quality. We get this problem when the DAC converts from the sampled data back to the continuous signal that we need to listen to the music. This is when the reconstruction filter comes into play, and we can reduce this issue by getting the filter as close as possible to a sinc function - then we can recover the bandwidth limited analogue signal with no timing errors, that is with all the delays being independent of the signal.

I agree with you about high frequency content - if you look at real recordings, once you get to 20kHz it's falling off steeply; at 20k is typically -60dB down. By the time you get to 50 kHz, all that is present is noise and distortion from the ADC - which actually needs to be filtered out (hence my HF filters on my DACs).

I hope this clarifies a rather complex situation for you.

Hi Rob,

Could you recommend just one or two good-quality tracks you'd recommend as demos for the M Scaler? Ideally, would you have recommendations for tracks you yourself use when listening to the sound-stage depth with those changes at -gazillion dBFS?

My depth track is from this album and you can get it from:
https://www.prestomusic.com/classic...-an-exhibition-three-movements-from-petrushka

(@Rob Watts) Rob, will you be at CanJam London this year?

Yes I will! I plan to do a couple of seminars about filter design, and DAC design. I am also doing CanJam SoCal - flying out to LA this Tuesday.

 

[csglinux]

My depth track is from this album and you can get it from:
https://www.prestomusic.com/classic...-an-exhibition-three-movements-from-petrushka

Got it - many thanks Rob. Which is your specific track for testing depth? I presume one of the Stravinsky tracks?

QUOTE="Rob Watts, post: 15009506, member: 394072"]
I am also doing CanJam SoCal - flying out to LA this Tuesday.[/QUOTE]
Thanks also for your presence in SoCal this year. I know it will be much appreciated. Would you, by any chance, have any time outside of your allotted seminar/panel sessions to answer lots of questions about DAC filters?

 

[Rob Watts]

It's the Mussorgsky: Pictures at an Exhibition I. Gnomus track.

You can't believe how many times I have used that track over and over and over to listen to the subtlest changes in depth...

Sure I will be available to chat - either at Chord's booth or after the seminars (I have one of my own and two Q&A sessions too).

 

[bhobba]

I agree with you about high frequency content - if you look at real recordings, once you get to 20kHz it's falling off steeply; at 20k is typically -60dB down. By the time you get to 50 kHz, all that is present is noise and distortion from the ADC - which actually needs to be filtered out (hence my HF filters on my DACs). I hope this clarifies a rather complex situation for you.

Wow - thanks or the very enlightening reply - it clarified a lot of things.

Now to pen my review - I stayed up too late last night and will be watching the cricket at 6.30 which, here in Aus, is about right now, so may not be finished for a bit.

Thanks
Bill

 

[bhobba]

This is my review of the M-Scalar and TT2 direct connected to some 89db speakers as per a previous post.

First this is one of the two best DAC's I have ever heard - I think actually the best - and I have heard and owned quite a few over the years. The other DAC is the Grandinote - but IMHO this beats it. Others present have heard another DAC they put in the same category - the top of the line MSB - but the price is way too high for me - I think someone said its the second most expensive DAC in the world.

OK what were the stand out characteristics of this setup? First thing I noticed was the volume was more than adequate - those 18W's are fine for normal listening. Then was the detail - I was hearing things in recordings I never heard before - it was also fast without any edge. The soundstage extended way beyond the boundaries of the speakers and did the good old speaker vanishing trick. It was very 3D with depth and spaciousness I had never heard before. Vocals were simply outstanding - you felt like you could reach out and touch Dean Martin.

There was nothing at all to dislike about it and considering you do not need an amp fantastic value - the other two DAC's with matching amps were at least over $10k more expensive.

My dealer builds and designs speakers and will be building a about 94db one specially for this DAC and amp.

My opinion - this is the best DAC I have yet heard. Others put it in their top 3.

Thanks
Bill

 

[csglinux]

It's the Mussorgsky: Pictures at an Exhibition I. Gnomus track.

You can't believe how many times I have used that track over and over and over to listen to the subtlest changes in depth...

Sure I will be available to chat - either at Chord's booth or after the seminars (I have one of my own and two Q&A sessions too).

Awesome - thanks Rob!

 

[musickid]

For Rob

Could i ask a very basic question: normally i get the overall jist of what is being described here but i find myself stuck on the following two quotes which seem contradictory to me?

"Yes a 96k file will preserve timing better than a 48k"

"if you look at real recordings, once you get to 20kHz it's falling off steeply; at 20k is typically -60dB down. By the time you get to 50 kHz, all that is present is noise and distortion from the ADC."

If anything above 50k is noise and distortion from the ADC how can a 96k file preserve timing better than a 48k file? Why bother with any recording above 50k then? Many thanks MK.

 

[musickid]

As an ex violinist i like this one by Brilliant Classics.