Watts Up...?
post-14454242
Post #1,008 of 1,846

Alifatemi

New Head-Fier
Joined
Aug 30, 2018
Messages
6
Reaction score
3
Location
Tehran
Joined
Aug 30, 2018
Location
Tehran
Posts
6
Likes
3
Hello Rob. I have Qutest DAC and like to know more about M scaler; I have never liked upsampling either through hardware DAC or software like Roon and always ending got back to the original. Does M Scaler add something to the sound like say, photoshop that when upsample a low resolution file, it double or triple neighbor pixels to increase the resolution which is nonsense or what? Regads
 
     Share This Post       
post-14455733
Post #1,009 of 1,846
Joined
Apr 1, 2014
Messages
2,368
Reaction score
6,316
Joined
Apr 1, 2014
Posts
2,368
Likes
6,316
It depends upon what you define as the "original". By original you are thinking about the digital data; when I think about the original I am thinking about the analogue signal in the ADC that existed before the signal was sampled; and these are very different things. The original data is simply a snapshot of the analogue signal, and we need to recover the original continuous (that is not sampled) analogue signal. Fortunately, we can recover a bandwidth limited signal absolutely perfectly if we use a sinc function filter; but a sinc function filter requires an infinite amount of processing.

The M scaler WTA algorithm provides enough processing power with it's million taps so that it is identical to the ideal sinc function with better than 16 bit accuracy; so we recover the original analogue signal to better than 16 bits under all conditions.

Why is this needed? It's because the timing of transients is a vital requirement for psychoacoustics; we get the perception of pitch, timbre, soundstage and the perception of timing from transients; and regular filters have transient timing errors (the reconstructed transient is either a little too early or a little too late); but with the M scaler we can reconstruct the transients to a guaranteed better than 16 bit accuracy. This is the reason why the M scaler sounds so fundamentally different.
 
Chord Electronics Stay updated on Chord Electronics at their sponsor page on Head-Fi.
 
https://www.facebook.com/chordelectronics https://twitter.com/chordaudio http://www.chordelectronics.co.uk/
post-14456072
Post #1,010 of 1,846

Alifatemi

New Head-Fier
Joined
Aug 30, 2018
Messages
6
Reaction score
3
Location
Tehran
Joined
Aug 30, 2018
Location
Tehran
Posts
6
Likes
3
It depends upon what you define as the "original". By original you are thinking about the digital data; when I think about the original I am thinking about the analogue signal in the ADC that existed before the signal was sampled; and these are very different things. The original data is simply a snapshot of the analogue signal, and we need to recover the original continuous (that is not sampled) analogue signal. Fortunately, we can recover a bandwidth limited signal absolutely perfectly if we use a sinc function filter; but a sinc function filter requires an infinite amount of processing.

The M scaler WTA algorithm provides enough processing power with it's million taps so that it is identical to the ideal sinc function with better than 16 bit accuracy; so we recover the original analogue signal to better than 16 bits under all conditions.

Why is this needed? It's because the timing of transients is a vital requirement for psychoacoustics; we get the perception of pitch, timbre, soundstage and the perception of timing from transients; and regular filters have transient timing errors (the reconstructed transient is either a little too early or a little too late); but with the M scaler we can reconstruct the transients to a guaranteed better than 16 bit accuracy. This is the reason why the M scaler sounds so fundamentally different.

Thanks Rob. By original I mean original digital redbook or high res files I have without upsampling. I like their sound better compare to when I upsample them via Roon lab. Also I have done lots of comparison beetween redbook/ highres files and their counterpart MQA on Tidal. I have a very high resolution and trsnsparent system comprising wilson sasha ll and Audio research amps supplied by my Chord Qutest. As You said, with Qutest its very obvius that MQA sound qualit is inferior even to redbook files( 16/44.1) Mqa got far less dynamic, less vibrant and silky and silly softer sound which mitgh misslead many listeners but shows its real weakness in long term. Grant Green’s Street of Dreams is one of them. Tidal cd’s quality is even better than Tidal MQA. I dont have MQA DAC to reach my final verdict but at least with Qutest I can say I dont like MQA. Of course I kerp my mind open for futhure listening and comparison and post here. Cheers...
 
     Share This Post       
post-14456253
Post #1,011 of 1,846

AndrewOld

500+ Head-Fier
Joined
Jun 19, 2014
Messages
718
Reaction score
545
Joined
Jun 19, 2014
Posts
718
Likes
545
Thanks Rob. By original I mean original digital redbook or high res files I have without upsampling. I like their sound better compare to when I upsample them via Roon lab. Also I have done lots of comparison beetween redbook/ highres files and their counterpart MQA on Tidal. I have a very high resolution and trsnsparent system comprising wilson sasha ll and Audio research amps supplied by my Chord Qutest. As You said, with Qutest its very obvius that MQA sound qualit is inferior even to redbook files( 16/44.1) Mqa got far less dynamic, less vibrant and silky and silly softer sound which mitgh misslead many listeners but shows its real weakness in long term. Grant Green’s Street of Dreams is one of them. Tidal cd’s quality is even better than Tidal MQA. I dont have MQA DAC to reach my final verdict but at least with Qutest I can say I dont like MQA. Of course I kerp my mind open for futhure listening and comparison and post here. Cheers...
Err.. Grant Greens Street of Dreams was recorded in 1964 onto analogue tape, way before digital recording. No way it is an original digital red book recording. Not that i think MQA makes any sense, but with a recording like that you have no idea what you are listening to, apart from degraded 50 year old analogue tape, or copies thereof, so using it as a basis for comparison is far from ideal.

https://en.m.wikipedia.org/wiki/Street_of_Dreams_(Grant_Green_album)
 
Last edited:
     Share This Post       
post-14457048
Post #1,012 of 1,846

musickid

Headphoneus Supremus
Joined
Sep 9, 2016
Messages
3,888
Reaction score
1,797
Location
U.K.
Joined
Sep 9, 2016
Location
U.K.
Posts
3,888
Likes
1,797
Is the recovered analogue signal post 1M tap processing fully identical to the analogue signal in the ADC that existed before the signal was sampled? Or is there a small margin of error? I think i asked this way back just recapping here.
 
     Share This Post       
post-14457869
Post #1,013 of 1,846
Joined
Apr 1, 2014
Messages
2,368
Reaction score
6,316
Joined
Apr 1, 2014
Posts
2,368
Likes
6,316
Is the recovered analogue signal post 1M tap processing fully identical to the analogue signal in the ADC that existed before the signal was sampled? Or is there a small margin of error? I think i asked this way back just recapping here.
In principle yes; in practice it depends upon the ADC itself, as ADCs have aliasing problems, so the decimation is not done correctly. And once an aliasing error is created, it's impossible to repair it. But assuming there is no aliasing errors, and the ADC itself creates no distortion or noise floor modulation, then it would be fully identical to the analogue signal to within 16 bit accuracy. Unfortunately, modern ADCs are not capable of this, but I fully expect Davina to be able to easily do this.
 
Chord Electronics Stay updated on Chord Electronics at their sponsor page on Head-Fi.
 
https://www.facebook.com/chordelectronics https://twitter.com/chordaudio http://www.chordelectronics.co.uk/
post-14457941
Post #1,014 of 1,846

Paul Bjernklo

100+ Head-Fier
Joined
Jul 8, 2014
Messages
108
Reaction score
54
Location
London, UK
Joined
Jul 8, 2014
Location
London, UK
Posts
108
Likes
54
In principle yes; in practice it depends upon the ADC itself, as ADCs have aliasing problems, so the decimation is not done correctly. And once an aliasing error is created, it's impossible to repair it. But assuming there is no aliasing errors, and the ADC itself creates no distortion or noise floor modulation, then it would be fully identical to the analogue signal to within 16 bit accuracy. Unfortunately, modern ADCs are not capable of this, but I fully expect Davina to be able to easily do this.
Hi, I am a bit confused if the 16 bits referred to here is in any way related to the number of bits that the file was recorded at, ie does it mean that finally 16 bit recorded files come out with 16 bit accuracy, and not the much fewer bits accuracy that I assume applies for other “ordinary” DACs? If so does the same apply for any higher bit recordings like 24 bit and they also comes out with 16 bits accuracy with the M scaler (and if so what’s the point with hi res apart from higher sampling rate for any DAC?) or am I comparing Apples with Oranges?

Edit: If bits refer to transient timing and not signal amplitude as per a following post, what do the 16 bits refer to and what is a normal benchmark /average value to compare to? Thank you....
 
Last edited:
     Share This Post       
post-14457966
Post #1,015 of 1,846

castleofargh

Sound Science Forum Moderator
Joined
Jul 2, 2011
Messages
8,885
Reaction score
4,243
Joined
Jul 2, 2011
Posts
8,885
Likes
4,243
Hello Rob. I have Qutest DAC and like to know more about M scaler; I have never liked upsampling either through hardware DAC or software like Roon and always ending got back to the original. Does M Scaler add something to the sound like say, photoshop that when upsample a low resolution file, it double or triple neighbor pixels to increase the resolution which is nonsense or what? Regads
just to point out that picture interpolation is a very bad analogy with audio upsampling/oversampling. when creating more pixels out of a pic, there is no actual information about what those extra pixels should be. the ultimate result would be to visually mask the lines formed by the original pixels. so 1) it's about making something up. and 2) a fine subjective result is the ideal target.

with music, the recorded signal is well defined up to about half the sample rate. we know that it's all sines and sums of sines so we don't have to guess where an in-between point is going to be. we know!
adding extra samples clearly doesn't add information, it just presents the information we already have in a form that might be more convenient for some specific operations.
 
     Share This Post       
  • Like
Reactions: AndrewOld
post-14458201
Post #1,016 of 1,846

Samuel Snoopy

New Head-Fier
Joined
Oct 25, 2015
Messages
44
Reaction score
18
Location
Hong Kong
Joined
Oct 25, 2015
Location
Hong Kong
Posts
44
Likes
18
In principle yes; in practice it depends upon the ADC itself, as ADCs have aliasing problems, so the decimation is not done correctly. And once an aliasing error is created, it's impossible to repair it. But assuming there is no aliasing errors, and the ADC itself creates no distortion or noise floor modulation, then it would be fully identical to the analogue signal to within 16 bit accuracy. Unfortunately, modern ADCs are not capable of this, but I fully expect Davina to be able to easily do this.
Hi Rob, Is the 16bits accuracy talking about transient timing but not signal amplitude?
 
     Share This Post       
  • Like
Reactions: Paul Bjernklo
post-14458316
Post #1,017 of 1,846

Alifatemi

New Head-Fier
Joined
Aug 30, 2018
Messages
6
Reaction score
3
Location
Tehran
Joined
Aug 30, 2018
Location
Tehran
Posts
6
Likes
3
Err.. Grant Greens Street of Dreams was recorded in 1964 onto analogue tape, way before digital recording. No way it is an original digital red book recording. Not that i think MQA makes any sense, but with a recording like that you have no idea what you are listening to, apart from degraded 50 year old analogue tape, or copies thereof, so using it as a basis for comparison is far from ideal.

https://en.m.wikipedia.org/wiki/Street_of_Dreams_(Grant_Green_album)

Thanks Andrew for comment but; we got an analogue tape from which, we have 3 copies of different file format digitally sampled: one in 16/44.1, one in 24/192 and another in MQA. Whats wrong comparing them together? I didn say my files come from original digitally recorded master tape. Besides many 50s or 60s records are so great I cant find that qualty in todays recording, not even close. We just comparing 3 file format hopfully have come from one analogue master tape. Cant be more fair than this.
 
Last edited:
     Share This Post       
  • Like
Reactions: 514077
post-14458363
Post #1,018 of 1,846

maxh22

1000+ Head-Fier
Joined
Apr 18, 2016
Messages
1,333
Reaction score
424
Joined
Apr 18, 2016
Posts
1,333
Likes
424
I’ve noticed that with all of Rob’s dacs ,old reel to reel recordings and even some recordings that were recorded to cassette tape sound so real and lifelike. It really brings out the best of them I think.

With some ESS dacs they can sound bland and unmusical, AKM chipsets sound much more musical to me but lack a lot of the realistic properties I hear in Robs dacs including the Mojo.

Looking forward to Rob’s eventual Davina recodings :)
 
     Share This Post       
  • Like
Reactions: 514077
post-14458495
Post #1,019 of 1,846

AndrewOld

500+ Head-Fier
Joined
Jun 19, 2014
Messages
718
Reaction score
545
Joined
Jun 19, 2014
Posts
718
Likes
545
Thanks Andrew for comment but; we got an analogue tape from which, we have 3 copies of different file format digitally sampled: one in 16/44.1, one in 24/192 and another in MQA. Whats wrong comparing them together? I didn say my files come from original digitally recorded master tape. Besides many 50s or 60s records are so great I cant find that qualty in todays recording, not even close. We just comparing 3 file format hopfully have come from one analogue master tape. Cant be more fair than this.
But there is no way of knowing that they came from the same master tape. And even if they did, you have no way of knowing that the mastering was the same in terms of eq, level, and compression etc. Different versions get released by different mastering houses at different times quite possibly from copy masters using different mastering. The only valid comparison to do is to take a known 24/192 master (say), compare it to itself downsampled to 16/44.1 (say) which is easy because you can use sox, or J River or any resampler you like, and also convert it to MQA and back which alas you can’t do because of the secret and proprietary nature of MQA.
 
     Share This Post       
post-14458535
Post #1,020 of 1,846

514077

Headphoneus Supremus
Joined
Mar 9, 2013
Messages
3,711
Reaction score
1,555
Joined
Mar 9, 2013
Posts
3,711
Likes
1,555
Thanks Andrew for comment but; we got an analogue tape from which, we have 3 copies of different file format digitally sampled: one in 16/44.1, one in 24/192 and another in MQA. Whats wrong comparing them together? I didn say my files come from original digitally recorded master tape. Besides many 50s or 60s records are so great I cant find that qualty in todays recording, not even close. We just comparing 3 file format hopfully have come from one analogue master tape. Cant be more fair than this.
How did you perceive the diference between your copy of 24/192 vs CD quality? I don't bother or worry about MQA, as Chord doesn't use it, and I don't have any other DACS. Oh yes; welcome to Head-fi.
 
     Share This Post       

Users Who Are Viewing This Thread (Users: 1, Guests: 1)

  • Top