avatar

gold

Platinum Blonde

Posts: 1,494 Member Since:27/01/2011

#41 [url]

The Prism AD2 will output two sample rates at once. I run a pitch and catch instance of Sequoia. I've done the test. On most things it's hard to tell the difference. I have two Prism DA2's so it was a pretty good test.

Quote    Reply   
avatar

zmix

Aqua Marine

Posts: 3,982 Member Since:20/01/2011

#42 [url]

gold wrote:
The Prism AD2 will output two sample rates at once. I run a pitch and catch instance of Sequoia. I've done the test. On most things it's hard to tell the difference. I have two Prism DA2's so it was a pretty good test.

Your comments are pretty typical of the majority of high end converters, there really isn't much difference at any sample rate...

Multiple sample rates is a nice feature, and so easy to implement, I wonder why more manufacturers don't do it? 

Same with master clocks which output several rates but are from the same master crystal, so the playback and record setups (to use the hetero terminology, not that there's anything wrong with that), don't drift...


Quote    Reply   
avatar

maarvold

Aqua Marine

Posts: 3,011 Member Since:23/01/2011

#43 [url]

zmix wrote:
I am the sort of person who has infinite curiosity and patience, and I always like to know "why" something seems "different", so I endeavor to learn the cause.


In my test, it turned out that there was RF and other ultrasonic interference within the individual tracks which were causing artifacts in the *audible* spectrum, which went away when I downsampled to 44.1, and these were simply files within a mix, and no difference in "converters" except for my monitoring DAC, which upsamples everything to the same rate.


I wouldn't just assume that listening to 2 channels at 44.1 or 96kHz is going to make the same difference as listening to 50-60 channels at 44.1 or 96kHz.

So I'm doing Broadway-type cast soundtrack album right now.  There is one male vocal that has been driving me nuts: I just can't get the top end to 'sit' in a way that I like and I've spent several hours trying every trick in the book.  And the project is 96k--not my choice; it was started that way and I just continued on with it.  So I highlighted a couple of lines and sent it into iZotope RX 5.  Although I don't see any oscillation, there is a ton of info above 22k and it looks just as loud as what's below.  I selected it and ran Spectral Repair at the strongest setting and, although it still leaves plenty of information, it quite visably removes a good amount of something.  When I put it back into Pro Tools and processed the original file (overwriting it with the new 'cleaned' file) for the first time my gut kind of relaxed when I listened to the playback.  Of course I have to acknowledge that there was potentially a lot of expectation bias involved, but when I hit command-z I just wasn't as happy with it.  Two other female vocals--both cut in a different iso booth and on different mics, but otherwise the exact same chain--are totally fine for me.  Oh sweet mysteries of life...

The top is the 'Before' and the bottom is 'After'.  The more-or-less black area toward the right is also quite mysterious.  And it a significantly more dramatic visual difference on a big monitor.  
 image

Last Edited By: maarvold . Edited 1 time.

Quote    Reply   
avatar

morespaceecho

Platinum Blonde

Posts: 2,243 Member Since:29/01/2011

#45 [url]

one of my defenses of poor old 44.1 is that i don't think a built in lowpass at 22.05k is a bad thing!

i tried chuck's test last night....took an 88.2 session, converted everything to 44.1, made mixes of both and compared. i'll have to listen again today with fresh ears but it seemed like a really tiny difference. if pressed i'd say i preferred the 88.2, but there was really nowt wrong with the 44.1, i dunno that i could pick which was which in a blind test.

the null was interesting....i had plugs on about half the tracks....all the tracks without plugs cancelled more or less completely (as you'd hope with a decent SRC), what was left sounded like a really scooped out version of my tune. so the plugs are certainly doing "something different" at 88.2 than 44.1, but how is it that the null shows a really clear difference, but a/b'ing the mixes i could barely tell them apart?

www.oldcolonymastering.com

morespaceecho.bandcamp.com

Quote    Reply   
avatar

maarvold

Aqua Marine

Posts: 3,011 Member Since:23/01/2011

#46 [url]

I think I'm going to use iZotope RX to SRC the 'offending' vocal track down to 48, then back up to 96 to see how I like that.  That might be a more effective use of my "expensive low-pass filter".  If I feel like something good (and real) happened, I'll report back.  

Quote    Reply   
avatar

zmix

Aqua Marine

Posts: 3,982 Member Since:20/01/2011

#47 [url]

morespaceecho wrote:
one of my defenses of poor old 44.1 is that i don't think a built in lowpass at 22.05k is a bad thing!

i tried chuck's test last night....took an 88.2 session, converted everything to 44.1, made mixes of both and compared. i'll have to listen again today with fresh ears but it seemed like a really tiny difference. if pressed i'd say i preferred the 88.2, but there was really nowt wrong with the 44.1, i dunno that i could pick which was which in a blind test.

the null was interesting....i had plugs on about half the tracks....all the tracks without plugs cancelled more or less completely (as you'd hope with a decent SRC), what was left sounded like a really scooped out version of my tune. so the plugs are certainly doing "something different" at 88.2 than 44.1, but how is it that the null shows a really clear difference, but a/b'ing the mixes i could barely tell them apart?

Question: How did you null a 44.1 file against an 88.2 file?

What plugins did you use?  They can and do generate aliasing, and possibly modulation that isn't repeated exactly across two passes, so it's hard to know from a null test which file has greater artifacts without making more specific tests to find the culprit.


Quote    Reply   
avatar

zakco

Platinum Blonde

Posts: 1,066 Member Since:02/03/2011

#48 [url]

maarvold wrote:
The more-or-less black area toward the right is also quite mysterious.  And it a significantly more dramatic visual difference on a big monitor. 
Wow, that's an odd one! Just curious as to what freq the "hole" starts and ends at? I've seen notches like that before (usually due to unusual offline processing or punching/editing) but never with a "ceiling" where content suddenly appears again...

Quote    Reply   
avatar

morespaceecho

Platinum Blonde

Posts: 2,243 Member Since:29/01/2011

#49 [url]

zmix wrote:
How did you null a 44.1 file against an 88.2 file?


i upsampled the 44.1 mix to 88.2 to make for easier a/b'ing. i also had a separate session where i downsampled the 88.2 mix to 44.1, so i could compare those. i wasn't sure which was a fairer fight! 

plugs were mostly equilibrium, blockfish (aliases for sure), black box analog hg-2 (doesn't really alias), and u-he presswerk (not sure). 

www.oldcolonymastering.com

morespaceecho.bandcamp.com

Quote    Reply   
avatar

bob olhsson

Aqua Marine

Posts: 3,315 Member Since:25/01/2011

#50 [url]

I'm not sure any sample rate or DSD is "best." We are really listening to different various filters that can sound better or worse at certain sample rates. I like 48 and 96 mostly because a number of plug-ins sound better at those rates. My guess is it's because the creative decisions about the code were made at those rates. Yes, there are creative decisions because there are a dozen ways to skin a cat and developers will choose the best sounding ones.

www.audiomastery.com Bob's room 615 562-4346 georgetownmasters.com Georgetown Masters 615 254-3233 www.thewombforums.com

Quote    Reply   
avatar

maarvold

Aqua Marine

Posts: 3,011 Member Since:23/01/2011

#51 [url]

zakco wrote:

maarvold wrote:The more-or-less black area toward the right is also quite mysterious.  And it a significantly more dramatic visual difference on a big monitor. 
 

Wow, that's an odd one! Just curious as to what freq the "hole" starts and ends at? I've seen notches like that before (usually due to unusual offline processing or punching/editing) but never with a "ceiling" where content suddenly appears again...

 
The bottom of the hole was roughly 22k and the top was roughly 44k (an octave, if there's any significance to that).  

Quote    Reply   
avatar

maarvold

Aqua Marine

Posts: 3,011 Member Since:23/01/2011

#52 [url]

maarvold wrote:
I think I'm going to use iZotope RX to SRC the 'offending' vocal track down to 48, then back up to 96 to see how I like that.  That might be a more effective use of my "expensive low-pass filter".  If I feel like something good (and real) happened, I'll report back.  

 
So maybe it's a complete wank, but something feels more natural about the top end to me with the vocal that was SRC'ed to 48 and then back to 96.  I'll grant you, it's subtle, but I'm happy.  It may also have something to do with how the 5 plugs I have on the track behave on the full bandwidth 96k file with that additional info that's above our range of hearing.  Full bandwidth feels subtly like I'm hearing a hologram singing and the file with the top SRC'ed out of it feels like a solid human being.  

One interesting thing I discovered is that there is a checkbox in iZotope that says "Change tag only" which, resumably would 'rebrand' audio tracked at the wrong clock speed at it's correct speed.  

Quote    Reply   
avatar

John Eppstein

Platinum Blonde

Posts: 1,175 Member Since:31/05/2015

#54 [url]

owlander wrote:

John Eppstein wrote:

owlander wrote:
I was fooling with my recording setup the other night and after buying some DVD+RW's decided to try DSD. However, when listening to playback I was underwhelmed vs 192khz. There's more crispness and likeness to the original signal vs the DSD version. Anyone else with DSD? What have you found?

Maybe I need to do some more listening, but for what I can tell at the moment, I much prefer 192 PCM.

The original signal?

What are you talking about?

To benefit from DSD that has to be your original format - you won't get any benefit from starting at 192 and converting to DSD - once something is lost, it's lost. And the extra layer of conversion will liklely have some adverse effect as well.

Most modern converters do actually start with DSD and downconvert to whatever PCM format you choose to work at, but it really only works one way. The problem is that you can't perform the usual processing and editing operations in the native DSD format - to do that stuff you HAVE to downconvert to PCM. You can mix to DSD because the mix is a new recording (it's kind of like taking a hi-rez photo of a collage composed of bits of lower rez original pictures), but it still isn't going to be as hi-rez as if iut was an original DSD track. That's why DSD is primarily regarded as an archival format. You record your original tracks in DSD, then you can downconvert to whatever mix/distribution PCM format you choose - but it doesn't work the other way.

If you had original DSD tracks and wanted to preserve resolution during mixdown you could play back your original DSD tracks in analog to do your mixing and processing, then capture your mix in DSD.

John...original signal to 192. Original signal to DSD. No converting.

Well, that's not what your original post said:
There's more crispness and likeness to the original signal vs the DSD version

So what was the "original signal, exactly?

Quote    Reply   
avatar

John Eppstein

Platinum Blonde

Posts: 1,175 Member Since:31/05/2015

#55 [url]

gtoledo3 wrote:
"Most modern converters do actually start with DSD and downconvert to whatever PCM format you choose to work at, but it really only works one way."

Really? I'd like to see proof of that. Not saying it isn't so, but darn... I will definitely admit that I do not *think* it is so! They oversample, sure, but I am pretty damn sure they do not process it to DSD then change it to PCM.

It's the basis for modern delta-sigma conversion, which starts with a high bitrate 1-bit (dsd) encoding function.

https://en.wikipedia.org/wiki/Delta-sigma_modulation

Quote    Reply   
avatar

John Eppstein

Platinum Blonde

Posts: 1,175 Member Since:31/05/2015

#56 [url]

owlander wrote:
Ok...wait a second. Something was screwed up with my machine, as now the files are showing up as de facto .dff and I am unable to play them. Before, (first time) when I recorded using the DSD format I was able to play the file, which doesn't make sense as none of my software actually supports .dff? Not sure what it was doing? It said DSD? ...anyway, now it seems to be recording DSD properly and I can see the file is a fair bit bigger vs 192...105MB vs 128MB DSD. If I could actually play the file, that would be good.

What are people using for editing software to manipulate these files? Any ideas? I use Sequoia but it don't support.

As I understand it, one of the major drawbacks of DSD is the inability to edit or mix the filkes without cdonverting to PCM.

Quote    Reply   
avatar

silvertone

Aqua Marine

Posts: 2,665 Member Since:26/01/2011

#57 [url]

No matter what your ears should be the final judge.

We all hear differently. Is Terry wrong in what he hears? Am I? Is Chuck? Or John?

I don't think so but I only trust my ears. They are what tell my brain, this is good, this is bad.

Personally, I don't care what can be technically proven or as shown superior. As we all know digital is perfect sound forever... and maybe it is to some. The majority of people now know it's not. Specifications do not always tell the whole truth.

I love old tube pres but a THAT chip is technically superior... yet I know which one I'd rather record with... YMWV

Silvertone Mastering, celebrating 28 years in business.

www.silvertonemastering.com

Quote    Reply   
avatar

owlander

Platinum Blonde

Posts: 1,022 Member Since:27/01/2011

#58 [url]

John Eppstein wrote:

owlander wrote:

John Eppstein wrote:

owlander wrote:
I was fooling with my recording setup the other night and after buying some DVD+RW's decided to try DSD. However, when listening to playback I was underwhelmed vs 192khz. There's more crispness and likeness to the original signal vs the DSD version. Anyone else with DSD? What have you found?

Maybe I need to do some more listening, but for what I can tell at the moment, I much prefer 192 PCM.

The original signal?

What are you talking about?

To benefit from DSD that has to be your original format - you won't get any benefit from starting at 192 and converting to DSD - once something is lost, it's lost. And the extra layer of conversion will liklely have some adverse effect as well.

Most modern converters do actually start with DSD and downconvert to whatever PCM format you choose to work at, but it really only works one way. The problem is that you can't perform the usual processing and editing operations in the native DSD format - to do that stuff you HAVE to downconvert to PCM. You can mix to DSD because the mix is a new recording (it's kind of like taking a hi-rez photo of a collage composed of bits of lower rez original pictures), but it still isn't going to be as hi-rez as if iut was an original DSD track. That's why DSD is primarily regarded as an archival format. You record your original tracks in DSD, then you can downconvert to whatever mix/distribution PCM format you choose - but it doesn't work the other way.

If you had original DSD tracks and wanted to preserve resolution during mixdown you could play back your original DSD tracks in analog to do your mixing and processing, then capture your mix in DSD.

John...original signal to 192. Original signal to DSD. No converting.

Well, that's not what your original post said:
There's more crispness and likeness to the original signal vs the DSD version

So what was the "original signal, exactly?


Me, singing and playing into it.

Quote    Reply   
avatar

zmix

Aqua Marine

Posts: 3,982 Member Since:20/01/2011

#59 [url]

silvertone wrote:
No matter what your ears should be the final judge.

We all hear differently. Is Terry wrong in what he hears? Am I? Is Chuck? Or John?

I don't think so but I only trust my ears. They are what tell my brain, this is good, this is bad.

Personally, I don't care what can be technically proven or as shown superior. As we all know digital is perfect sound forever... and maybe it is to some. The majority of people now know it's not. Specifications do not always tell the whole truth.

I love old tube pres but a THAT chip is technically superior... yet I know which one I'd rather record with... YMWV

First of all, no.  The majority of people now have been cynically been persuaded that "digital audio is bad, m'kay?"  by trumped up demonstrations of transcoded  data compressed audio, but in reality most people prefer digital these days.   Perhaps you're just being hyperbolic, in which case, fine... 

It's absolutely true, however that you should never take anyone's opinions as a fact unless you've actually done the work of verifying their claims by reproducing their tests...

In the meantime:

xfiles-trust-no-one.jpg



Last Edited By: zmix . Edited 1 time.

Quote    Reply   
avatar

owlander

Platinum Blonde

Posts: 1,022 Member Since:27/01/2011

#60 [url]

Lack of trust shows low self esteem. Just kidding.

Also, what Larry is referring to in his post, Chuck, are alternative facts...

image

Last Edited By: owlander . Edited 1 time.

Quote    Reply   
Add Reply

Quick Reply

bbcode help