Go Back   Cockos Incorporated Forums > REAPER Forums > REAPER General Discussion Forum

Reply
 
Thread Tools Display Modes
Old 03-09-2019, 09:25 PM   #1
Tesgin
Human being with feelings
 
Join Date: Mar 2011
Posts: 222
Default DAW says peaks are +0.3, but true peaks are -0.6

Title of post says it all. My DAW (REAPER) tells me that my mix is clipping, 0.3. However, my LUFS meter, which is YouLean Loudness Meter 2, says my true peak is -0.6. I also have a limiter on my master buss, with a ceiling set to -0.7.

Ugh. What am I missing? I’m “thinking” that that’s not unusual, and that a slight bit of clipping is okay, but that what I really wanna watch is my true peaks, cuz that’s what’s gonna determine whether it clips on another audio system after digital to audio conversion.

Am I correct in that? Or is something screwed up? (please tell me I’m correct)
Tesgin is offline   Reply With Quote
Old 03-09-2019, 11:57 PM   #2
mschnell
Human being with feelings
 
mschnell's Avatar
 
Join Date: Jun 2013
Location: Krefeld, Germany
Posts: 14,688
Default

Where exactly do you see these values ?

I use a Brick Wall Limiter set to -0.4 db at the end of the Maser FX chain, and when rendering, Reaper says -0,4 dB in the Rendering status value Window.

I did not try to use an independent tool to analyze the resulting wav file.

-Michael

Last edited by mschnell; 03-10-2019 at 12:42 AM.
mschnell is offline   Reply With Quote
Old 03-10-2019, 12:07 AM   #3
urednik
Human being with feelings
 
urednik's Avatar
 
Join Date: Apr 2010
Posts: 1,247
Default

Reaper meters almost always show clipped values on imported normalised/compressed soundtracks. It is a living organism. Unfortunately there is no solution to this than try living with it. This is a part of soundengineering life we have.
__________________
W10 (64) Lenovo E540 - SSD; Lenovo B590; W7 (32), Compaq 610 (2.1Ghz core 2 duo, L2 cache, 2GB RAM); DPA 4018, Schoeps MK2, Schoeps MTSC 64, Neumann mk184, AEA Ribbon 88mk, AKG SolidTUBE; Focusrite Scarlett 18i20, recording merely live acoustic music.
urednik is offline   Reply With Quote
Old 03-10-2019, 12:42 AM   #4
synkrotron
Human being with feelings
 
synkrotron's Avatar
 
Join Date: May 2015
Location: Warrington, UK
Posts: 1,444
Default

I must admit that I never look at the master track level anymore as I use my limiter, Pro-L, on the master bus to monitor mix level, which is set to a full -1 dB and on top of that I like to preserve dynamics and it is very rare that Pro-L has to actually limit.

And, finally, I never master within REAPER. Old habits die hard... I render the WAV and then open in Sound Forge for final trimming and fine level adjustment.
__________________
Bandcamp // YouTube // SoundCloud
synkrotron is offline   Reply With Quote
Old 03-10-2019, 12:45 AM   #5
mschnell
Human being with feelings
 
mschnell's Avatar
 
Join Date: Jun 2013
Location: Krefeld, Germany
Posts: 14,688
Default

A peak > 0 dB does not mean that it's clipped at that point. It can be a perfectly viable result of a re-sampling process, and will be processed perfectly well without any clipping artifacts.

-Michael
mschnell is offline   Reply With Quote
Old 03-10-2019, 07:45 PM   #6
Tesgin
Human being with feelings
 
Join Date: Mar 2011
Posts: 222
Default

Quote:
Originally Posted by mschnell View Post
A peak > 0 dB does not mean that it's clipped at that point. It can be a perfectly viable result of a re-sampling process, and will be processed perfectly well without any clipping artifacts.

-Michael
Thanks, all. To clarify, these are not imported files, but tracks containing recorded guitars, bass and vocal.

So, Michael, you're saying that it's okay and not unusual that the digital peaking indicator shows clipping. But the other thing I cannot figure out is why the digital peak is actually higher than the true peak in YouLean. I never saw that before. I thought the true peak will always be higher than the digital peak, no?

Second, why or how is the digital peak getting up past 0dB when I have a brickwall limiter as the last plugin on the mix buss? It's the Waves L1 + ultramaximizer, and the ceiling is set to -0.7 dB with the domain set as truepeak (but even when set at digital, it still goes beyond the ceiling!).
Tesgin is offline   Reply With Quote
Old 03-10-2019, 10:52 PM   #7
mschnell
Human being with feelings
 
mschnell's Avatar
 
Join Date: Jun 2013
Location: Krefeld, Germany
Posts: 14,688
Default

The common recommendation I read (and follow) is to set a brick wall limiter to -0.4 dB, if after same there is a format translation (i.e. "rendering") from floating point (i.e. >0 dB = no problem) to fixed point (0 dB = limit).

-Michael
mschnell is offline   Reply With Quote
Old 03-11-2019, 01:25 AM   #8
bigjoe
Human being with feelings
 
bigjoe's Avatar
 
Join Date: Jun 2015
Posts: 323
Default

clipped values are caused by Inter-sample peaks, ISP for short.
They usually does not appear in the "waveform", instead they are generated during playback because of interpolation approximation.
That's the reason why -1db True Peak is now the standard peak value for Youtube, Spotify, Itunes and almost all the streaming services.
Also you can go higher than -1dbTP by using an Oversampled limiter, or one with an ISP function.
Regular limiters does not catch ISPs
bigjoe is offline   Reply With Quote
Old 03-11-2019, 12:12 PM   #9
clepsydrae
Human being with feelings
 
clepsydrae's Avatar
 
Join Date: Nov 2011
Posts: 3,409
Default

Quote:
Originally Posted by bigjoe View Post
That's the reason why -1db True Peak is now the standard peak value for Youtube, Spotify, Itunes and almost all the streaming services.
Also you can go higher than -1dbTP by using an Oversampled limiter, or one with an ISP function.
Regular limiters does not catch ISPs
"True Peak" and ISP are the same thing, by most definitions, hence the OP's confusion. I share that confusion: by my understanding ISP should be >= digital peak.

OP: have you confirmed that your master fader is at 0.00dbFS? Have you right clicked on the master track fader in the mix panel and made sure you understand the various settings there? E.g. the "clipping" you're seeing is actually instantaneous clipping and not RMS over a threshold, etc?

You could either upload the entire project somewhere so we can look at it, or if that's not feasible you could try to make a reduced version; e.g. find the spot where the peak happens, clip everything to one second around that spot, render the master channel to a 32bit float WAV file, import that in to a new project and see if you can reproduce, etc.

First guess would be user error of some kind, but second guess is a bug in a plugin.

(Unless someone enlightens us as to how reaper's peak metering works.)
clepsydrae is offline   Reply With Quote
Old 03-11-2019, 01:37 PM   #10
DVDdoug
Human being with feelings
 
Join Date: Jul 2010
Location: Silicon Valley, CA
Posts: 2,779
Default

I don't know what's wrong but the "true peak" CAN'T be lower than the digital-numerical peak.


….Personally, I'm not worried about the "true peak". There is no inter-sample information in the digital audio data. Inter-sample peaks can ONLY exist in the reconstructed analog waveform.* The true-peak meter is calculating (estimating?) what the analog peak will be.


The digital-side is numerically hard-limited to 0dB. In integer formats, 0dB is defined as the maximum you can "count to" with a given number of bits so integer audio data simply cannot go over 0dB.

DACs are integer-based so you can't feed a "number" bigger than 0dB into a DAC. If you have floating-point data over 0dB, it will be clipped when the driver converts to integer.

But, there is no reason the continuous analog waveform on the analog-side of the DAC can't peak over 0dB without clipping. ...Some DACs may not accurately reproduce the true-peaks over 0dB but there's no engineering, scientific, or mathematical reason that it can't be done.







* You can also get higher peaks if you re-sample.
DVDdoug is online now   Reply With Quote
Old 03-16-2019, 07:21 PM   #11
brainwreck
Human being with feelings
 
Join Date: Jul 2006
Posts: 20,859
Default

Quote:
Originally Posted by DVDdoug View Post
….Personally, I'm not worried about the "true peak". There is no inter-sample information in the digital audio data. Inter-sample peaks can ONLY exist in the reconstructed analog waveform.* The true-peak meter is calculating (estimating?) what the analog peak will be.
I think talk of inter-sample peaks can be confusing by the name alone but also because the term has been thrown around so much without maybe any explanation for what it means. It should be called something like reconstruction peaks, because inter-sample peaks happen at the DAC reconstruction filter, not having anything to do with anything at the software level. I have seen it said before that inter-sample peaks are happening in the daw software (that isn't what is happening!).
__________________
It's time to take a stand against the synthesizer.
brainwreck is offline   Reply With Quote
Old 03-16-2019, 07:42 PM   #12
clepsydrae
Human being with feelings
 
clepsydrae's Avatar
 
Join Date: Nov 2011
Posts: 3,409
Default

They can happen via oversampling (in software) as well. (Edit: although whether or not that's an issue depends on what happens to the data; often it's going to be floating point and not matter.)
clepsydrae is offline   Reply With Quote
Old 03-18-2019, 03:41 PM   #13
Tesgin
Human being with feelings
 
Join Date: Mar 2011
Posts: 222
Default

Quote:
Originally Posted by clepsydrae View Post
"True Peak" and ISP are the same thing, by most definitions, hence the OP's confusion. I share that confusion: by my understanding ISP should be >= digital peak.

OP: have you confirmed that your master fader is at 0.00dbFS? Have you right clicked on the master track fader in the mix panel and made sure you understand the various settings there? E.g. the "clipping" you're seeing is actually instantaneous clipping and not RMS over a threshold, etc?

You could either upload the entire project somewhere so we can look at it, or if that's not feasible you could try to make a reduced version; e.g. find the spot where the peak happens, clip everything to one second around that spot, render the master channel to a 32bit float WAV file, import that in to a new project and see if you can reproduce, etc.

First guess would be user error of some kind, but second guess is a bug in a plugin.

(Unless someone enlightens us as to how reaper's peak metering works.)
I am SO sorry I haven't replied for a bit. I've been very swamped and pretty much out of the loop. Excellent input. THANK YOU all.

So, yes, I continue to be stumped by this project. I am suspecting it has to do with some inaccurate metering problems, based on the comments herein.

I did as you suggested and rendered the project to a wav file, just around the section where the problem occurs. Here's the link to the wav file:

https://we.tl/t-j2fRvOy2uE

So, in the original project, my master volume is at 0.0. The meter peak clips at +0.2, IN SPITE of the fact that I have L1 as the last thing on the master buss (well, except YouLean after that, but that's only a meter) and the ceiling is set to -0.7. So, that's my first question is why it's clipping if I have a brickwall limiter at -0.7. But it does clip, even though (my second question), YouLean is giving me a true peak reading of -0.7. How? As DVDdoug said, the true peak can't be lower than the digital-numerical peak.

When I load that wave file back into a new project, the master meter peaks at 0.0; AND YouLean now gives a true peak of +0.1. UGH! Makes no sense to me.

Thank you for your suggestion, clepsydrae, of perhaps uploading the project for someone to look at. That'd be awesome. Here's the link to that zip file (virtually all of the plugins are Waves' plugins, plus YouLean, SlickEQ M, and TDR Kotelnikov, and FreeG, and some ReaPlugs). It's 426.1 MB:

https://we.tl/t-iGjrnnsJg3

Oh, and I can add that if I lower the L1 so that the ceiling is -2.0, the meter on the master buss no longer clips, but still goes to -1.0. Same with JS: hard limiter. Set at -2.0, I also get a peak reading on the master buss of -1.0. Ugh again.

I know I can make my problem go away by just pulling my faders down and steering "way clear" of clipping, but it puzzles me and bugs me that the metering is off. That the signal peak is getting past my brickwall limiter, that the digital peak is louder than the true peak in YouLean. That the reading on the rendered wav is different than what the meters in the project said.

It really seems like some kind of anomaly in the metering to me. Some kind of glitch somewhere, either in REAPER or YouLean or the Waves plugs. Thank you all for helping me with this. Just trying to understand and learn.
Tesgin is offline   Reply With Quote
Old 03-19-2019, 10:23 AM   #14
xpander
Human being with feelings
 
xpander's Avatar
 
Join Date: Jun 2007
Location: Terra incognita
Posts: 7,670
Default

Quote:
Originally Posted by Tesgin View Post
Oh, and I can add that if I lower the L1 so that the ceiling is -2.0, the meter on the master buss no longer clips, but still goes to -1.0. Same with JS: hard limiter. Set at -2.0, I also get a peak reading on the master buss of -1.0.

I know I can make my problem go away by just pulling my faders down and steering "way clear" of clipping, but it puzzles me and bugs me that the metering is off. That the signal peak is getting past my brickwall limiter, that the digital peak is louder than the true peak in YouLean. That the reading on the rendered wav is different than what the meters in the project said.
One thing to remember is that limiters are not equal when it comes to True Peak settings/readings. Here's some food for thought:
https://www.saintpid.se/en/isp-true-peak-limiters-test/

I loaded up your example project, but since I'm missing almost all of the FX you have in there, I don't get the real conditions for a test.

You have Sonarworks Reference 4 after the hard limiter, before the Youlean Meter. Any difference if you bypass that one?
xpander is offline   Reply With Quote
Old 03-19-2019, 10:43 AM   #15
poetnprophet
Human being with feelings
 
poetnprophet's Avatar
 
Join Date: Jan 2018
Posts: 1,651
Default

This is something I have noticed in cubase, sonar/cakewalk, and reaper. I have just gotten used to the fact I need to lower the input/clip gain or track gain whenever I import anything....it's the first thing I check. It is especially common with mp3s and seems less so with wav files.

shrugs
__________________
https://www.kdubbproductions.com/
https://www.youtube.com/channel/UCpC...2dGA3qUWBKrXQQ
i7 8700k,4.9Ghz,Win10,Reaper 6,Motu 828es, Cranborne ADAT500
poetnprophet is offline   Reply With Quote
Old 03-19-2019, 10:55 AM   #16
karbomusic
Human being with feelings
 
karbomusic's Avatar
 
Join Date: May 2009
Posts: 29,260
Default

TP meters are somewhat educated guesses of what might happen during conversion. If one uses waveforms which typically don't exist in music, a TP can be several dB above digital peaks but since that doesn't happen in music so much, not a big deal - meaning we don't need to sit at -3dB to escape a 3dB TP in music material. I still think TP meters should be considered ballpark TP rather than an exact value...

https://techblog.izotope.com/2015/08...eak-detection/

Quote:
How can we measure the quality of a true peak meter?

While most true peak meters follow the same basic algorithm as the one described in BS.1770, they can vary significantly in two dimensions: the quality of the upsampling algorithm, and also in the ratio of upsampling. BS.1770 includes a description of a simple upsampling algorithm, but many true peak meters actually perform more accurate upsampling than required by the specification. Also, many meters upsample by more than the required four times. This means that true peak meters can vary significantly in the accuracy of their output.
That said, not sure it should ever be lower than the sample peak though.
__________________
Music is what feelings sound like.
karbomusic is offline   Reply With Quote
Old 03-19-2019, 06:10 PM   #17
Tesgin
Human being with feelings
 
Join Date: Mar 2011
Posts: 222
Default

Quote:
Originally Posted by xpander View Post
One thing to remember is that limiters are not equal when it comes to True Peak settings/readings. Here's some food for thought:
https://www.saintpid.se/en/isp-true-peak-limiters-test/

I loaded up your example project, but since I'm missing almost all of the FX you have in there, I don't get the real conditions for a test.

You have Sonarworks Reference 4 after the hard limiter, before the Youlean Meter. Any difference if you bypass that one?
Wow. Great article. That confirms my suspicion that part of the problem is that my L1 actually is indeed having problems with ISPs. Interesting. I'm still puzzled as to why REAPER's digital peak is actually higher than the true peak measured by dpmeter3 or YouLean. It would have been interesting to have seen them run similar tests using YouLean/dpmeter3 as plugins measuring true peak "live" before rendering.

Thanks for loading the project. Yeah, mainly Waves stuff.

And yes, I have a free demo of Ref 4, good until fall. No, whether bypassing or not, same results.
Tesgin is offline   Reply With Quote
Old 03-19-2019, 06:11 PM   #18
Tesgin
Human being with feelings
 
Join Date: Mar 2011
Posts: 222
Default

Quote:
Originally Posted by poetnprophet View Post
This is something I have noticed in cubase, sonar/cakewalk, and reaper. I have just gotten used to the fact I need to lower the input/clip gain or track gain whenever I import anything....it's the first thing I check. It is especially common with mp3s and seems less so with wav files.

shrugs
Interesting. Seems consistent with what I found. Disappointing. I was hoping the metering would be more accurate than that.
Tesgin is offline   Reply With Quote
Old 03-19-2019, 06:25 PM   #19
Tesgin
Human being with feelings
 
Join Date: Mar 2011
Posts: 222
Default

Quote:
Originally Posted by karbomusic View Post
TP meters are somewhat educated guesses of what might happen during conversion. If one uses waveforms which typically don't exist in music, a TP can be several dB above digital peaks but since that doesn't happen in music so much, not a big deal - meaning we don't need to sit at -3dB to escape a 3dB TP in music material. I still think TP meters should be considered ballpark TP rather than an exact value...

https://techblog.izotope.com/2015/08...eak-detection/



That said, not sure it should ever be lower than the sample peak though.
Excellent article. I can follow the conclusions and the explanations, but when they start breaking it down into advanced mathematical formulas, I feel like Charlie Brown when the adults are talking ("wah-wah-wah-wah-wah ... wah-wah-wah). .

I have so much to learn.

And yes, I agree: I'm getting it that the metering is not as precise as I was assuming it to be, but for the life of me I can't figure out why the TP estimates from YouLean would actually be lower than REAPER's digital meters.
Tesgin is offline   Reply With Quote
Old 03-19-2019, 10:57 PM   #20
clepsydrae
Human being with feelings
 
clepsydrae's Avatar
 
Join Date: Nov 2011
Posts: 3,409
Default

Quote:
Originally Posted by poetnprophet View Post
I have just gotten used to the fact I need to lower the input/clip gain or track gain whenever I import anything....it's the first thing I check. It is especially common with mp3s and seems less so with wav files.
If you have an example of uncompressed audio that goes over 0dBFS on import, please post the snippet that does so.

For MP3s (and other compressed formats) it is normal to get >0dBFS with loud files. It's a side effect of the type of compression used. Short version: converting to a lossy format changes the waveform, and sometimes those changes result in instantaneous levels that are slightly "louder". (This doesn't mean the file in general is "louder" in any practical sense.)

Quote:
Originally Posted by Tesgin View Post
And yes, I agree: I'm getting it that the metering is not as precise as I was assuming it to be, but for the life of me I can't figure out why the TP estimates from YouLean would actually be lower than REAPER's digital meters.
Yes, let's get to the bottom of this. Kudos for posting a project, but what we need is something different: a minimal project just complex enough to demonstrate the issue. [edit: i recall now that I said above "post the entire project somewhere"; I haven't looked at it but it's clear from what others said about it that we need to get a reduced form to check out]

I suggest opening your project as it is, playing through until you find a place that triggers the peak issue you see, making a time selection around that point (1 second? 5 seconds?), and confirming that it works as you expect when you play just that region.

Then disable the youlean plugin (if I understand you, it's the last thing on your master bus?) and render the master track just within the time selection to a 32bit or 64bit float WAV file.

Make a new project, import the rendered audio, put youlean on the master track, and see if you can duplicate the issue. If you can, post that project (or even just the rendered audio file). Make sense? Then we can try to see what you're seeing without needing all the other plugins.

Last edited by clepsydrae; 03-19-2019 at 11:32 PM.
clepsydrae is offline   Reply With Quote
Old 03-19-2019, 11:04 PM   #21
clepsydrae
Human being with feelings
 
clepsydrae's Avatar
 
Join Date: Nov 2011
Posts: 3,409
Default

Also: I'm surprised that there is much difference in ISP measurement... that saintpid.se link says they saw .1-.2dB variance among high quality ISP measurements, which isn't a lot, but is a lot more than I would have expected.

I don't think it's that ISP detection is a fuzzy art, I guess it probably just comes down to how much you upsample (the quality of upsampling seems less likely to explain this?): I can easily see 4X not giving enough resolution to catch the peakest peak of the peak, and if you devote more CPU you're going to find that tip top more often. Stands to reason, I just assumed the top companies would all be upsampling a ton.
clepsydrae is offline   Reply With Quote
Old 03-19-2019, 11:16 PM   #22
bezusheist
Human being with feelings
 
bezusheist's Avatar
 
Join Date: Nov 2010
Location: Mullet
Posts: 829
Default

You must have moved your master fader.
__________________
I like turtles
bezusheist is offline   Reply With Quote
Old 03-19-2019, 11:29 PM   #23
clepsydrae
Human being with feelings
 
clepsydrae's Avatar
 
Join Date: Nov 2011
Posts: 3,409
Default

Quote:
Originally Posted by bezusheist View Post
You must have moved your master fader.
See post #13 -- he says it's at 0.00...
clepsydrae is offline   Reply With Quote
Old 03-20-2019, 12:14 AM   #24
bezusheist
Human being with feelings
 
bezusheist's Avatar
 
Join Date: Nov 2010
Location: Mullet
Posts: 829
Default

but that is the only logical answer.

*did you check the project file ? i did...

(OP, master track volume automation is POST efx...)
__________________
I like turtles

Last edited by bezusheist; 03-20-2019 at 12:28 AM.
bezusheist is offline   Reply With Quote
Old 03-20-2019, 09:04 AM   #25
clepsydrae
Human being with feelings
 
clepsydrae's Avatar
 
Join Date: Nov 2011
Posts: 3,409
Default

Quote:
Originally Posted by bezusheist View Post
but that is the only logical answer.
...well, there could have been bugs, or other user error.

Quote:
*did you check the project file ? i did...
Nope!

Quote:
(OP, master track volume automation is POST efx...)
Bingo -- OP: sounds like we have the solution. (I still haven't checked the project personally, but assuming this is the case, that will explain it. Edit: yep, confirmed.)

Last edited by clepsydrae; 03-20-2019 at 10:05 AM.
clepsydrae is offline   Reply With Quote
Old 03-20-2019, 09:07 AM   #26
brainwreck
Human being with feelings
 
Join Date: Jul 2006
Posts: 20,859
Default

Quote:
Originally Posted by clepsydrae View Post
They can happen via oversampling (in software) as well. (Edit: although whether or not that's an issue depends on what happens to the data; often it's going to be floating point and not matter.)
In what cases would it not be floating point?
__________________
It's time to take a stand against the synthesizer.
brainwreck is offline   Reply With Quote
Old 03-20-2019, 09:24 AM   #27
clepsydrae
Human being with feelings
 
clepsydrae's Avatar
 
Join Date: Nov 2011
Posts: 3,409
Default

Quote:
Originally Posted by brainwreck View Post
In what cases would it not be floating point?
I can't think of any (reaper users that set track mixing depth to an integer format, maybe?, for whatever reason? edit: after the ISP was created via upsampling and DSP was done that generated an over as a result which persisted after downsampling); one could also imagine a scenario where a plugin oversamples (to floating point) but isn't programmed to be aware that the signal might then go over 0 dBFS, which could be somehow relevant.

I don't think this is something a user needs to worry about, I was just pedantically reacting to your post that made it sound like ISPs don't exist except when doing ADC. (I think you added that asterisk later; otherwise, sorry for missing it.) Just pointing out that they do exist and are handled in general DSP as well, when upsampling occurs.

Last edited by clepsydrae; 03-20-2019 at 09:32 AM.
clepsydrae is offline   Reply With Quote
Old 03-20-2019, 09:33 AM   #28
brainwreck
Human being with feelings
 
Join Date: Jul 2006
Posts: 20,859
Default

Quote:
Originally Posted by clepsydrae View Post
I can't think of any (reaper users that set track mixing depth to an integer format, maybe?, for whatever reason?); one could also imagine a scenario where a plugin oversamples (to floating point) but isn't programmed to be aware that the signal might then go over 0 dBFS, which could be somehow relevant.

I don't think this is something a user needs to worry about, I was just pedantically reacting to your post that made it sound like ISPs don't exist except when doing ADC. (I think you added that asterisk later; otherwise, sorry for missing it.) Just pointing out that they do exist and are handled in general DSP as well, when upsampling occurs.
I think it wasn't pedantic to point it out. My statement was an over-generalization. The more we know, the less we are going about things like it's all voodoo.
__________________
It's time to take a stand against the synthesizer.
brainwreck is offline   Reply With Quote
Old 03-20-2019, 09:38 AM   #29
clepsydrae
Human being with feelings
 
clepsydrae's Avatar
 
Join Date: Nov 2011
Posts: 3,409
Default

<ignore> definitely groggy this morning...
clepsydrae is offline   Reply With Quote
Old 03-20-2019, 09:40 AM   #30
TBProAudio
Human being with feelings
 
TBProAudio's Avatar
 
Join Date: May 2014
Location: Germany
Posts: 643
Default

Quote:
Originally Posted by karbomusic View Post
TP meters are somewhat educated guesses of what might happen during conversion.

Hmm, TruePeak detection is clearly defined here: https://www.itu.int/dms_pubrec/itu-r...0-I!!PDF-E.pdf
Just to avoid guessing:-)

The relevance in the analogue world is another story!
__________________
www.tbproaudio.de
TBProAudio is offline   Reply With Quote
Old 03-20-2019, 09:44 AM   #31
karbomusic
Human being with feelings
 
karbomusic's Avatar
 
Join Date: May 2009
Posts: 29,260
Default

Quote:
Originally Posted by TBProAudio View Post
Hmm, TruePeak detection is clearly defined here: https://www.itu.int/dms_pubrec/itu-r...0-I!!PDF-E.pdf
Just to avoid guessing:-)

The relevance in the analogue world is another story!
Isn't the analog world aka reconstruction the entire point of TP?
__________________
Music is what feelings sound like.

Last edited by karbomusic; 03-20-2019 at 09:49 AM.
karbomusic is offline   Reply With Quote
Old 03-20-2019, 09:44 AM   #32
clepsydrae
Human being with feelings
 
clepsydrae's Avatar
 
Join Date: Nov 2011
Posts: 3,409
Default

Quote:
Originally Posted by TBProAudio View Post
Hmm, TruePeak detection is clearly defined here: https://www.itu.int/dms_pubrec/itu-r...0-I!!PDF-E.pdf
Just to avoid guessing:-)

The relevance in the analogue world is another story!
I think his point is just that the implications of the determined TP value are a bit vague, given that ADC varies in its implementation. (And the determination of the TP depends on the quality of the upsampling used, the lower bounds of which is well defined, but as one of those links above points out, implementations can exceed the requirement at their discretion.)
clepsydrae is offline   Reply With Quote
Old 03-20-2019, 09:46 AM   #33
TBProAudio
Human being with feelings
 
TBProAudio's Avatar
 
Join Date: May 2014
Location: Germany
Posts: 643
Default

Quote:
Originally Posted by clepsydrae View Post
Also: I'm surprised that there is much difference in ISP measurement... that saintpid.se link says they saw .1-.2dB variance among high quality ISP measurements, which isn't a lot, but is a lot more than I would have expected.

I don't think it's that ISP detection is a fuzzy art, I guess it probably just comes down to how much you upsample (the quality of upsampling seems less likely to explain this?): I can easily see 4X not giving enough resolution to catch the peakest peak of the peak, and if you devote more CPU you're going to find that tip top more often. Stands to reason, I just assumed the top companies would all be upsampling a ton.

Well, TP measurement is not only OSx4 (at least according EBU standard).
So the quality of the OS-filter determines the quality of TP detection.
__________________
www.tbproaudio.de
TBProAudio is offline   Reply With Quote
Old 03-20-2019, 09:52 AM   #34
TBProAudio
Human being with feelings
 
TBProAudio's Avatar
 
Join Date: May 2014
Location: Germany
Posts: 643
Default

Quote:
Originally Posted by karbomusic View Post
Isn't the analog world the entire point of TP?

Right, but at this point you can only reach your target indirectly:


"If you use the standard measurement, you are on the save side in the analogue world."


For some it is an academic discussion, for some it is business critical:-)
__________________
www.tbproaudio.de
TBProAudio is offline   Reply With Quote
Old 03-20-2019, 09:55 AM   #35
karbomusic
Human being with feelings
 
karbomusic's Avatar
 
Join Date: May 2009
Posts: 29,260
Default

Quote:
Originally Posted by clepsydrae View Post
I think his point is just that the implications of the determined TP value are a bit vague, given that ADC varies in its implementation. (And the determination of the TP depends on the quality of the upsampling used, the lower bounds of which is well defined, but as one of those links above points out, implementations can exceed the requirement at their discretion.)
The Izotope article explains it best I think.

I know I can load up several TP meters and get different values from most of them (at least that was true a year ago). From a caveman perspective, I don't care why, I care that they aren't generally consistent and anyone who say's "but mine is accurate" needs to demonstrate it somehow beyond saying they followed the spec (see article)? Because if so cut and dried, you'd think everyone's would match more reasonably than they do.
__________________
Music is what feelings sound like.
karbomusic is offline   Reply With Quote
Old 03-20-2019, 09:59 AM   #36
TBProAudio
Human being with feelings
 
TBProAudio's Avatar
 
Join Date: May 2014
Location: Germany
Posts: 643
Default

Quote:
Originally Posted by clepsydrae View Post
I think his point is just that the implications of the determined TP value are a bit vague, given that ADC varies in its implementation.

Correct: The TP value does not answer how well the signal is translated by the DAC.
__________________
www.tbproaudio.de
TBProAudio is offline   Reply With Quote
Old 03-20-2019, 10:00 AM   #37
clepsydrae
Human being with feelings
 
clepsydrae's Avatar
 
Join Date: Nov 2011
Posts: 3,409
Default

Interesting tidbits from the ITU pdf:

"Higher sampling rates and over-sampling ratios are preferred" (as several have mentioned already, just affirming it)

"Incoming signals that are at higher sampling rates require proportionately less over-sampling (e.g. for an incoming signal at 96 kHz sample rate a 2X over-sampling would be sufficient.)" (4X is the amount for e.g. 48k) Thought this was interesting -- I would have expected them to keep 4X across the board. Not that I care personally, but those that are attached to higher sample rates might wonder why ISPs between their 96k samples are less important than those at 48k. :-)

Also an interesting table on page 22 that shows the maximum "underread" of TP at various sample rates.

E.g. at 4X oversampling the absolute worst case TP would be 0.688 dB louder than the TP meter indicated. At 32X oversampling it would only be 0.01 dB louder. The ITU spec only requires 4X, so this is kind of a big difference.
clepsydrae is offline   Reply With Quote
Old 03-20-2019, 10:01 AM   #38
karbomusic
Human being with feelings
 
karbomusic's Avatar
 
Join Date: May 2009
Posts: 29,260
Default

Quote:
Originally Posted by TBProAudio View Post
Right, but at this point you can only reach your target indirectly:
Quote:
Originally Posted by TBProAudio View Post
Correct: The TP value does not answer how well the signal is translated by the DAC.
I think that's precisely why I call it a guess. I think for most end-users, it sort of doesn't matter why, just that it is.
__________________
Music is what feelings sound like.
karbomusic is offline   Reply With Quote
Old 03-20-2019, 10:31 AM   #39
TBProAudio
Human being with feelings
 
TBProAudio's Avatar
 
Join Date: May 2014
Location: Germany
Posts: 643
Default

Quote:
Originally Posted by Tesgin View Post
Here's the link to the wav file:

https://we.tl/t-j2fRvOy2uE

Hmm, dpMeter3: Peak: 0.0, TP +0.1
Do I miss something?
__________________
www.tbproaudio.de
TBProAudio is offline   Reply With Quote
Old 03-20-2019, 10:34 AM   #40
clepsydrae
Human being with feelings
 
clepsydrae's Avatar
 
Join Date: Nov 2011
Posts: 3,409
Default

Jesus my reading comprehension has been terrible on this thread; OP, thanks for posting that WAV file and sorry to miss that the first time.

Quote:
Originally Posted by TBProAudio View Post
Hmm, dpMeter3: Peak: 0.0, TP +0.1
Do I miss something?
OP had volume automation on the master track. Hence the WAV file doesn't exhibit the issue, as OP saw themselves, but the original project did.
clepsydrae 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 06:23 PM.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.