Music Player Network
Previous Thread
Next Thread
Print Thread
Page 2 of 7 1 2 3 4 5 6 7
Joined: Jan 2000
Posts: 1,032
Platinum Member
Offline
Platinum Member
Joined: Jan 2000
Posts: 1,032
Quote:
Originally posted by gm:
Quote:
Originally posted by 3D Audio:
I'll have to re-record them. They aren't on the drive anymore. [...]

I'll try to record samples of as many of them as possible. The format is RADAR with Nyquist converters. I'll do one at 24.48 and the other at 24/96. I can then transfer the files to Masterlink when we're done.
ah-HA!!! You're not testing only differences due to sampling rate changes! You're testing different filters! You testing different clocks!
Wait a minute. You're saying that recording the same material to two different but identical recorders at 48K and 96K is an inadequate way to compare sampling frequencies?

I'm not contesting that at all. I'm eager to learn.

The differences I heard were real and instantly recognizable. To me at least, and the second on the session.

So George, to accurately compare 48K and 96K, and exclude the factors you mention, how would YOU set up the comparison? I have the source material and the recorders set to go for Tuesday. How do you duplicate filters and clocks for two different Fs? Is there any way to remove the other variables?

More importantly, if recording at 96K on the same recorder sounds different than recording on it at 48K, then should I care what is causing the difference? If it sounds better, it is better, right?

Please don't think I'm being argumentative. You know me--always digging for more information and quantifiable evidence. I am eager to learn.


Lynn Fuston
3D Audio Inc
Home of 3dB
Joined: Oct 2000
Posts: 649
K
Gold Member
Offline
Gold Member
K
Joined: Oct 2000
Posts: 649
Quote:
Originally posted by Imagine:
I'd be interested in knowing your results...

Dean
In general, 96 and 192 have better transients on "easy to make horrible" sounds like triangle and keys. Overall clarity and better soundstage, too.


KK Proffitt
Chief Audio Engineer, JamSync, Nashville
JamSync
Joined: Feb 2001
Posts: 1,041
V
Platinum Member
Offline
Platinum Member
V
Joined: Feb 2001
Posts: 1,041
Recording triangle is very important.

Joined: Oct 2000
Posts: 649
K
Gold Member
Offline
Gold Member
K
Joined: Oct 2000
Posts: 649
Quote:
Originally posted by gm:

Sorry Lynn, my dear friend, I really don't believe you'd be able to hear the difference between sample rates on electric bass guitar in a supervised double-blindfold test. I just don't. And, I'd be willing to put money on it. A lot of money. Any takers?


George[/QB]
Thumb/pop, fretless, finger picked or pick style? Maybe with thumb/pop or pick style....it would be an interesting test. Hello...Larry Graham?


KK Proffitt
Chief Audio Engineer, JamSync, Nashville
JamSync
Joined: Feb 2000
Posts: 2,184
gm Offline
MP Hall of Fame Member
Offline
MP Hall of Fame Member
Joined: Feb 2000
Posts: 2,184
Quote:
Originally posted by 3D Audio:
Quote:
Originally posted by gm:
Quote:
Originally posted by 3D Audio:
I'll have to re-record them. They aren't on the drive anymore. [...]

I'll try to record samples of as many of them as possible. The format is RADAR with Nyquist converters. I'll do one at 24.48 and the other at 24/96. I can then transfer the files to Masterlink when we're done.
ah-HA!!! You're not testing only differences due to sampling rate changes! You're testing different filters! You testing different clocks!
Wait a minute. You're saying that recording the same material to two different but identical recorders at 48K and 96K is an inadequate way to compare sampling frequencies?

I'm not contesting that at all. I'm eager to learn.

The differences I heard were real and instantly recognizable. To me at least, and the second on the session.

So George, to accurately compare 48K and 96K, and exclude the factors you mention, how would YOU set up the comparison? I have the source material and the recorders set to go for Tuesday. How do you duplicate filters and clocks for two different Fs? Is there any way to remove the other variables?

More importantly, if recording at 96K on the same recorder sounds different than recording on it at 48K, then should I care what is causing the difference? If it sounds better, it is better, right?

Please don't think I'm being argumentative. You know me--always digging for more information and quantifiable evidence. I am eager to learn.
Lynn,

And you know me, I love arguments.

No, I'd record at the highest possible clock rate - the higher the better. Then I'd filter that sample using a very precise FIR (off-line, of course).

If I had a 96k original, I'd filter down to 48k. Mind you, both of the samples run at 96k sample-rate. The filtered one, however is in every way indistinguishable from a 48k-sampled signal, save that it's clocked at 96k.

Put both the samples on, say, a Masterlink disk...et, voila! See if you can hear the difference.

By the way, this is exactly the test that the Studio Production and Practices Tech Committee fashioned. It's up on a web site if you have a Masterlink, a fast connection and a computer with a CD-R burner and you'd like to check it out.

George


George Massenburg

http://www.massenburg.com
Joined: Mar 2000
Posts: 7,950
MP Hall of Fame Member
Offline
MP Hall of Fame Member
Joined: Mar 2000
Posts: 7,950
Quote:
Originally posted by gm:
You know, I hate to be the sand in everyone's KY
good thing i use astroglide.


alphajerk
FATcompilation
"if god is truly just, i tremble for the fate of my country" -thomas jefferson
Joined: Jan 2000
Posts: 1,032
Platinum Member
Offline
Platinum Member
Joined: Jan 2000
Posts: 1,032
Quote:
Originally posted by gm:
[QUOTE]By the way, this is exactly the test that the Studio Production and Practices Tech Committee fashioned. It's up on a web site if you have a Masterlink, a fast connection and a computer with a CD-R burner and you'd like to check it out.
Yes, yes, yes, yes and yes.

URL, please.


Lynn Fuston
3D Audio Inc
Home of 3dB
Joined: Apr 2001
Posts: 590
Gold Member
Offline
Gold Member
Joined: Apr 2001
Posts: 590
Quote:
Originally posted by 3D Audio:
Quote:
Originally posted by gm:
[QUOTE]By the way, this is exactly the test that the Studio Production and Practices Tech Committee fashioned. It's up on a web site if you have a Masterlink, a fast connection and a computer with a CD-R burner and you'd like to check it out.
Yes, yes, yes, yes and yes.

URL, please.
||: yup :||

I'd very much like to hear this too.

-dg


DAVE GREENBERG
SONOPOD MASTERING
Joined: Jun 2003
Posts: 12
M
Member
Offline
Member
M
Joined: Jun 2003
Posts: 12
Quote:
Originally posted by gm:
No, I'd record at the highest possible clock rate - the higher the better. Then I'd filter that sample using a very precise FIR (off-line, of course).

If I had a 96k original, I'd filter down to 48k. Mind you, both of the samples run at 96k sample-rate. The filtered one, however is in every way indistinguishable from a 48k-sampled signal, save that it's clocked at 96k.
Sure, tested that way, you are no doubt right about the differences. However, when we set our A/D converters to record at 96khz vs. 44.1, we are not using that same process you describe. Wasn't the original discussion more about the clock rate we set our converters to in order to get the best sound? Since with some converters, clocking at higher rates sounds better, why not do it?

Joined: Feb 2000
Posts: 2,184
gm Offline
MP Hall of Fame Member
Offline
MP Hall of Fame Member
Joined: Feb 2000
Posts: 2,184
Quote:
Originally posted by Mark_W:
Quote:
Originally posted by gm:
No, I'd record at the highest possible clock rate - the higher the better. Then I'd filter that sample using a very precise FIR (off-line, of course).

If I had a 96k original, I'd filter down to 48k. Mind you, both of the samples run at 96k sample-rate. The filtered one, however is in every way indistinguishable from a 48k-sampled signal, save that it's clocked at 96k.
Sure, tested that way, you are no doubt right about the differences. However, when we set our A/D converters to record at 96khz vs. 44.1, we are not using that same process you describe. Wasn't the original discussion more about the clock rate we set our converters to in order to get the best sound? Since with some converters, clocking at higher rates sounds better, why not do it?
Really good question. And the answer is specific & important, which is: so that we understand what we're listening to, to better inform our choices.

I don't disagree with you liking a given converter better, 48, 96 or whatever. In fact, our job is to chooses from among available sound choices - whether that means the best possible sound, or some specific textural quality.

What I'd like to point out is that you're better served to make these sonic choices, 1> blindfolded, with for instance as little attention paid to that shiny, hi-tech front front panel, & 2> without being swayed by numbers, such as 48, 96, 192 and whatever. It should be self-expanatory why.

I've said elsewhere that my favorite front end even at 48k is at this moment either a Meitner or dCS DSD front-end, resampled with a high-precision resampler.

George


George Massenburg

http://www.massenburg.com
Joined: Feb 2002
Posts: 479
Senior Member
Offline
Senior Member
Joined: Feb 2002
Posts: 479
I have never recorded a bass guitar @ 192k Only vocal's and acoustic instruments. I can for sure say that not only is there more sonic clarity but recording headroom is improved a lot over 44.1/48.

Joined: Nov 2002
Posts: 24
J
Member
Offline
Member
J
Joined: Nov 2002
Posts: 24
George, may i make a sort of related comment. in the TC 6000 your EQ runs at 96k all the time and uses 3 of the 4 engines, not leaving much room in the box for other functions. what is the reason that it does not run at 48k if what i am reading above is correct, you are not endorsing 96k or higher for any better sonic qualities. if i have read your intentions incorrectly i apologize to all.

thanks john

Joined: Jan 2000
Posts: 1,032
Platinum Member
Offline
Platinum Member
Joined: Jan 2000
Posts: 1,032
Quote:
Originally posted by gm:
Quote:
Originally posted by Mark_W:
Quote:
Originally posted by gm:
No, I'd record at the highest possible clock rate - the higher the better. Then I'd filter that sample using a very precise FIR (off-line, of course).

If I had a 96k original, I'd filter down to 48k. Mind you, both of the samples run at 96k sample-rate. The filtered one, however is in every way indistinguishable from a 48k-sampled signal, save that it's clocked at 96k.
Sure, tested that way, you are no doubt right about the differences. However, when we set our A/D converters to record at 96khz vs. 44.1, we are not using that same process you describe. Wasn't the original discussion more about the clock rate we set our converters to in order to get the best sound? Since with some converters, clocking at higher rates sounds better, why not do it?
Really good question. And the answer is specific & important, which is: so that we understand what we're listening to, to better inform our choices.

I don't disagree with you liking a given converter better, 48, 96 or whatever. In fact, our job is to chooses from among available sound choices - whether that means the best possible sound, or some specific textural quality.

What I'd like to point out is that you're better served to make these sonic choices, 1> blindfolded, with for instance as little attention paid to that shiny, hi-tech front front panel, & 2> without being swayed by numbers, such as 48, 96, 192 and whatever. It should be self-expanatory why.

I've said elsewhere that my favorite front end even at 48k is at this moment either a Meitner or dCS DSD front-end, resampled with a high-precision resampler.

George
FWIW, when I did the listening comparisons I referred to earlier in this thread, they were completely blind. And I was completely shocked by the outcome.

So there was no "shiny front panel" influence. But I won't contest your opinion about the differences revolving more around the filters and clocking.

So you actually did the 48K vs. 96K comparison yourself, the way you suggested, and heard no differences? Or subtle, occasionally audible differences only?

BTW, I did find the samples on another drive. I made analog mixes to 24/44.1 on the Masterlink to see if the differences we were hearing live actually translated to 44.1. So I still have those files. But, as GM pointed out, the differences between them are not only SR. Which renders them interesting but useless for this discussion.

Although, I still think that recording at 48K and then turning around and recording the same thing at 96K and comparing the two offers valuable information about the comparative sound. It's just that you can't point the finger at SR only. I guess the only thing that really makes a difference is what it sounds like coming out as opposed to what it sounded like going in. And that includes everything in the signal path.


Lynn Fuston
3D Audio Inc
Home of 3dB
Joined: Feb 2000
Posts: 2,184
gm Offline
MP Hall of Fame Member
Offline
MP Hall of Fame Member
Joined: Feb 2000
Posts: 2,184
Here is the path to the first entry in the "Pairs Test". It's two files - a recording of Chris Thiele's "Butt Trumpet" that Gary Paczosa made - both 96/24. The only difference between them is that one of them has a very high quality FIR filter at 20kHz.

You tell me which one is which.

ftp://ftp.ednet.net/

login: spapuser
password: spapus3r
account: (blank)

Download...

021202.0.iso (the Masterlink image)

...and...

Instructions for Masterlink.pdf

...for how to do it.

George


George Massenburg

http://www.massenburg.com
Joined: Jan 2000
Posts: 1,032
Platinum Member
Offline
Platinum Member
Joined: Jan 2000
Posts: 1,032
Quote:
Originally posted by gm:
a recording of Chris Thiele's "Butt Trumpet" that Gary Paczosa made
Sounds, uh..., interesting.


Lynn Fuston
3D Audio Inc
Home of 3dB
Joined: Aug 2001
Posts: 104
J
Senior Member
Offline
Senior Member
J
Joined: Aug 2001
Posts: 104
Quote:
Isn't that the truth. I find it increasingly difficult to get people to just listen objectively anymore. They would much rather argue about technical minutiae. Not just the engineers and producers, but EVERYBODY. It's no longer just whether you like the music, but if you approve of it's method of creation. Very frustrating.
* Good point. Lately Ive been advocating two paper cups connected by a string, but no one seems to keen on it as a recording method.

Seriously though, Ive done some recording recently at 96K on my home rig running Cubase, clocked by an Aardsync.

Overall, I much prefer the "sound" of 48 as opposed to 96.

yeah, I could hear some overall clarity and "tightness" at 96, but its too squeeky clean...too Steely Dan.. for my tastes. 48 just sounds more "real" to my ears (imagine that).

I do agree with GM's point about the inability to discern it on an individual Fender Bass track. However, throw up an entire mix at 96, and the differences are apparent.... but just not desirable IMHO.

Joined: Jul 2001
Posts: 709
A
Gold Member
Offline
Gold Member
A
Joined: Jul 2001
Posts: 709
Quote:
Overall, I much prefer the "sound" of 48 as opposed to 96.

yeah, I could hear some overall clarity and "tightness" at 96, but its too squeeky clean...too Steely Dan.. for my tastes. 48 just sounds more "real" to my ears (imagine that).
I call that pseudo presence what some people prefer on the lower sample rates. With a high quality converter you don´t have this effect too.

Sometines I also like this effect on choirs, but only for a short time. Mostly after a while it is fatiguing to hear this sounds. Even if it seems to be fresher in the first moment.

Joined: Oct 2000
Posts: 649
K
Gold Member
Offline
Gold Member
K
Joined: Oct 2000
Posts: 649
Quote:
Originally posted by 3D Audio:
Quote:
Originally posted by gm:
a recording of Chris Thiele's "Butt Trumpet" that Gary Paczosa made
Sounds, uh..., interesting.
Doesn't sound like any trumpet I've ever heard. My guess is that BT_1L4 is the non-filtered one. However, it's (as noted before) very close.

The "scientific" method employed here will probably not yield significant results (statistically speaking), since the chance of "getting it right" will be 50%. Kind of hard to find a solid standard deviation on that...simple correlation perhaps, but since it's that close, I doubt the correlation will be strong.

Also the Masterlink doc is woefully behind for Toast (2 revs, in fact). With modern (OSX Toast Titanium), you don't have to make a CD...you can mount the image on the desktop.


KK Proffitt
Chief Audio Engineer, JamSync, Nashville
JamSync
Joined: Oct 2000
Posts: 649
K
Gold Member
Offline
Gold Member
K
Joined: Oct 2000
Posts: 649
Test...sorry, my sig disappeared in the last message.

This is troubleshooting only.


KK Proffitt
Chief Audio Engineer, JamSync, Nashville
JamSync
Joined: Oct 2000
Posts: 649
K
Gold Member
Offline
Gold Member
K
Joined: Oct 2000
Posts: 649
Quote:
Originally posted by kk@jamsync.com:
Test...sorry, my sig disappeared in the last message.

This is troubleshooting only.
Admin please delete.


KK Proffitt
Chief Audio Engineer, JamSync, Nashville
JamSync
Joined: Jun 2001
Posts: 169
E
Senior Member
Offline
Senior Member
E
Joined: Jun 2001
Posts: 169
Quote:
Originally posted by 3D Audio:
recording the same material to two different but identical recorders
Different... but yet, identical.

\:D \:D \:D

Sorry... it's been a long day, and that little oxymoron just pushed my 'funny button' for some reason.

Joined: Mar 2000
Posts: 7,950
MP Hall of Fame Member
Offline
MP Hall of Fame Member
Joined: Mar 2000
Posts: 7,950
i feel the need to comment on REAL WORLD applications of this and GM's proposed test. i think its more REAL to take a source split to two different recorders as lynn described. it just makes more sense to me. recording something that 96khz and then putting some superduper filter on it just isnt realistic to any comparison. we must judge the process in real world applications such as lynn did. clocking, filtering and all.

what GM proposes, i would have to literally record all at 96khz and then add some superduper filter to the source to properly work at 48khz? im just unsure how that is a proper comarison... how does the system react at 48khz legit? [not downsampled]


alphajerk
FATcompilation
"if god is truly just, i tremble for the fate of my country" -thomas jefferson
Joined: Jan 2000
Posts: 1,032
Platinum Member
Offline
Platinum Member
Joined: Jan 2000
Posts: 1,032
Quote:
Originally posted by gm:
Here is the path to the first entry in the "Pairs Test". It's two files - a recording of Chris Thiele's "Butt Trumpet" that Gary Paczosa made - both 96/24. The only difference between them is that one of them has a very high quality FIR filter at 20kHz.

You tell me which one is which.

ftp://ftp.ednet.net/
Does it matter which one is which if they can are audibly distinguishable from each other?

BTW, I don't know whose server that is, but I was DL'ing the file last night and it did not complete. Tried again and got a "connection error." Maybe the server overheated from everyone trying to DL at the same time. I know this is a hot topic.

Can you check on that? It's not working now as I type.

(Forget it. It's working now-7 AM. But it's dreadfully slow, ~16K/sec. About a five hour download. But it's working now.)


Lynn Fuston
3D Audio Inc
Home of 3dB
Joined: Nov 2001
Posts: 168
F
Senior Member
Offline
Senior Member
F
Joined: Nov 2001
Posts: 168
I haven’t read the whole thread, so excuse me if this has been covered yet.
As far as I know, most people are hearing differences between 48k and 96k/192k because of their “Lower Quality” converters.
When recording @ 48k, there is some DSP involved in the process.
Nyquest filtering and oversampling.
Good DSP is expensive. (And hardly used)
When recording @ 96k/192k there is no DSP involved (the signal passes full bandwidth).
That way you are bypassing the nasty artifacts of the DSP process.
With high quality converters there is no significant difference between 48k and 96k/192k.
I’m with GM on this one.


Fredo Gevaert
Temple Of Tune
fredo@temple.be
http://www.temple.be
Joined: Jan 2000
Posts: 1,032
Platinum Member
Offline
Platinum Member
Joined: Jan 2000
Posts: 1,032
All I'm concerned about are "in band" differences. The ones we can definitely hear. I know when I listened to 192, 96, 48 and 44.1 side by side on the same converter (dCS), I was shocked at what I heard, and it wasn't just UHF information.

I don't necessarily believe that having info above 20K is something that many people will notice. It's the difference in the 20-20K range that concerns me.


Lynn Fuston
3D Audio Inc
Home of 3dB
Joined: Sep 2000
Posts: 15,398
10k Club
Offline
10k Club
Joined: Sep 2000
Posts: 15,398
Quote:
Originally posted by alphajerk:

what GM proposes, i would have to literally record all at 96khz and then add some superduper filter to the source to properly work at 48khz? im just unsure how that is a proper comarison...
Yeah, certainly the filter itself would affect the sound, eh?

Joined: Feb 2000
Posts: 2,184
gm Offline
MP Hall of Fame Member
Offline
MP Hall of Fame Member
Joined: Feb 2000
Posts: 2,184
Quote:
Originally posted by alphajerk:
[...]how does the system react at 48khz legit? [not downsampled]
Um...if I understand you're questions correctly (what happens if you play back a 96k file at 48k) then it plays back at 1/2 speed. Duh.

But that couldn't be what you meant, could it?

George


George Massenburg

http://www.massenburg.com
Joined: Mar 2000
Posts: 7,950
MP Hall of Fame Member
Offline
MP Hall of Fame Member
Joined: Mar 2000
Posts: 7,950
no, not a 96khz... what did i mean? um, 48khz legit means the 48khz source file for comparison to the 96khz source file recorded AT 48khz, not filtered/downsampled from a 96khz file. we must look at how ADC's operate in real time.


alphajerk
FATcompilation
"if god is truly just, i tremble for the fate of my country" -thomas jefferson
Joined: Oct 2000
Posts: 649
K
Gold Member
Offline
Gold Member
K
Joined: Oct 2000
Posts: 649
Quote:
Originally posted by alphajerk:
no, not a 96khz... what did i mean? um, 48khz legit means the 48khz source file for comparison to the 96khz source file recorded AT 48khz, not filtered/downsampled from a 96khz file. we must look at how ADC's operate in real time.
That would be two converters linked and driven by a master clock that can accurately sync 48K/96K so that the frame edges are the same (er, every other one for 48K)? It's possible....


KK Proffitt
Chief Audio Engineer, JamSync, Nashville
JamSync
Joined: Feb 2000
Posts: 2,184
gm Offline
MP Hall of Fame Member
Offline
MP Hall of Fame Member
Joined: Feb 2000
Posts: 2,184
Quote:
Originally posted by alphajerk:
no, not a 96khz... what did i mean? um, 48khz legit means the 48khz source file for comparison to the 96khz source file recorded AT 48khz, not filtered/downsampled from a 96khz file. we must look at how ADC's operate in real time.
Well, you should look at how ADC's work...at how boxes are constructed...at the electronics topology.

George


George Massenburg

http://www.massenburg.com
Page 2 of 7 1 2 3 4 5 6 7

Moderated by  gm 

Link Copied to Clipboard
Powered by UBB.threads™ PHP Forum Software 7.7.5