|
Post by Clive on May 27, 2019 18:29:06 GMT
I've been intrigued by the Chord M Scaler, there've been quite a number of good reports about it with its 768kHz upsampling in Chord DACs such as my Qutest. At 3,495 quid I didn't fancy splashing out. I also saw the Rob Watts video on it, which was compelling. Then someone said, try HQPlayer and this does much the same thing in software for a whole lot less money, though it's still around 230 quid. Much as Rob Watts says in the video, returning to normal sample rates is quite a let down. 768kHz into my Qutest has been a revelation. Music is more alive and possesses great presence. Following Oscar Peterson on piano...just one small example - the individual notes are so very well defined. I don't think I've heard this level of "realness" and musicality in my room before. Certainly my Qutest has take a significant set up. HQPlayer
|
|
|
Post by MartinT on May 27, 2019 18:48:15 GMT
I was impressed when I heard the mScaler. However, it's not cheap.
I'm ambivalent about upsampling. Having tried it in Volumio, I returned to native resolution and just let my LKS do its job with the data. I think it's better without the upsampling.
|
|
|
Post by John on May 27, 2019 19:12:38 GMT
I think this might be source dependent I hear it outperforms Jplay Fermto which when I heard it I was very impressed. I have been trying to figure out how to use it on the SOtM to see if it makes a difference on that as the SOtM works as a Network Audio Adapter But so far not been able to get it to work. Looking forward to hearing what it does when I next visit Vic
|
|
|
Post by Clive on May 27, 2019 19:14:43 GMT
I upsampled with Volumio on the Pi and Sparky...this is a totally different effect. Some of it is due to the Chord DAC being able to take a high sampling rate.
It's worth watching the Rob Watts video even if it is 28 mins long. The effect is probably very similar to M Scaler.
|
|
|
Post by Clive on May 27, 2019 19:23:37 GMT
Yes John, it does indeed improve on Jplay Femto, it's the greatest improvement I've heard with digital for a long time.
I'm scared about playing a record as I think LP will seriously struggle. Never thought I'd find digital would outclass vinyl...equal with differences yes but ..better.. will be difficult to take.
|
|
|
Post by John on May 27, 2019 19:32:05 GMT
That some statement Clive I think I figured how to use it on my SOtM so will have a play Tuesday night
|
|
|
Post by Clive on May 27, 2019 19:38:58 GMT
John, what sample rate will your DAC accept? I can't say how much of what I'm hearing is due to the Chord DAC benefiting from 768khz vs any other DAC which can take the same sample rate. Btw HQPlayer can deliver up to 1.5M!
|
|
|
Post by John on May 27, 2019 19:51:03 GMT
I think it will only go to 192 so perhaps not worth the effort At some point, I might get a dac that will go up to 768 but no idea when
|
|
|
Post by MartinT on May 27, 2019 20:15:45 GMT
I'm scared about playing a record as I think LP will seriously struggle. I reached that point six months ago, Clive. There was no turning back. SQ beats nostalgia for me.
|
|
|
Post by Clive on May 27, 2019 20:44:26 GMT
I'm scared about playing a record as I think LP will seriously struggle. I reached that point six months ago, Clive. There was no turning back. SQ beats nostalgia for me. I've been there or thereabouts since moving to Jplay Femto then with Qutest. Now pushing Qutest to the limit the step up isn't small. It's as though I'm using a very high end DAC. There are many ways to achieve this but this is a good and affordable one... certainly it's much better value than an M Scaler.
|
|
|
Post by Clive on May 28, 2019 7:54:19 GMT
For those who CBA to watch the Rob Watts video in the OP as it's 28 mins long, here's a major point I took from it. Whether he's correct, I'm not qualified to say.
The contention is that the high resampling (upscaling?) rate of M Scaler properly reproduces transients. OK, I'm not using M Scaler but I am doing pretty much then same in software. I'm not saying I have the best digital sound, what I have though is the biggest step up with digital sound I've heard in my own setup - which is of course the one I know best. Hearing other good kit in show environments doesn't really tell me much, eg the latest Innuos server was fabulous but I can't say how it compares to my system.
|
|
|
Post by jandl100 on May 28, 2019 9:58:36 GMT
Hmm, it's still "making it up". The higher rez is faked. It might sound better, but whether it is more realistic and true to the original sound is perhaps open to debate. Perhaps it's one version of the truth!
|
|
|
Post by Clive on May 28, 2019 10:13:05 GMT
Hmm, it's still "making it up". The higher rez is faked. It might sound better, but whether it is more realistic and true to the original sound is perhaps open to debate. Perhaps it's one version of the truth! I'm not sure, is it faked? The higher sampling rate isn't making up data any more than 44.1 is, it's just linking adjacent data points, yes there are more of them as they are closer together in time but the shape of the curve should be accurate. The main point from Rob Watts is that the start of the waveform is more accurately reproduced, hence the improved transients. He claims this is mathematically proven. He used some simple graphs in the video to illustrate this, I can't say whether this is all kosher but he's more reliable than some "experts" (in that he is actually an expert). All I can say is that from three of us who've installed the HQPlayer software over the weekend (all Chord Qutest users), we are all very impressed by the change and it's not a small change. Is it more accurate? I can't say.
|
|
|
Post by jandl100 on May 28, 2019 10:20:15 GMT
If something is recorded at, say, 44KHz and that is mapped onto a 500KHz (or whatever) file using interpolation algorithms (presumably), then that certainly seems to be using its imagination - the data simply wasn't there on the recording. You may be able to say that the result is a plausible approximation, but one of many depending on the interpolations algorithms used. Maybe I'm missing something.
EDIT: I don't doubt that it sounds "better".
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on May 28, 2019 10:21:08 GMT
IMHO the only way correct genuine high resolution is produced is to actually record at that particular sample/bit rate from the outset. A few year ago I wrote an up-sampling program to compliment a HQ player I was toying with at the time, yes it worked really well on red book multiples, however you had the extra current draw of the processors and hardware to calculate the math's and manipulate the PCM signals. Granted the chord used PGFA's to accomplish this I believe, I remember David Vivian enthusing about a recent visit he had just returned from the Chord factory after being given the 768Khz treatment lol
I understand why a lot of folks are impressed with it especially if you have not listened to a quality digital set up before, however in the last 18 months we have had traded in seven Chord Dave's including two M-scaler / Dave packages with the comments, the wow factor has worn off, the seduction has left the building and there is not much to engage us please help! Of coarse YMMV
Thinking about this a little more, using the devices usually incorporates some for of extra PLL or re-clocking architecture as well as adding dither, this is possible one of the reasons that the perception is given of improved SQ?
Fro me perspective actually using a 24bit depth as opposed to 16 bits imho delivers much more palpability and realism, as these are log scale not linear, so even if we stayed with the red book sample rate of 44100 samples per second you not have a resolution window that is 24 x 24 bits as opposed to 16 x 16 bits.
Either way you could knock up a decent up-sampling device from a couple of chipsets, quality power supply and some nice 75 Ohm bnc cables, infact Clive I am surprised Clive you have not attempted this yet? Personally I would be looking at improving the clock architecture and power supply lines to the clock.
Interestingly if you switched in say a pro audio SRC (Drawmer/Apogee etc) even at the red book the sound would still improve, again this I feel is due to the extra PLL circuitry inside?
Just a few thoughts
|
|
|
Post by Clive on May 28, 2019 10:43:42 GMT
If something is recorded at, say, 44KHz and that is mapped onto a 500KHz (or whatever) file using interpolation algorithms (presumably), then that certainly seems to be using its imagination - the data simply wasn't there on the recording. You may be able to say that the result is a plausible approximation, but one of many depending on the interpolations algorithms used. Maybe I'm missing something. EDIT: I don't doubt that it sounds "better". I'm at the edge or beyond the edge of my understanding here so I make no claims for being correct. I think higher res also gives you more samples per second in the audio band. Even with 44.1 two data points have a line drawn between them - is this "accurate"? When 768k sampling is applied the extra data points sit on the line drawn between the two 44.1 data points, so in reality the resulting line of the graph is no different. For some mathematical reason transients are more accurately reproduced, maybe because when they occur on 44.1 they might have just missed a sample so the next one has to "catch up". This is less likely to occur with extra samples. I've seen it said that the ear is more sensitive to timing than 44.1 resolution provides. This is where people argue all sorts....obviously I don't know, I doubt many if any here do either. As Mr Flux says, the effect may wear off over time, at least I've not splashed out on an M Scaler. I do though find I have gained musical insight, the Oscar Peterson piano example - it is so much better, individual notes are much easier to distinguish and make more musical sense now. I don't believe my original setup was in anyway a poor one, indeed it was pretty good (I have heard a lot).
|
|
|
Post by Clive on May 28, 2019 10:51:30 GMT
Either way you could knock up a decent up-sampling device from a couple of chipsets, quality power supply and some nice 75 Ohm bnc cables, infact Clive I am surprised Clive you have not attempted this yet? Personally I would be looking at improving the clock architecture and power supply lines to the clock.
I have tried a few ways to resample or re-constitute the USB data mostly with good results but these have been at much lower sampling rates. Installing a piece of software is so much easier...sand the benefits were much smaller. Anyway time will tell, maybe I'll become bored. I've not bought it yet, this has all cost me nothing but a small amount of time. It takes about 4 mins to install (once downloaded). It takes longer to suss out the software but it's quite easy really.
|
|
|
Post by petea on May 28, 2019 10:54:58 GMT
|
|
|
Post by MartinT on May 28, 2019 10:59:20 GMT
The main point from Rob Watts is that the start of the waveform is more accurately reproduced, hence the improved transients. Now that I do understand, and it's less to do with the resolution/sample rate and more to do with the digital reconstruction filter used. It's why I had my previous Ayre SACD player upgraded with the Minimum Phase filter and why I select SLOW-M (Minimum Phase Slow roll-off) in my current LKS DAC. It better emulates transient leading edges, as opposed to traditional DACs that have a pre-ring before the actual transient. Below is an example diagram from Wolfson showing how MP better simulates the real-world transient. The reality is that it back-loads more ringing than the Wolfson diagram shows, but it does sound more natural to the ear.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on May 28, 2019 14:00:35 GMT
To FIR or not to FIR that is the question, whether to suffer the outrageous post and pre-ringing of fortunes of anti phase distortion filtering or just plump for a Wadia filter 'A' that is the question?
|
|