[Rob Watts]

Just to clarify - I have only one 768k recording, and it has the same virtues of a 44.1k track that is M scaled - that is a cavernous sound stage, natural flow, and timbre variations. I have not (yet) taken a 768k recording, decimated it to 48k, then M scaled it back to 768k, to see what the overall losses are involved. This is something I certainly plan to do (actually since the decimation filter has been done I can actually do this now).

As too your second question, this is actually much more difficult to answer - as the 16 bit interpolation accuracy applies at the M scaler OP sampling rate (although in practice it would be 16 bit analogue accurate in terms of transient timing reconstruction) as the next WTA2 filter will continue the reconstruction to 16 bit accuracy, at ever finer OP resolution. As you double the SR, you reduce the area of the error by a factor of 4, or two bits, without adding more interpolation. But the issue with other DACs relate to the DAC analogue reconstruction accuracy, and R2R DACs are simply not fast enough to be accurate above 768; and conventional DACs with their typical 2.8 MHz noise shapers are not much better either. So switching glitches and/or noise shaper noise degrades accuracy.

 

[Chester Rockwell]

Very interesting, thanks. It just struck me as an interesting topic when I read that. Is 768k the foundation to reproducing the analog signal perfectly? As we know, there is always more to it than that, but it gets you thinking about the future of sound reproduction though, hence my question.

 

[PJDubyaM]

When we’re talking $4-5K on headphones or the M-Scaler, I have to do my research in hopes that I make an educated purchase that I’ll like.

...and I'm going to go the other direction. You already have a lot of headphones. I'm using the M Scaler just with a Mojo until my TT2 arrives (so not the full million-tap experience), and what I've found is that it makes it possible to enjoy LPs that before I found hard going. Stuff like Jlin's 'Black Origami' and Blanck Mass's 'World Eater'. Before M Scaler – even with my Empyreans – I found stuff like that a difficult listen. With the M Scaler, and it's a completely different story. I can't put my finger on how it's happened, or what's changed, but something about the M Scaler, even before a million taps are involved, just makes intense / wall-of-sound / lots-to-unpack music much easier to listen to. Just yesterday, gave Duke's 'Uingizaji Hewa' a try – it's Tanzanian singeli music, sitting somewhere between 180-300 bpm, and probably the most intense thing I've ever listened to. Loved it!

Given that you list listening to metal as one of your favoured genres, it might just be that the M Scaler would deconstruct / open out your music in the same sort of way?

 

[ZappaMan]

I’d say new headphones will give more immediate gratification. Just depends, if you like hearing the little details and subtleties. For casual listening, say you’re working, you’ll probably miss most of it anyway.
But, now I have aeon flow closed headphones, I can easily a/b the difference between passthrough.
Yesterday when I tried it, activating passthrough made the music seem tinny, with full taps, it’s sounded natural.
Easy enough to hear with anouar brahem , with his soaring instrumental style.

 

[jarnopp]

I’d say new headphones will give more immediate gratification. Just depends, if you like hearing the little details and subtleties. For casual listening, say you’re working, you’ll probably miss most of it anyway.
But, now I have aeon flow closed headphones, I can easily a/b the difference between passthrough.
Yesterday when I tried it, activating passthrough made the music seem tinny, with full taps, it’s sounded natural.
Easy enough to hear with anouar brahem , with his soaring instrumental style.

There is another option, not just new headphones or mScaler. You could add speakers to TT2 if you have the space, particularly if they are efficient and transparent like the Omega Super Alnico Monitors. That is actually quite an amazing presentation and even a bargain compared to TOTL phones at under $2000.

 

[Amberlamps]

Just to clarify - I have only one 768k recording, and it has the same virtues of a 44.1k track that is M scaled - that is a cavernous sound stage, natural flow, and timbre variations. I have not (yet) taken a 768k recording, decimated it to 48k, then M scaled it back to 768k, to see what the overall losses are involved. This is something I certainly plan to do (actually since the decimation filter has been done I can actually do this now).

As too your second question, this is actually much more difficult to answer - as the 16 bit interpolation accuracy applies at the M scaler OP sampling rate (although in practice it would be 16 bit analogue accurate in terms of transient timing reconstruction) as the next WTA2 filter will continue the reconstruction to 16 bit accuracy, at ever finer OP resolution. As you double the SR, you reduce the area of the error by a factor of 4, or two bits, without adding more interpolation. But the issue with other DACs relate to the DAC analogue reconstruction accuracy, and R2R DACs are simply not fast enough to be accurate above 768; and conventional DACs with their typical 2.8 MHz noise shapers are not much better either. So switching glitches and/or noise shaper noise degrades accuracy.

Question Rob,

Why does normal flac files and redbook use khz for sample rate and why does dsd go much higher and use megahertz instead, what benefit does mhz have over khz, if any ?

I’m just wondering, as I keep reading that dsd isn’t that great of a format. I would of thought that the more samples taken per second ( in the millions of hz per second for dsd and in the thousands of hz for normal RBCD and normal flac files ) the better the quality of the audio it would be.

But that doesn’t seem to be the case. So why doesn’t dsd 64 - 512 sound crazy better than normal 44.1khz - 384khz or 768khz.

I can’t figure out why millions of samples taken per second, doesn’t sound orders of magnitude better than tracks in the khz range in flac format.

I know that dsd 64 is 64 x 44.1khz and dsd 512 is 512 x 44.1khz, but the increase in samples taken per second doesn’t seem to sound better than what we already have. I don’t understand that ?

 

[Mediahound]

Why does normal flac files and redbook use khz for sample rate and why does dsd go much higher and use megahertz instead,

Same unit of measurement. 1000kHz = 1mHz

 

[AndrewOld]

Question Rob,

Why does normal flac files and redbook use khz for sample rate and why does dsd go much higher and use megahertz instead, what benefit does mhz have over khz, if any ?

I’m just wondering, as I keep reading that dsd isn’t that great of a format. I would of thought that the more samples taken per second ( in the millions per second for dsd and in the thousands for normal RBCD and normal flac files ) the better the quality of the audio it would be.

But that doesn’t seem to be the case. So why doesn’t dsd 64 - 512 sound crazy better than normal 44.1khz - 384khz or 768khz.

I can’t figure out why millions of samples taken per second, doesn’t sound orders of magnitude better than tracks in the khz range in flac format.

I know that dsd 64 is 64 x 44.1khz and dsd 512 is 512 x 44.1khz, but the increase in samples taken per second doesn’t seem to sound better than what we already have. I don’t understand that ?

DSD has a sampling rate of 2.88 MHz, which sure enough is 64 times redbook PCM (swings), but that’s only for 1 bit so it is 32768 times less resolution (roundabouts).

Here’s the wiki: https://en.m.wikipedia.org/wiki/Direct_Stream_Digital

My take on it is that it was for a short while a potentially good way of archiving analogue tape, but it is very difficult to record in DSD, because it is hard to mix and edit, so if you hunt around for true DSD recordings made from DSD masters you will find precious few. Most so called DSD releases have either analogue or PCM masters, and some of the rest went through analogue for editing.

 

[Amberlamps]

DSD has a sampling rate of 2.88 MHz, which sure enough is 64 times redbook PCM (swings), but that’s only for 1 bit so it is 32768 times less resolution (roundabouts).

Here’s the wiki: https://en.m.wikipedia.org/wiki/Direct_Stream_Digital

My take on it is that it was for a short while a potentially good way of archiving analogue tape, but it is very difficult to record in DSD, because it is hard to mix and edit, so if you hunt around for true DSD recordings made from DSD masters you will find precious few. Most so called DSD releases have either analogue or PCM masters, and some of the rest went through analogue for editing.

Cool, something I never do is read up about all of this, I just see bits and pieces in peoples posts and from my limited knowledge of the subject, I wonder why somethings are what they are.

Is 2.88mhz fixed or is that just for dsd 64 ? As onkyo player on my phone can be set at 5.6mhz, not that I use it, I have tried in the past but I noticed 0 difference.

I will have a look at the wiki link, I should of done it before I posted but I’m kinda too lazy, as at this time of night this website crawls for me on my ipad, I don’t know why, as on my phone it’s still fast.

Cheers

 

[Amberlamps]

Same unit of measurement. 1000kHz = 1mHz

Yes, but say for example 44.1khz, it is 44100hz, thats a normal RBCD track, dsd 64 is 2’880’000hz, but yet both tracks would basically sound the same. I thought that the fact that it was sampling many more times than RB, I thought it should sound much better.

Before I read andrews post, I wondered why dsd didn’t sound sooper dooper, but it looks like it’s the dsd recordings and technical problems that are the cause of why dsd doesn’t sound like an mscaler on steroids.

Is dsd likely going to be the future standard ? Anybody ?

 

[AndrewOld]

Yes, but say for example 44.1khz, it is 44100hz, thats a normal RBCD track, dsd 64 is 2’880’000hz, but yet both tracks would basically sound the same. I thought that the fact that it was sampling many more times than RB, I thought it should sound much better.

Before I read andrews post, I wondered why dsd didn’t sound sooper dooper, but it looks like it’s the dsd recordings and technical problems that are the cause of why dsd doesn’t sound like an mscaler on steroids.

Is dsd likely going to be the future standard ? Anybody ?

DSD is dead in the water. It never really began, but it is for sure end of life game over no chance historical interest only. Nobody is making DSD recordings. Well, very nearly almost just about nobody. Make a list of all the true DSD masters produced in the last year. Nearly nada.

 

[Amberlamps]

I was reading that wiki link, I forgot my fat playstation 3 was also a SACD capable player.

If no one is making dsd recordings or just a handful of them are, then surely it will die off, I wonder what will take over from RB, sooner or later some albums won’t fit on a 700mb CD, I’m not speaking about multi disc albums, I just mean that higher resolution tracks being hundreds of megabytes each.

I wonder what will take over from the cd, or will that die off since everything seems to be streaming nowadays, maybe albums will come on a micro sd card or flash drive in the future ?

 

[x RELIC x]

In the simplest of terms, DSD is 1bit (one value for every sample) so requires much higher sampling to play back a variety of sample values per second. PCM at CD quality is 16bit meaning there are much more value variations available (65,536) in each of the 44,100 samples per second.

 

[AndrewOld]

I was reading that wiki link, I forgot my fat playstation 3 was also a SACD capable player.

If no one is making dsd recordings or just a handful of them are, then surely it will die off, I wonder what will take over from RB, sooner or later some albums won’t fit on a 700mb CD, I’m not speaking about multi disc albums, I just mean that higher resolution tracks being hundreds of megabytes each.

I wonder what will take over from the cd, or will that die off since everything seems to be streaming nowadays, maybe albums will come on a micro sd card or flash drive in the future ?

DSD barely lived for a brief period; it has for sure died off, except in obscure corners of audiophile-land where of course it is touted as the holy grail. As far as CDs are concerned .. I haven’t bought one for years. Why would you? You can download or stream thousands of 24/96 or better masters from Qobuz or buy direct from record labels. You can’t ask for more than the master.

 

[Amberlamps]

DSD barely lived for a brief period; it has for sure died off, except in obscure corners of audiophile-land where of course it is touted as the holy grail. As far as CDs are concerned .. I haven’t bought one for years. Why would you? You can download or stream thousands of 24/96 or better masters from Qobuz or buy direct from record labels. You can’t ask for more than the master.

I agree when it comes to buying CD’s, the last time I bought an actual music cd was probably late 90’s, early 2000-2001, as lets not sugar coat it, probably every person here at somepoint grabbed a song here or album there from places, whether they knew it or not, and I don’t condone it, as it’s wrong and illegal.

Nowadays for 10, 20 or 25 quid a month you can get access to as much songs as anyone could want/need. Qobuz does me, even though I spent £20 quid last week to get a song from tidal as I couldn’t find it on qobuz, surprise surprise, it wasn’t on tidal either, now I wish I did use one of my throw away email addys to sign up for a months free trial, but it’s too much of a hassle, I just re’subbed to my old tidal account, it took me seconds but cost £20. We live and learn.

The last album I bought was from hdtracks and it was Otis Blue, Otis Redding.