Old 08-13-2011, 10:25 AM   #81
Justin
Administrator
 
Justin's Avatar
 
Join Date: Jan 2005
Location: NYC
Posts: 10,247
Default

Quote:
Originally Posted by alanofoz View Post
The purpose of my original post was to point out the very significant mathematical difference between integer and floating point. I find it very difficult to see how integer mixing could ever be preferred to floating point. I'd love to see some examples posted for ABX testing.
I think generally the 48 bit integer mode ("39" bit is what we call it, since it uses 9.39 bits like a certain other DAW) is preferable in many respects to the 32 bit floating point mode... primarily just for the additional bits of mantissa (having sacrificed some range at the extremes). At any rate, 64 bit FP ftw.
Justin is offline   Reply With Quote
Old 08-13-2011, 10:33 AM   #82
bennetng
Human being with feelings
 
Join Date: Jun 2010
Location: Hong Kong
Posts: 244
Default

http://www.sendspace.com/file/7k2mq3

I am offering another test:
1.render the project to the preferred bit depth (32fp/64fp/24int...) *in 48khz*
2.compare the rendered files by abxing or nulling or analyzing the differences with anything you have.
3.share your results

project is saved with reaper v4.0 rc4
__________________
My game and music channel
http://www.nicovideo.jp/mylist/8322681
bennetng is offline   Reply With Quote
Old 08-13-2011, 04:54 PM   #83
alanofoz
Human being with feelings
 
alanofoz's Avatar
 
Join Date: Sep 2009
Location: Oz - Blue Mountains NSW, formerly Geelong
Posts: 612
Default

Quote:
Originally Posted by Justin View Post
I think generally the 48 bit integer mode ("39" bit is what we call it, since it uses 9.39 bits like a certain other DAW) is preferable in many respects to the 32 bit floating point mode... primarily just for the additional bits of mantissa (having sacrificed some range at the extremes). At any rate, 64 bit FP ftw.
Thanks Justin, especially the last sentence.

In my earlier posts I was forgetting that we were talking about 39 bit (mea culpa), so *in my mind* I was comparing 32 bit float with 24 bit int, where 32 bit float is clearly the winner. By extension it seems to me that 64 bit float must be superior to 39 bit int.

I'm too far away from home at the moment to try out any of this.
alanofoz is offline   Reply With Quote
Old 08-13-2011, 07:15 PM   #84
Billoon
Human being with feelings
 
Join Date: Sep 2006
Location: Arse end of the earth.
Posts: 2,980
Default

OK, here's a project with 2 files, 1 rendered in 64bf mode and 1 rendered in 39bi mode, that are otherwise identical.

Press 'alt+c' to toggle solo between the tracks. (Note: make sure you have 'Options>Solo in front' unticked).

Can you really(repeatably, every time) tell the difference?

http://www.sendspace.com/file/n8mrer
Billoon is offline   Reply With Quote
Old 08-14-2011, 02:29 AM   #85
bblue
Human being with feelings
 
Join Date: Mar 2010
Location: El Cajon, CA (San Diego)
Posts: 593
Default

Quote:
Originally Posted by Billoon View Post
OK, here's a project with 2 files, 1 rendered in 64bf mode and 1 rendered in 39bi mode, that are otherwise identical.

Press 'alt+c' to toggle solo between the tracks. (Note: make sure you have 'Options>Solo in front' unticked).

Can you really(repeatably, every time) tell the difference?

http://www.sendspace.com/file/n8mrer
Probably not with that source material. With well recorded full-range source, probably. GIGO.

--Bill
bblue is offline   Reply With Quote
Old 08-14-2011, 02:44 AM   #86
timlloyd
Human being with feelings
 
Join Date: Mar 2010
Posts: 4,713
Default

I couldn't, and while I realise that I may be wrong I very much doubt that anyone could because they cancel down to below -200dBFS.

I would be fascinated if it could be proved otherwise.
timlloyd is offline   Reply With Quote
Old 08-14-2011, 03:00 AM   #87
jens
Human being with feelings
 
jens's Avatar
 
Join Date: Feb 2006
Location: Basel, Switzerland
Posts: 4,434
Default

Quote:
Originally Posted by bblue View Post
Probably not with that source material. With well recorded full-range source, probably. GIGO.

--Bill
ah... you would hear a difference only when using Neumanns, Neve, Monstercables, right?

So let's hear some of your awesomely recorded, non-garbage, pristine sounds with which we all will spot the difference with ease! Giz a choon!
jens is offline   Reply With Quote
Old 08-14-2011, 04:21 AM   #88
Billoon
Human being with feelings
 
Join Date: Sep 2006
Location: Arse end of the earth.
Posts: 2,980
Default

Quote:
Originally Posted by bblue View Post
Probably not with that source material. With well recorded full-range source, probably. GIGO.

--Bill
So that's a (qualified) no then.

Can you explain how the source material could detract from the (implied) 'obvious superiority' of the integer mode?

...and how 'better' source material wouldn't just null below -200db but would actually extend the difference into the audible range?
Billoon is offline   Reply With Quote
Old 08-14-2011, 04:46 AM   #89
Subz
Human being with feelings
 
Subz's Avatar
 
Join Date: Jun 2006
Location: UK
Posts: 2,919
Default

Quote:
Originally Posted by timlloyd View Post
You're not required to call me "Sir" except on special occasions.



glad you took my post as it was intended

Subz
Subz is offline   Reply With Quote
Old 08-14-2011, 05:57 AM   #90
Lawrence
Human being with feelings
 
Join Date: Mar 2007
Posts: 20,876
Default

Reading with interest. Can someone explain this to me?

Is integer mode used throughout Reaper in that case or is this bus summing? How does it all work, is it also plugin DSP?
Lawrence is offline   Reply With Quote
Old 08-14-2011, 06:22 AM   #91
timlloyd
Human being with feelings
 
Join Date: Mar 2010
Posts: 4,713
Default

These might help:

Quote:
Originally Posted by Justin View Post
The integer modes involve losing information before mixing.. so in theory you can end up with slightly different noise qualities. Some people seem to swear by this difference, though I have my doubts. Without doing the math, but guesstimating it, in 24 bit mode you would need about 128 tracks before the biggest possible difference in signal between 64FP and 24 int would get close to -90dB.
Not noise though, signal-correlated distortion.

Quote:
Originally Posted by schwa View Post
Regarding integer track mixing modes -- if there is a measurable difference in the audio quality, it should be attributable only to the fact that the integer-based modes accumulate tiny amounts of truncation noise.

Personally I don't see a DSP-based "sound quality" argument for wanting to accumulate truncation noise, but REAPER is all about options. If a particular user feels that integer summing results in different or better-quality noise, they can use the mixing bit depth options.
Quote:
Originally Posted by Justin View Post
Right, though in this case if you have the 12 bit mode set, it just forces the output of each non-master track to 12 bit, so if you have master FX or the master fader set to some value then you might have activity in the lower bits as well.
So to sum up, the int modes truncate the output of each non-master track and in doing so cause signal-correlated distortion, the amount of which depends on the int mode you choose ... lower bit-depth = more distortion.

Plug-ins and other processing remains 64-float.

To hear what this distortion sounds like, just take an existing project, set the mixing bit-depth to 8int and hit play. The distortions that you hear are the same that will exist in the 39int mode, but due to the massive range of numbers that 39int can express, the distortion will be so far below 0dBFS that it is completely irrelevant for anything that remotely resembles normal musical DAW use.
timlloyd is offline   Reply With Quote
Old 08-14-2011, 06:26 AM   #92
bblue
Human being with feelings
 
Join Date: Mar 2010
Location: El Cajon, CA (San Diego)
Posts: 593
Default

Quote:
Originally Posted by jens View Post
ah... you would hear a difference only when using Neumanns, Neve, Monstercables, right?

So let's hear some of your awesomely recorded, non-garbage, pristine sounds with which we all will spot the difference with ease! Giz a choon!
The higher quality the original recording, the more obvious it is when it has been degraded (in any form).

There are other factors as well. How good of a listener a person is (lots of listening/mixing experience and has heard good reference quality on which to base an opinion) and how well their monitoring (speakers, amp, and final D/A conversion) can resolve this type of detail. I won't name names but there are various combinations of amplifiers and speakers out there being used for monitoring that can't resolve much detail at all.

The difference between 64FP and 39i here was obvious, just on a quick test within the project I'm currently working on. Good, but not great audio. Where is was the most obvious in this case was during hard strumming of an acoustic guitar. The upper harmonics and ring-out of the chords were far more obvious at 64FP, whereas they took on a more metallic sound that blended with the hard attack of the chord at 39i. 32FP, I didn't hear any obvious differences with this source at this time.

Once you recognize how this degradation sounds, it's a whole lot easier to spot it on a variety of source and different listening conditions.

--Bill
bblue is offline   Reply With Quote
Old 08-14-2011, 06:35 AM   #93
Billoon
Human being with feelings
 
Join Date: Sep 2006
Location: Arse end of the earth.
Posts: 2,980
Default

Quote:
Originally Posted by bblue View Post
The difference between 64FP and 39i here was obvious, just on a quick test within the project I'm currently working on. Good, but not great audio.
OK, well if its actually audible then render the track to stem at both 64f and 39i and null them and the difference should be audible...if its not audible not, with respect, you're just imagining it.

If it is audible, upload the difference file so we can hear it please.
Billoon is offline   Reply With Quote
Old 08-14-2011, 07:14 AM   #94
Subz
Human being with feelings
 
Subz's Avatar
 
Join Date: Jun 2006
Location: UK
Posts: 2,919
Default

Quote:
Originally Posted by Billoon View Post
OK, well if its actually audible then render the track to stem at both 64f and 39i and null them and the difference should be audible...if its not audible not, with respect, you're just imagining it.

If it is audible, upload the difference file so we can hear it please.
+1

i smell cod! but would love to be proved wrong!?



Any how if it dose or dose not sound different (as better is subjective) makes no difference as options is the Reaper Way! & i love it!

as i could hear no difference i'm going to run in 39bit mode for a while as its the "Pro" way lol
Subz is offline   Reply With Quote
Old 08-14-2011, 07:27 AM   #95
timlloyd
Human being with feelings
 
Join Date: Mar 2010
Posts: 4,713
Default

Quote:
Originally Posted by Subz View Post
as i could hear no difference i'm going to run in 39bit mode for a while as its the "Pro" way lol
Not anymore it's not ... that ol' chestnut now uses 64float
timlloyd is offline   Reply With Quote
Old 08-15-2011, 04:28 AM   #96
vinx
Human being with feelings
 
Join Date: May 2007
Location: France
Posts: 662
Default

there's something in the summing engine behaviour that makes me ask myself:

It is "relative" based:
If you clip a track output to reach "+2" and then lower the master to "-2.1", you won't have any clipping at all.
This at first sight seems to be a cool feature, but the counterpart is that any time a clip really occurs (the master isn't lowered as much), which happens quiet often in the "home studio"'s world, then the whole mix is clipped! (all instruments are distorded).

If the summing engine would be what I'd call "straight", then if a track (say the kick) reach "+2", this track output would be clipped, even if you lower the master. Then , everytime a clip would occur (more often than you may think, in the "real world"), it would clip only the related track. Your kick would not distord every instrument anymore...
So everytime a clip would occur, "straight" would sound better than "relative"

Moreover you could easily hear a clip as it occurs when mixing, and you'd know it because anytime the meter would go red, you would have distortion for real, just as every audio gear did until now.
I think it would also lead to best mixing habits for the "average" mixer, and maybe less clipped mixes received by mastering engineers.

An other feature of "straight" mixing engine would be the possibility for "Lo fi" obsessed guys to clip natively each instrument at its own level, which could lead to more insane Lo-fi !!!!!

so what?

Last edited by vinx; 08-15-2011 at 04:47 AM.
vinx is offline   Reply With Quote
Old 08-15-2011, 05:24 AM   #97
jens
Human being with feelings
 
jens's Avatar
 
Join Date: Feb 2006
Location: Basel, Switzerland
Posts: 4,434
Default

oh boy... where to start?

your idea of digital summing (or summing in general) is all upside down, it appears.

The master clips because it clips, that's why it clips - i.e. the outpout exceeds 0dB. Clipping means exceeding odB and if the sum of all the signals a mix consists of exceeds 0dB then the master clips. If the sum after the master fader doesn't exceed odB then the output doesn't clip, no matter how many individual tracks might ever clip.

So you can have 150 tracks clipping and still not have the outputclipping, just as long as you pull down the master fader far enough. And if the output clips, then you just need to appropriately pull down the master fader.

B.t.w.: in a digital enviroment, such as Reaper, it actually doesn't matter whether the individual tracks clip or not.

But: effect plugins such as many compressors, saturators, etc. which apply non-linear processes still depend on proper gain-staging, which means you yourself have to take care of this during mixing, all clipping aside. A track can have completely fucked up gain-staging (and thus sound shite) while it doesn't clip at all. Likewise a track can clip and still have proper gain-staging and sound marvellous.
jens is offline   Reply With Quote
Old 08-15-2011, 05:41 AM   #98
vinx
Human being with feelings
 
Join Date: May 2007
Location: France
Posts: 662
Default

Quote:
Originally Posted by jens View Post
oh boy... where to start?

your idea of digital summing (or summing in general) is all upside down, it appears.
I don't think so, but my frenchy-english may be confusing .

Quote:
Originally Posted by jens View Post
So you can have 150 tracks clipping and still not have the outputclipping, just as long as you pull down the master fader far enough. And if the output clips, then you just need to appropriately pull down the master fader.
That's the "relative"'s feature. But I can tell you most people don't pull down the master, or worse, they mix with a limiter on it.

Quote:
Originally Posted by jens View Post
B.t.w.: in a digital enviroment, such as Reaper, it actually doesn't matter whether the individual tracks clip or not.
False, if the master isn't correctly set, which is often happening.

Quote:
Originally Posted by jens View Post
But: effect plugins such as many compressors, saturators, etc. which apply non-linear processes still depend on proper gain-staging, which means you yourself have to take care of this during mixing, all clipping aside. A track can have completely fucked up gain-staging (and thus sound shite) while it doesn't clip at all. Likewise a track can clip and still have proper gain-staging and sound marvellous.
not related to the subject
vinx is offline   Reply With Quote
Old 08-15-2011, 05:56 AM   #99
jens
Human being with feelings
 
jens's Avatar
 
Join Date: Feb 2006
Location: Basel, Switzerland
Posts: 4,434
Default

Quote:
Originally Posted by vinx View Post
I don't think so, but my frenchy-english may be confusing .


No, your English is fine. It's just that your proposal of 'straight' summing is akin to this:

jens is offline   Reply With Quote
Old 08-15-2011, 07:13 AM   #100
n0rd
Human being with feelings
 
n0rd's Avatar
 
Join Date: Dec 2010
Location: Down Under
Posts: 396
Default

I *love* Escher pics!!!
n0rd is offline   Reply With Quote
Old 08-15-2011, 07:43 AM   #101
JasonSpatola
Human being with feelings
 
Join Date: Aug 2009
Posts: 257
Default

Quote:
Originally Posted by bblue View Post
The difference between 64FP and 39i here was obvious, just on a quick test within the project I'm currently working on. Good, but not great audio. Where is was the most obvious in this case was during hard strumming of an acoustic guitar. The upper harmonics and ring-out of the chords were far more obvious at 64FP, whereas they took on a more metallic sound that blended with the hard attack of the chord at 39i. 32FP, I didn't hear any obvious differences with this source at this time.
Upload guitar track please?
JasonSpatola is offline   Reply With Quote
Old 08-15-2011, 07:49 AM   #102
timlloyd
Human being with feelings
 
Join Date: Mar 2010
Posts: 4,713
Default

ahhh Escher ...

I understand what Vinx is saying. However, I don't think there is a problem with how things function at the moment.

There is a problem with the widespread misunderstanding of the importance of safe gain staging habits within audio software and the workflow habits which lead to users touching 0dBFS on individual tracks and their master output.

It's not the software that needs to change, it's these bad habits. Bad habits "within the home studio world" are not a valid reason for changing the way reaper or any other DAW works just to coddle people into continuing on with these habits ...

Maybe Cockos could have a little section on the website that explains this stuff to people who are so far "not getting it?" Ethan Winer is currently writing a book that might include such things, as is Skip Burrows from Sunrise Sound (which deals specifically with this stuff).
timlloyd is offline   Reply With Quote
Old 08-15-2011, 08:21 AM   #103
karbomusic
Human being with feelings
 
karbomusic's Avatar
 
Join Date: May 2009
Posts: 18,831
Default

Quote:
It's not the software that needs to change, it's these bad habits. Bad habits "within the home studio world" are not a valid reason for changing the way reaper or any other DAW works just to coddle people into continuing on with these habits ...
+1

"sometimes" misinformation out of habit completely overwhelms everything else.
__________________
Your whole life people will tell you what you can't do. Getting past them is the first step to actually getting things done.
karbomusic is online now   Reply With Quote
Old 08-15-2011, 10:14 AM   #104
vinx
Human being with feelings
 
Join Date: May 2007
Location: France
Posts: 662
Default

ok ok...
What could be realistically done:
Change the master meter so it will be "ultra-red" compared to track meters, with the word "clip" flashing
A rendering option, on standard: "prevent output clipping" that would simply do that by lowering the master to the right level, just while rendering the file.
vinx is offline   Reply With Quote
Old 08-15-2011, 01:28 PM   #105
Billoon
Human being with feelings
 
Join Date: Sep 2006
Location: Arse end of the earth.
Posts: 2,980
Default

Quote:
Originally Posted by vinx View Post
ok ok...
What could be realistically done:..
This is getting a bit OT here but i think to start with the idea that its ok to exceed 0db in float ANYWHERE in the signal path(because you can turn it down later) should be debunked.

There is no reason ever to exceed 0db and there are good reasons not to (eg. proper gain staging, prevent clipping plugins etc).

I've even toyed with the idea of FR an option to remove the fader on the master (but leave the meter) which would prevent this behavior(letting tracks exceed 0db then 'fixing' it by turning down the master fader). Mine never leaves 0db.

If you want a fade out, you could still automate a volume plugin.
Billoon is offline   Reply With Quote
Old 08-15-2011, 01:34 PM   #106
timlloyd
Human being with feelings
 
Join Date: Mar 2010
Posts: 4,713
Default

Absolutely Billoon ... there is never a reason to exceed 0dBFS.

Re. the master track stuff, it's now possible with WALTER in v4.

Here's my master track:

[IMG]http://img690.**************/img690/2746/mastertrack.png[/IMG]

My master output never clips ... I make sure of that through good habits

Last edited by timlloyd; 08-15-2011 at 02:19 PM. Reason: spelling :)
timlloyd is offline   Reply With Quote
Old 08-15-2011, 01:44 PM   #107
JasonSpatola
Human being with feelings
 
Join Date: Aug 2009
Posts: 257
Default

Quote:
Originally Posted by timlloyd View Post
there is never a reason to exceend 0dBFS.
Uhhh, what if you're covering songs from Rush's 2002 release Vapour Trails?
JasonSpatola is offline   Reply With Quote
Old 08-15-2011, 02:18 PM   #108
timlloyd
Human being with feelings
 
Join Date: Mar 2010
Posts: 4,713
Default

Do it better than Rush ...
timlloyd is offline   Reply With Quote
Old 08-15-2011, 02:29 PM   #109
griz lee
Human being with feelings
 
griz lee's Avatar
 
Join Date: Jul 2006
Location: in a hotel room near you
Posts: 1,175
Default

Quote:
Originally Posted by timlloyd View Post
ahhh Escher ...

Ethan Winer is currently writing a book that might include such things
Which I certainly won't be buying after seeing what went down at the daw myths presentation at AES a while ago.
griz lee is offline   Reply With Quote
Old 08-15-2011, 02:42 PM   #110
Subz
Human being with feelings
 
Subz's Avatar
 
Join Date: Jun 2006
Location: UK
Posts: 2,919
Default

1 solution is a option to default all new tracks to -3db/-6db/-9db/-12db (maybe a option in the panning law?) [or dose this option already exist?]

i admit i use bad habits!

my master often peaks (even though my tracks don't) & i add a simple volume plugin as the first FX on the master & turn down from there (i always leave my master at 0db & make sure the only FX that peak on the master is the first volume fx)

after reading this thread i will be more mindful of my levels from now on

Subz
Subz is offline   Reply With Quote
Old 08-15-2011, 03:01 PM   #111
Justin
Administrator
 
Justin's Avatar
 
Join Date: Jan 2005
Location: NYC
Posts: 10,247
Default

Quote:
Originally Posted by timlloyd View Post
Not noise though, signal-correlated distortion.
Hmm, I'd still call it noise, just not white or pink or any other specific kind of noise. It would be signal-correlated noise.

The rationale for not calling it be distortion is that the difference is not correlated with the amplitude or power of the signal. If you're going to 8-bit mode, even, the added noise volume will not change meaningfully whether you are pushing +0dBFS or -30dBFS... It really is the equivalent of summing the good (full quality) signal with some limited peak volume, but signal dependent noise.

Distortion, to me, implies some nonlinear process...
Justin is offline   Reply With Quote
Old 08-15-2011, 03:05 PM   #112
griz lee
Human being with feelings
 
griz lee's Avatar
 
Join Date: Jul 2006
Location: in a hotel room near you
Posts: 1,175
Default Err, sorry.

Right Tim. I've got to give it to you, you're right.

I had a minute just now to investigate my mix from the other day, which started me worrying about this whole business.

Good news is, yes, I'm not deluding myself. My 39i mix and my 64fp mix do actually sound subtly different. Not by miles, but different!

Unfortunatlely, whilst I had printed my reverbs (which i have to do anyway before I sum to a file as I'm using hardware) I had completely overlooked a little line-trim plugin, which on investigation, I find was adding random noise above the 24bit threshold at -108db. I have this over a lot of tracks in my mix, to just get the faders in the right place.

Unfortunately, the mix I printed with 39i enabled is, for some reason, better sounding. However I conclude that it's not the fact it is an integer that is the cause - just a more 'pleasing' set of random low level summings. If I remove the line trim plug, I am unable to tell the difference between the 39i mix and the 64bit fp mix.

Phew. I like the options though. But I think I'm cured of the need for a >39i setting. I will be looking for a new gain plugin though..

Apologies for any inconvenience caused in the process of arriving at this conclusion (:
griz lee is offline   Reply With Quote
Old 08-15-2011, 03:43 PM   #113
timlloyd
Human being with feelings
 
Join Date: Mar 2010
Posts: 4,713
Default

Quote:
Originally Posted by Justin View Post
Hmm, I'd still call it noise, just not white or pink or any other specific kind of noise. It would be signal-correlated noise.

The rationale for not calling it be distortion is that the difference is not correlated with the amplitude or power of the signal. If you're going to 8-bit mode, even, the added noise volume will not change meaningfully whether you are pushing +0dBFS or -30dBFS... It really is the equivalent of summing the good (full quality) signal with some limited peak volume, but signal dependent noise.

Distortion, to me, implies some nonlinear process...
That makes sense. I did write a bit of a post, but actually you're right and correlated noise seems more accurate It's a bit like noise gated by the signal.

Quote:
Originally Posted by griz lee View Post
a little line-trim plugin, which on investigation, I find was adding random noise above the 24bit threshold at -108db. I have this over a lot of tracks in my mix, to just get the faders in the right place.

Unfortunately, the mix I printed with 39i enabled is, for some reason, better sounding. However I conclude that it's not the fact it is an integer that is the cause - just a more 'pleasing' set of random low level summings. If I remove the line trim plug, I am unable to tell the difference between the 39i mix and the 64bit fp mix.
That's rather interesting I'm still surprised that you can hear differences down that far though, especially since the noise from the trim plug-in presumably has the same distribution in each render ... I'm really going to have to delve into this more deeply soon, perception is pretty fascinating. Need to do some more abx-ing myself.

Last edited by timlloyd; 08-15-2011 at 03:58 PM.
timlloyd is offline   Reply With Quote
Old 08-15-2011, 04:07 PM   #114
griz lee
Human being with feelings
 
griz lee's Avatar
 
Join Date: Jul 2006
Location: in a hotel room near you
Posts: 1,175
Default

I'm still not talking about anything on a level that would make me do a remix. The best I can describe it is it like evaluating the differences between dithers.
griz lee is offline   Reply With Quote
Old 08-15-2011, 04:20 PM   #115
Guod3
Human being with feelings
 
Guod3's Avatar
 
Join Date: Jan 2008
Posts: 419
Default

Quote:
Originally Posted by griz lee View Post
Right Tim. I've got to give it to you, you're right.

I had a minute just now to investigate my mix from the other day, which started me worrying about this whole business.

Good news is, yes, I'm not deluding myself. My 39i mix and my 64fp mix do actually sound subtly different. Not by miles, but different!

Unfortunatlely, whilst I had printed my reverbs (which i have to do anyway before I sum to a file as I'm using hardware) I had completely overlooked a little line-trim plugin, which on investigation, I find was adding random noise above the 24bit threshold at -108db. I have this over a lot of tracks in my mix, to just get the faders in the right place.

Unfortunately, the mix I printed with 39i enabled is, for some reason, better sounding. However I conclude that it's not the fact it is an integer that is the cause - just a more 'pleasing' set of random low level summings. If I remove the line trim plug, I am unable to tell the difference between the 39i mix and the 64bit fp mix.

Phew. I like the options though. But I think I'm cured of the need for a >39i setting. I will be looking for a new gain plugin though..

Apologies for any inconvenience caused in the process of arriving at this conclusion (:
priceless!

When people start claiming to hear the difference between signals (that in fact null to zilch), then science is all we have left!
Guod3 is offline   Reply With Quote
Old 08-15-2011, 04:30 PM   #116
griz lee
Human being with feelings
 
griz lee's Avatar
 
Join Date: Jul 2006
Location: in a hotel room near you
Posts: 1,175
Default

Quote:
Originally Posted by Guod3 View Post
priceless!

When people start claiming to hear the difference between signals (that in fact null to zilch), then science is all we have left!
I couldnt hear the difference after having conducted a valid test. I made an error in the test process.
griz lee is offline   Reply With Quote
Old 08-15-2011, 04:56 PM   #117
Guod3
Human being with feelings
 
Guod3's Avatar
 
Join Date: Jan 2008
Posts: 419
Default

yes, indeed you did. (I didn't mean to sound contrary)
Interesting thread.
Guod3 is offline   Reply With Quote
Old 08-15-2011, 05:05 PM   #118
timlloyd
Human being with feelings
 
Join Date: Mar 2010
Posts: 4,713
Default

I'm going to try something tomorrow that has probably been investigated before that I'm not aware of ...

Abx two different files of the same distribution of noise ... I don't expect to be able to tell them apart, but we'll see ...

hmm ... could be interesting
timlloyd is offline   Reply With Quote
Old 08-15-2011, 11:07 PM   #119
jens
Human being with feelings
 
jens's Avatar
 
Join Date: Feb 2006
Location: Basel, Switzerland
Posts: 4,434
Default

Quote:
Originally Posted by Billoon View Post
This is getting a bit OT here but i think to start with the idea that its ok to exceed 0db in float ANYWHERE in the signal path(because you can turn it down later) should be debunked.

There is no reason ever to exceed 0db and there are good reasons not to (eg. proper gain staging, prevent clipping plugins etc).

Quote:
Originally Posted by jens View Post
B.t.w.: in a digital enviroment, such as Reaper, it actually doesn't matter whether the individual tracks clip or not.

But: effect plugins such as many compressors, saturators, etc. which apply non-linear processes still depend on proper gain-staging, which means you yourself have to take care of this during mixing, all clipping aside.A track can have completely fucked up gain-staging (and thus sound shite) while it doesn't clip at all. Likewise a track can clip and still have proper gain-staging and sound marvellous.


If I have spent time to take care that my track has proper gain-staging and I later on think this track is not loud enough in the mix, then I will
simply pull up the track fader, instead of messing with the gain-stage once again. And if it then clips, then so be it.
There's no reason to avoid it. Actually you could rather say it's the other way around: In order to preserve your carefully adjusted gain-stage, you occassionally have to clip some tracks. And there is no negative side effect to it at all.

Getting rid of that myth helps in understanding what proper gain-staging actually means.
jens is offline   Reply With Quote
Old 08-15-2011, 11:13 PM   #120
jens
Human being with feelings
 
jens's Avatar
 
Join Date: Feb 2006
Location: Basel, Switzerland
Posts: 4,434
Default

Quote:
Originally Posted by timlloyd View Post

My master output never clips ... I make sure of that through good habits
What's the big deal? As if that would be any difficult. Either you hear it (you usually will,) or Reaper notifies you about the clipping during rendering.

Managing to get a clipped file out of Reaper most probably would mean you're approximately braindead. It's virtually impossible, even if you're an idiot of tremendous proportions.
jens is offline   Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -7. The time now is 02:19 PM.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2017, vBulletin Solutions Inc.