Sample rates question

[Peter Simonsen]

[quote author="TomWaterman"] However for those that think PT runs at 24-bit, I recall its actually 48-bit fixed point system
Cheers Tom[/quote]

Tom, Rochey

Thanks for your answers...now where is JPrisus ??? *S*

I dont know for sure, but I belive that Tom is correct..PT is 48bit fixed point math today !!!

Thanks for sharing..an interesting topic where I sadly fall on my a.. to often, but I try my best to gain new knogledge about the digital side of recording..Its just soo very difficult to find information that is not clouded by old "myths"..imho ..Poul Frindle, Nika, Dan levry is the only ones who really are on top of digital these day (and GM too, but he seldom speaks to much about it )...just a little hard for someone with little understanding to follow these gents :-D. I belive Nika has a book comming out later this year..written so even a recording engineer has a chance to understand digital..*GGG*

Kind regards

Peter

 

[JPrisus]

[quote author="Peter Simonsen"]

So I would like to know if JPrisus could tell me as to what "new" math problems he refers to when he so utterly speaks of bad PT math, and which problems the bad math are the results in, and where in the signal chain it goes so wrong *S*

[/quote]

Sorry, forgot about this thread, got too long about 3 pages ago. When I process something within PTHD, it affects the original source negatively to ym ears, and is something I try to refrain from as much as possible. I'm talking about things as slight as crossfades over edit points. When I do the same in DP4 or Nuendo, it doesn't seem to affect the source as much. I don't mean to come off negatively or anything, but PT has had proven issues with their processing, and it hasn't been fixed in PTHD. I'm not talking about converters either. To my knowledge, they never even addressed the mix bus concerns, and still refute any complaints about it. Internal plug-in processing might be 48-bit now, i think that's correct IIRC. But i'm no techie for this stuff, i just use what I feel sounds best!

I have tremendous respect for the knowledge you gentlemen have regarding digital processing and the like, so please forgive me as I know I can come off strong at times. But the bottom line is, what does it sound like? If I can hear a graininess in crossfades or an overall lack of headroom compared to competetive products, that tells me there's a problem somewhere. Regardless of what looks good on paper, I trust my ears, and so do my clients. So when someone asks publicly about upsampling in PT while bouncing to disk, I have no problems sharing my opinion that there's no sonic benefit in upsampling at that point, PT SRC has an audible detriment to the source which I have heard (maybe it's no different than other systems, either way I wouldn't do it), and bouncing to disk in PT has an audible detriment to the stereo mix. I've experienced this on both Mix+ and HD.

 

[alk509]

[quote author="JPrisus"]and bouncing to disk in PT has an audible detriment to the stereo mix.[/quote]

:roll:

This just isn't true. There was a rather lengthy discussion about this in the Logic Users Group about a month ago ("Logic bounces sound better than PT bounces") and the argument was settled when people started bouncing things in Logic and PT (in the same system), inverting one of them and adding them again, getting complete silence as a result.

The sound of a bounce does NOT depend on the mixing algorithm because there is only one way to add any two numbers, with only one possible result! 2 + 2 = 4 whether you use PT, Logic, DP, Nuendo or whatever. It is the hardware you use the software with, that determines the accuracy of the samples in a mixed file (specifically the precision of the FPU doing the adding).

When people say that ProTools has 48-bit precision, they're in reality talking about the precision of the HD hardware's FPU. If you're using Logic with your HD hardware, the precision is still determined by the FPU in the HD hardware, and stuff will sound exactly the same as it did in PT! If you use Logic with your computer's internal 32-bit FP processor, then it will sound different than PT with HD hardware, but the same as PT LE, which uses your computer's FPU.

Peace,
Al.

 

[DrFrankencopter]

Just thought I'd chime in here with some of my thoughts. I use the Emu/Ensoniq Paris DAW, which is an integer based system (I think). One of the great things about this particular DAW mixer is that it is possible to clip individual channels in a mix just by cranking up trim gains, or running the faders up high. This obviously results in a different sound than non clipped, properly gain staged channels. In a mix situation you can use this as a creative effect....you play the console. It's much like with an analog board, you can play off the board's headroom.

Systems that use floating point math for channel calculations can only clip at the summing buss where the audio gets folded back into a 24 bit int. This means you're clipping everything, not individual channels...so it's not really desireable to go there.

Many folks argue that digital mixers should be pristine/accurate math machines, but I like to think of a mixer as forming part of a creative feedback loop that is closed by the operators ears and hands. A mixer doesn't need to be linear or precise, in fact if it isn't then users will find ways to exploit those nonlinearities to creative effect.

Cheers,

Kris

 

[Peter Simonsen]

JPrisus,


Thanks for climing in on this again ;-)

Regardless of what looks good on paper, I trust my ears, and so do my clients

I do too..one of the most important things to do when working in the audio field I belive.*S* What I very often find interesting is how "we"..very often find it hard to agree on what we hear..I for one has not heard problems with the things you talk about "trouble with x-fading etc" and if I understand you correctly it has to be VERY bad if one can hear it so clearly among two, or more systems...Some people even say they hear the fader moves in PT ruins good vocal takes..!!!

JPrisus could you explain to me how you conduct these listening tests among the different systems..or is it soo obvius as you say that one just needs to listen, and one should be able to hear this without even knowing that an edit was done to a audio file..???

Thanks

Kind regards

Peter

 

[TomWaterman]

Hey Kris,

Do you find the digital clipping in Paris sounds musical? I always thought of digtal clipping as a big no no once inside the box...what kind of effect do you get from it - is it pleasing?

At the end of the day, whatever sounds best to your own ear is the only way.....there are lots of opinions on the net about things and its easy to get caught up in it all.

BTW Have any of you checked out the AWEsome DAWsum CD from Lynn Fuston at 3D Audio...I've been meaning to get a copy at some point?

Cheers Tom

 

[JPrisus]

[quote author="alk509"][quote author="JPrisus"]and bouncing to disk in PT has an audible detriment to the stereo mix.[/quote]

:roll:

This just isn't true. [/quote]

I don't know, If i can hear the difference, then it's true to me. Again, something that might seem fine on paper, but in reality, i hear a major difference when I bounce something to disk compared to printing it digitally to an external 2trk (no extra conversion or processing involved). I'm not familiar with Logic so I can't comment. I'm not saying PT is worse than its competitors in that dept, and i'm sorry if I implied such. I think DP4 sucks at this as well. Never tried it in Nuendo/Cubase. But the original post was regarding protools specifically, so I made my statement concerning PT.

 

[JPrisus]

[quote author="Peter Simonsen"]JPrisus,


Thanks for climing in on this again ;-)

Regardless of what looks good on paper, I trust my ears, and so do my clients

I do too..one of the most important things to do when working in the audio field I belive.*S* What I very often find interesting is how "we"..very often find it hard to agree on what we hear..I for one has not heard problems with the things you talk about "trouble with x-fading etc" and if I understand you correctly it has to be VERY bad if one can hear it so clearly among two, or more systems...Some people even say they hear the fader moves in PT ruins good vocal takes..!!!

JPrisus could you explain to me how you conduct these listening tests among the different systems..or is it soo obvius as you say that one just needs to listen, and one should be able to hear this without even knowing that an edit was done to a audio file..???

Thanks

Kind regards

Peter[/quote]

My 'listening tests' have been conducted during tracking/mixing sessions using the formats discussed. I'm using them in real-life scenarios and posting what i've noticed, thats all. On the older Mix+ systems, yes I could hear the audible effects of a fader move on a lead vocal. I don't hear that on HD, so maybe it has to do with the 48-bit processing. The examples i've given are very obvious to me, and i'm not trying to pick a fight or anything but this is why some prefer Neve to SSL etc. In my taste and opinion, i choose not to use PT for anything more than a multitrack and sometimes editor, because in all other facets i've heard things I don't like very much from it.

I can't tell if you're being sarcastic or serious, and i'm sorry if my experiences differ from the 'facts', but that is how I feel and I know many others who would agree with me. I doubt you'd be able to pick out an edit point within a stereo mix (unless it was horribly executed), but i think if you crossfaded an edit and really focused on hearing *just that edit* for a minute or so, you'd develop a sense of awareness for the crossfade, just like hearing a fader ride on a vocal. It's splitting hairs really, i don't think a crossfade will make or break an album lol, but i do take pride in the little things (as most AEs would). And i have no problems stating that I feel other systems execute these processes in a more musical manner than PT.

If you really wanna talk about good processing, look at Paris... they had it down before anyone else. Damn shame really. They've become the first 'vintage' DAW!

 

[DrFrankencopter]

Do you find the digital clipping in Paris sounds musical? I always thought of digtal clipping as a big no no once inside the box...what kind of effect do you get from it - is it pleasing?

I've done some preliminary investigations of clipping in Paris, and it's really just saturation (i.e. clipping without 'wrapping'). I think what makes it cool is that you can do this on individual tracks as opposed to clipping an entire submix (or the whole mix). A little bit of saturation on a kick or snare track is hardly noticeable. A little bit of saturation on a bass track actually sounds cool. Add all this together and you end up finding that mixes done in paris are louder/hotter, and have more 'attitude' than those done in systems with 'infinite headroom'. I don't clip vocals, or overheads or acoustic guitars (except maybe for creative effect), but it's cool to be able to vary the sound based on how you gain stage a channel.

Cheers,

Kris

 

[TomWaterman]

How would you describe the saturation? I would have thought all you would be left with would be digital fizz?? Afterall its a not a tube or transformer we're talking about....

Have you tried doing it to a simple sine wave, pulse or whatever and looked at an FFT analysis to see if it creates any kind of extra harmonics - odd or even?? I mean thats what I would expect from a tube or something but in digital clipping??

I know that clipping converters can have a similar effect but I thought that relied heavily on the analog stage prior to the AD??

I'm not trying to say it can't happen but I'm interested in your findings....

Cheers Tom

 

[Peter Simonsen]

[quote author="JPrisus"]
I can't tell if you're being sarcastic or serious, and i'm sorry if my experiences differ from the 'facts',![/quote]

I´m sorry if I sounded that way..not my intension *s* It´s just that I really want to try to understand what makes people hear something I dont..is it me, or "them". If it is me...I like to learn and listen harder.

You basicly say something like if I took a 2-3 min long audio file that you knew very well (something you may even have played or recorded yourself to your own standarts)... and I did, (or did not do) some cuts, and x-fading in the file (you would not know if, and where) you could in a double blind test tell me where the x-fades where ..say 8-9 times out of 10...because it ruined the sound soo much as you say. Now this really makes my mind tick very hard as it tells me I cant hear **** *GGG* and I mean this in the most friendly way..*S* I really find this amazing !!! Your ears must be among the "golden ones"

Thanks for your reply

Kind regards

Peter

 

[JPrisus]

[quote author="TomWaterman"]

I know that clipping converters can have a similar effect but I thought that relied heavily on the analog stage prior to the AD??
[/quote]

it has everything to do with the analog circuit, because that is what's being clipped/saturated. Same premise as the Lavry Blue ADCs. I'm interested in this 'internal' clipping too, never heard of it before!

FWIW, i found that raising the input level of my Lucid AD feeding a Masterlink changes the sonic landscape of a mix noticeably. Upper mids are more forward, bottom seems a touch thinner. And this is only a 4-5dB increase, not clipping (meter-wise) at all. Goes to show there's more to digital than just keeping an eye on the meters.

 

[JPrisus]

[quote author="Peter Simonsen"]

You basicly say something like if I took a 2-3 min long audio file that you knew very well (something you may even have played or recorded yourself to your own standarts)... and did, (or did not do) some cuts, and x-fading in the file you could in a double blind test tell me where the x-fades where ..say 8-9 times out of 10...because it ruined the sound soo much as you say. Now this really makes my mind tick very hard as it tells me I cant hear **** *GGG* and I mean this in the most friendly way..*S* I really find this amazing !!! Your ears must be among the "golden ones"

[/quote]

No i don't think that would be possible... i'm far from one of the golden ones, and i doubt they could do this either if the edits were performed well enough. I'm saying that in the microscopic sense, the inner audio geek inside of me, if I take say a bass track, edit it in whatever manner was necessary, crossfade it, and really focus on *one* crossfade at a critical point, like say a change of key or song section, that yes, i think DP4 and Nuendo/Cubase perform these fades in a much smoother and less abrasive fashion than Protools, and i think it's faily noticeable. I don't know if you will notice it, as I don't know you at all. I may not have noticed it 2 years ago. Will it make a difference in the grand scheme of things? nope. But when it's up to me, i'll choose to use what i feel sounds best. The only time I choose to edit in PT is when a drummer needs beat detective, and that's very occassionaly, maybe once a year. And next time I do that, i think i will import the file into another program before I do my crossfades and merge the edits.

 

[chrissugar]

"FWIW, i found that raising the input level of my Lucid AD feeding a Masterlink changes the sonic landscape of a mix noticeably. Upper mids are more forward, bottom seems a touch thinner. And this is only a 4-5dB increase, not clipping (meter-wise) at all. Goes to show there's more to digital than just keeping an eye on the meters."

there can be two reasons for this:
1-the analog stage when it is pushed harder change the sound because it can't handle that level without problems.
2-even if the analog part is perfect the problem can be the peak metering. Unfortunately more than 90 percent of the digital recording devices do not show the peak level of the reconstructed signal, but the momentary sample value and that is wrong. As Paul Fridle, Dan Lavry, Nika and some other AD/DA conversion gurus said, it is important to measure the peak of the reconstructed wave because you can have overload even if the peak meter is at -1dB. THAT'S WHY IT IS GOOD TO RECORD WITHOUT GOING OVER -6dB PEAK if you record at 24 bit. At 16 bit every lost bit is a problem so it is about a compromise.[the diference in dinamic between 24bit and 16 bit is 48dB, so loosing 6dB at 24bit is not a problem if you want to be safe.

chrissugar

 

[TomWaterman]

Nuendo does fades in realtime where as PT processes offline and replaces the selection with a new region......you are effectively bouncing to disk in PT.

If you notice a difference in sound quality when you BTD then you will hear the same difference when you x-fade in PT.

Thats why Nuendo has the upper edge when it comes to x-fades in my opinion.....

Cheers Tom

 

[JPrisus]

[quote author="chrissugar"]

there can be two reasons for this:
1-the analog stage when it is pushed harder change the sound because it can't handle that level without problems.
[/quote]

interesting stuff about the metering... it doesn't sound like digital clipping though, so i think i'm fine in that dept. Regarding the analog stage, what do you mean by perfect? I don't know if i'd really wanna hear something *perfect*, as it's the non-linearities that make it interesting to me.

just a thought... the whole 'louder is better' regarding digital recording, well i know this isn't necessarily true any longer. 24-bit is reached at what, -22dB or something like that? But anyway, I think the PT mix bus headroom issue could possibly be a direct result of people stuffing each track to it's near-clipping point and mixing with all faders relative to 0VU, not taking into account the mix bus struggling with these ridiculous levels. any thoughts?

FWIW, A friend just pointed out to me that a mackie 8-bus has better specs than a 9000J. Go figure.

 

[JPrisus]

[quote author="TomWaterman"]Nuendo does fades in realtime where as PT processes offline and replaces the selection with a new region......you are effectively bouncing to disk in PT.

If you notice a difference in sound quality when you BTD then you will hear the same difference when you x-fade in PT.

Thats why Nuendo has the upper edge when it comes to x-fades in my opinion.....

Cheers Tom[/quote]

Thanks Tom, that's gotta be what i'm hearing then. Good to know!

 

[Peter Simonsen]

[quote author="JPrisus"][
FWIW, i found that raising the input level of my Lucid AD feeding a Masterlink changes the sonic landscape of a mix noticeably. Upper mids are more forward, bottom seems a touch thinner. And this is only a 4-5dB increase, not clipping (meter-wise) at all. Goes to show there's more to digital than just keeping an eye on the meters.[/quote]

Im sorry..Its me again..* :grin: *..What do you compare this to...whats your ref..???. 4-5 db is alot gain change imho but one has to have a reference to meassure to or listen to or else..the difference could be something going on in your head..

To be able to compare you have to have 2 identical files mathed within .1db with each other..Just raising the input will 100% make difference in hearing..hell pulling 4-5 db in vol on a cheap mono radio will sound different..

Again..I´m interested seriously..I try very hard to understand where you are comming from..

Kind regards

Peter

 

[Peter Simonsen]

[quote author="JPrisus"]
No i don't think that would be possible... i'm far from one of the golden ones, and i doubt they could do this either if the edits were performed well enough. I'm saying that in the microscopic sense, the inner audio geek inside of me, if I take say a bass track, edit it in whatever manner was necessary, crossfade it, and really focus on *one* crossfade at a critical point, like say a change of key or song section, that yes, i think DP4 and Nuendo/Cubase perform these fades in a much smoother and less abrasive fashion than Protools, and i think it's faily noticeable. .[/quote]

I´m sorry..I just dont get it...how can you know then..??? I mean If I focus and know where a edit is..yes the brain cells will tell me I can hear it if I want them to, and sometimes they do it anyway..but this imho does not matter if one can not when not knowing..!!! How can you be sure at all..I just dont get it... it does not add up.. Do a double blind test with audio files -xfaded in PT, nuendo etc and tell me you can hear which one did what...in a blind test...again 8-9 times out of 10. I think otherwice its a little on the strong side to go out in a public forum and claim PT has very big problems with the math..so much it ***** up audio..!!!

Kind regards

Peter

 

[JPrisus]

Because i'm not talking about hearing an edit in context or not. That's not the point at all. If a fade is performed well by the editor, it should be impossible to hear regardless of the platform used. I'm talking about listening to a fade, knowing that it's a fade, the edit itself is performed musically and acurately, and hearing a distinct difference between the way PT crossfades and Nuendo/DP4 crossfades. I do believe Tom is right about the difference in the way the programs perform these processes, at least it justifies what i'm hearing anyway. I don't care about having identical files at .1dB blah blah because ya know what? that's not a real-life practical application, and that's what i'm concerned with. I understand that it's the scientific way of A/Bing something, and if you wish to do that yourself, go for it. But I can hear a difference and I don't need a test to justify what I hear.

regarding the converter thing I posted, i'm prety sure that what i'm hearing is the non-linearity in the analog section of the Lucid. When i've sat in front of speakers for 10 hours hearing the same song over and over, tweaking and tweazing, and then i give my converter a little more juice since in that particular instance the mix was a bit lower in level than usual, and suddenly I hear a subtle yet noticeable shift in the HZ response of the overall mix, i'm concerned. I'm constantly changing my volume when I work, from whisper soft to a good 80dB, and i know all about volume affecting Hz response. But this was no level mismatch, it clearly affected things, significantly enough to justify a post IMHO!