clipping indicator options

[the brok]

hi all,

is there a way to editthe clipping indicator, so that it _only_ lits up, if a signal is overshooting? i have a normalized signal, perfectly levelled to 0db, and i would like to not see the clipping indicator clippng, as the signal is exactly 0db max .... but the clipping indicator lits up when the signal is 0db ... so is there a way to set up the clipping indicator,that it only lits up when the signal actually exeeds 0db?

[labyrinth]

This would not be wise, because many devices clip at exactly 0dB.

[the brok]

Originally Posted by labyrinth

This would not be wise, because many devices clip at exactly 0dB.

i don`t care. my rme interface doesn`t clip at 0db, but anything over 0db. and that`s the way it should be. hardclipping occours not at 0db, but over 0db, that`s the technical specification. if you normalize a wave file in i.e.wavelab, it mustn`t be indicated clipping when played back in reaper, as it simply doesnt clip.
also, i`m talking about the individual channels, where you have lot`s of headroom over 0db because of 32bit processing, before actually hardclipping occours ... and these channels also show clipping at 0db, rather than over 0db ... so i`d like t be able to change that ... those who don`t, just leave it the way it is ...

[labyrinth]

I'm good with an option, maybe to optimize to gear.

[martygras]

Could you change the master meter range shown?
Like-6 to +2

[the brok]

labyrinth, yeah, that`s actually a good idea ... even more, there could be some sort of file to download from the reaper site, which holds the settings for the meters, etc for a speciffic hardware, so that reaper is set up accordingly when loading that file ... but that would require the cockos team to check out every hardware they come across ... and a user upload ... well, that depends on the user, but that could be an by the time increasing database ...
interresting ...

[labyrinth]

We would basically need to have a few audio file of white noise at say -2dB thru +2dB and depending on when your meters clipped on your interfaces meters (software, not hardware) show clipping this is when/where you adjust to.

I must point out though that regardless of what level your device clips at dBFS is exact...0 is the stop...after that your sign waves turn to square waves or basically Shit!

[the brok]

Originally Posted by labyrinth

We would basically need to have a few audio file of white noise at say -2dB thru +2dB and depending on when your meters clipped on your interfaces meters (software, not hardware) show clipping this is when/where you adjust to.

exactly ...

Originally Posted by labyrinth

I must point out though that regardless of what level your device clips at dBFS is exact...0 is the stop...after that your sign waves turn to square waves or basically Shit!

i did not say anything else. what that techically means is, that whenever a signal reaches 0db, there`s zero headroom anymore. at exactly 0db the signal just fills out the complete possible range, but the signal is not clipping, thus the meters in reaper should not indicate a clipping when the signal is not exeeding 0db. everything above clips (i.e. 0.000001db). that`s when the led should light up. so the way it is in reaper is just wrong.

[labyrinth]

Originally Posted by the brok

......thus the meters in reaper should not indicate a clipping when the signal is not exeeding 0db. everything above clips (i.e. 0.000001db). that`s when the led should light up. so the way it is in reaper is just wrong.

I agree that it should let you move to 0, but not above.

[the brok]

exactly ... so, either they give us a possibility to change the behave of the clipping indicators, or they hardcode it as a fix in an update ... now that i think of it:
was it always like that? or is is since reaper 2.5 in the end?? hmmm .... if so, i`ll post that to the bugreports ... but i need someone to confirmthat it was the right way in reaper pre-v2.5 ...

[the brok]

buuuuuump!!

doesn`t that missbehave annoy you?? come one, guys, it needs to be changed!
what`s a clipping indicator worth if it doesn`t indicate the clipping the correct way???

[IIRs]

There is no "correct" way to determine clipping. Some meters show an over after a single full scale sample, others will require a number of consecutive full scale samples (eg: 3) to register an over. Others still will oversample the audio signal in an attempt to recognize inter-sample peaks that can occur in oversampling DACs (this is why it is considered good practise not to normalise or limit to 0dB but to leave approx 0.5db headroom.)

The best advice I can give is: don't peak your signals to full scale. Leave a tiny bit of headroom so that you can be sure of a clean analogue output signal from any DAC. This will also cure your Reaper meters 'problem'.

[the brok]

Originally Posted by IIRs

There is no "correct" way to determine clipping. Some meters show an over after a single full scale sample, others will require a number of consecutive full scale samples (eg: 3) to register an over. Others still will oversample the audio signal in an attempt to recognize inter-sample peaks that can occur in oversampling DACs (this is why it is considered good practise not to normalise or limit to 0dB but to leave approx 0.5db headroom.)

The best advice I can give is: don't peak your signals to full scale. Leave a tiny bit of headroom so that you can be sure of a clean analogue output signal from any DAC. This will also cure your Reaper meters 'problem'.

thanks for that advice, i know that. but:

1. clipping indicators should indicate clipping (commonly over, not at 0db is clipping). otherwise they`re wrong implemented.
the fact that some do like you say above doesn`t change that as a) reaper itself doesn`t oversample the signal chain, and b) even if we have a lot of headroom on 32/64 bit systems on individual channels, the channel metering is scaled with a reason ... to be balanced out over all tracks including busses and masters (what good is a metering for if not properly used?) ...
2. the problem is, that a lot of plugins, like above stated, have a brickwall at the end of the signal chain, fixed at 0db. the limiter is driven by its input, and often that just sounds good ... now, when you have a fader at 0db, and you drive the brickwall limiter, the channel metering indicateds clipping, even if there is no clipping, because you`re at, not over 0db ...
good sequencer programs, beeing aware of that problem, give you a preference where you can set the point where the clipping indicator lits up by yourself ... this is what i would like to have ...

[the brok]

Originally Posted by IIRs

(this is why it is considered good practise not to normalise or limit to 0dB but to leave approx 0.5db headroom.)

i do that, always ... at the masterchannel ... but anyway, that is not my point ...

[jonespnice]

Originally Posted by the brok

thanks for that advice, i know that. but:

1. clipping indicators should indicate clipping (commonly over, not at 0db is clipping). otherwise they`re wrong implemented.
the fact that some do like you say above doesn`t change that as a) reaper itself doesn`t oversample the signal chain, and b) even if we have a lot of headroom on 32/64 bit systems on individual channels, the channel metering is scaled with a reason ... to be balanced out over all tracks including busses and masters (what good is a metering for if not properly used?) ...
2. the problem is, that a lot of plugins, like above stated, have a brickwall at the end of the signal chain, fixed at 0db. the limiter is driven by its input, and often that just sounds good ... now, when you have a fader at 0db, and you drive the brickwall limiter, the channel metering indicateds clipping, even if there is no clipping, because you`re at, not over 0db ...
good sequencer programs, beeing aware of that problem, give you a preference where you can set the point where the clipping indicator lits up by yourself ... this is what i would like to have ...

No disrespect but I would like to emphasize the importance of not mixing/ recording really close to the red (especially of you're using 24 bit or higher), from everything that I've read the accumulation of tracks that are in the red will cause your song to sound more digital and sterile. Eventhough there are saturation plugins they usually will have a drive parameter where you can really squash it.
It's cool to have more option (i.e. being able to adjust the clipping indicators) and I'm all for it but just wanted drop that tid bit of information at you.
BTW, I think pro tools has the most awful metering display/ indicator (I haven't seen pro tools 8 yet).

[the brok]

no problem ... i don`t take it as offence at all ...
well, you know, i`m mixing since close to 25 years now, i know all about that ... fact is: yes, stay away from the clipping/red, especially when recording, or outputting to the analog domain again ... fact is too, that i actually am, and that`s what i was about: i am not clipping, yet the clipping indicator tells me i am ... it`s true that real digital overs willdo strange things not only to the track that is actually clipped, but to the ones it is going to be summed to ... but that wasn`t my point at all .. again, my point was that the clipping indicator in reaper indicated a clipping even there is none. this has to be changed, either user-adjustable or fixed. but it has to be changed, as a clipping indicator that shows clipping when there is none, is simply useless ...

[IIRs]

The reason digital meters have over lights that respond to samples at 0dB is because usually there is no higher level available. If you record a signal too hot to 16 bit DAT tape, then play it back, there will be no 'overs'.. there will only be strings of full scale 0dB samples to indicate a clipped peak.

In Reaper its a different situation, because the floating point architecture does allow overs without clipping them to full scale... so you could argue that the red clip light should never come on for channels but only on the master.

Either way, there are two ways you can remove the problem without needing to change Reaper: either ignore the over lights, or drop your channel fader to -0.1 dB

[the brok]

Originally Posted by IIRs

The reason digital meters have over lights that respond to samples at 0dB is because usually there is no higher level available. If you record a signal too hot to 16 bit DAT tape, then play it back, there will be no 'overs'.. there will only be strings of full scale 0dB samples to indicate a clipped peak.

i know this. it`s still wrong, as, again, at 0db is not clipping, but anything over 0db is. the over indicators (as the name says) are there to reliably tell you when you are over 0db, as this can cause digital clipping. that`s when they should light up. if you are at 0db, you`re not clipping digitally, your audio isn`t damaged then, so there shouldn`t be an over indicator lighting up at all.

Originally Posted by IIRs

In Reaper its a different situation, because the floating point architecture does allow overs without clipping them to full scale... so you could argue that the red clip light should never come on for channels but only on the master.

that`s another situation. when going over 0db internally on an 32 bit system (i.e.win xp), you have a lot more headroom anyway (not to speak of the fact that the bottleneck is the output to the converters in all cases anyway). it´s not reapers fault, but the os`s fault (well, reaper could work in 16 bit instead of 32bit, but no modern app on xp is, there`s just no reason) ... still, that`s not the point:
here the point is: if one ignores the metering on the tracks, pushing up all tracks, he can`t rely on the metering at all anymore, so he puts the use of meterings out of it`s sense ...the meterings are there to follow them, therefore, even if you could, you shouldn`t have over indications on any track, as, if you level correctly, the hardware bus metering (busses and outputs) are showing the correct values, and at those positions in the signal flow the over indicators are max. important, as there you mustn`t have overs at the converters ...

Originally Posted by IIRs

Either way, there are two ways you can remove the problem without needing to change Reaper: either ignore the over lights, or drop your channel fader to -0.1 dB

i know that, there`s even more ways to work around this issue ... come on ... i want reaper to behave right, not me working around the failures of it ... it`s reapers failure. it`s not correct and should be corrected. i mean, seriously, reaper is claimed to be one of the most accurate audio apps out there, and then this??
and it`s still not corrected in v2.54, even though it would be just maximum easy to fix this bug ... this is annoying ...

[mahasandi]

Originally Posted by the brok

hi all,

is there a way to editthe clipping indicator, so that it _only_ lits up, if a signal is overshooting? i have a normalized signal, perfectly levelled to 0db, and i would like to not see the clipping indicator clippng, as the signal is exactly 0db max .... but the clipping indicator lits up when the signal is 0db ... so is there a way to set up the clipping indicator,that it only lits up when the signal actually exeeds 0db?

Its a good point i would like to see clipping indicators ( 0.0db is not clipping its 0 dbfs fullscale)
at all outputs. (harware outputs have no clip indacators for those of us not using the master
buss.
and the floating point nature of reaper is not the point;

its when we send signals out to dac/analog they should show what will happen in the "real world" of voltages and non floating point dacs!
inter sample peaks can be higher then .5 db as well.

bottom line clip indicators on all points in the signal chain that leave the floating point
mixer should be available.

[Dstruct]

It's all fine in Reaper. Nothing needs to be changed IMO.

Clipping indicator comes up at 1 0dBFS sample. What do you want more? For internal mixing you can ignore it. When going to external hardware output it's useful.

[Evan]

Originally Posted by IIRs

so you could argue that the red clip light should never come on for channels but only on the master.

Very interesting suggestion. Is there such an option? Should there be? It might make it easier to take everything 'too loud' for those who are not experienced with mixing. On the other hand it's a more accurate and up-to-date digital-age approach.

[Dstruct]

Originally Posted by Evan

Should there be?

Not really. Why?

1. Tracks can be sent directly to hardware outputs -> here we need the clipping indicator

2. You might want to render the output of a track to a fixed point format -> here we need the clipping indicator too


IMO there's no need to change anything.

[the brok]

Originally Posted by Dstruct

It's all fine in Reaper. Nothing needs to be changed IMO.

Clipping indicator comes up at 1 0dBFS sample. What do you want more? For internal mixing you can ignore it. When going to external hardware output it's useful.

some of you don`t really seem to understand:
it`s important to have the over indicators on all tracks, as they do give you a clue, they assist you for having a balanced level-average, before summing on busses/masters ... of course i could ignore the over indicators on tracks internally and raise the signal way over 0db as we have plenty of internal headroom, but a)i end up having to lower the bus/masterfader all the time in order to not clip at the converters, ending up to have a tiny range on the master/bus strip (bad for automation, if there is), and b) i then have no real overview on the channels, the metering doesn`t tell me the correct values anymore ... we have the metering and over indicators on tracks aswell for a reason ...
understand ... f.e., when mixing the loudest signal in the mix to -6db, you mostly end up with a level on the master where you don`t have to lower the masterfader too much (depending on the mix)... all levels stay within the ideal visual range on all meters ... the visual sweetspot so to say ... get what i mean? you can then trust the meters on all channels ... and that`s the ideal way ...

however, get thisone already, too:
the clipping indicators in reaper are wrong behaving. they show clipping at 0db (where no clipping occours), instead of over 0db. this has to be changed. it`s plain wrong.
and forget inter sample peaks, has nothing to do with this issue ... if an inter sample peak occours, you simply have a different problem (which has to be overcome of course) ... in the end, it either disturbs you, if it`s noticable audibly or if it`s killing your level because of drastic level differences caused by these peaks, which don`t allow you to use the peaklevel of the original signal, or they`re so tiny to be ignored ... either way, it has nothing to do with the over indicators, exept the fact the similarity it has with all signals:
up to at 0db = no overs (ideal range on the meters), over 0db = overs (not ideal meters, aswell as clipping, if on any mixerstrip that is routed to a converter directly).

[Dstruct]

Originally Posted by the brok

the clipping indicators in reaper are wrong behaving. they show clipping at 0db (where no clipping occours), instead of over 0db. this has to be changed.

It doesn't need to be changed.

1. If you peak at -0.1dBFS you practically don't loose any loudness or resolution.

2. Say you render a song which peaks at 0dBFS. Now you give this song to another person. Now this person can't tell for sure if the render was clipping or not. If the mix peaks at -0.1dB you CAN say that the render was not clipping.

Originally Posted by the brok

my rme interface doesn`t clip at 0db, but anything over 0db.

Wrong. You're talking about TotalMix? Well by default it's set to show clipping if there are 3 samples at 0dBFS in series. You can set it to "1 sample" to get the same reading as Reaper (which is a bit safer).

[the brok]

sorry dstruct,
i don`t have the time to reply detailed.
however, what is it you don`t understand in my latest post???
all you say is/might be right, it doesn`t change a thing in what i say and what was my point ...
use a brickwall limiter which exactly limits the signal to 0dbfs.
2 facts are showing up:

1. the output signal of the limiter is not clipping.
2. reaper indicates clipping, which is wrong, as the signal is not.

conclusion: it`s wrong and has to be changed.
what`s so hard to understand?

[Dstruct]

I understand what your're trying to say. Of course the current mode is technically not really correct. But it makes sense the way it is (as explained). All other applications do it the same way (0dBFS=clip).

[zildjohn01]

There's no reason to change REAPER's behavior. But if this is a huge deal for you, I'll write you a JS plugin. Stick this at the end of the tracks where metering is important. Take it off before mixdown.

@init
  // 32 bits
  //clip_to = 0.99999999976716935634613037109375;
  // 64 bits
  clip_to = 0.99999999999999999994578989137572;

@sample
  spl0 == 1 ? spl0 = clip_to : spl0 == -1 ? spl0 = -clip_to;
  spl1 == 1 ? spl1 = clip_to : spl1 == -1 ? spl1 = -clip_to;

Haven't tested it myself. Does it work?

[the brok]

Originally Posted by Dstruct

I understand what your're trying to say. Of course the current mode is technically not really correct. But it makes sense the way it is (as explained).

hmm ... though that way you`ve explained makes sense, i don`t really see the relevance of that... i mean, you rendered the file you give someone, so you know it`s not clipping ... it`s not necessary for the other person to unviel if your rendering is clipping or not, as he cannot change it anyway afterwards, on the rendered stereofile ...
it`s the "renderer" that has to take care of that, in this case you ... right?

Originally Posted by Dstruct

All other applications do it the same way (0dBFS=clip).

not at all: samplitude, logic, and some others are showing clippings at +0,0001dbfs, not at 0dbfs (in samplitude you even have an option to tell the metering when to show overs, this is what i requested here).

however, you can check yourself that the over indication in reaper is wrong:

use george yhoung`s w1 limiter, a very precise waves l1 clone, freeware (mentioning thatone so that you can check yourself). set the ceiling to 0db, and lower the threshold so that the signal should hit the threshold in whatever ammount. the output signal is not clipping, yet reapers clipping indicator lights up. this is plain wrong. and it doesn`t happen in sam10, not in logic 5.5.1, not in ext 1.4.1 or xt2x (not that those are a reference), not in wavelab. in all these apps the clipping indication doesn`t light up in the above scenario, just the way it should be ...

reapers implementation is wrong, and has to at least give an option to set it up accordingly.

[the brok]

Originally Posted by zildjohn01

There's no reason to change REAPER's behavior. But if this is a huge deal for you, I'll write you a JS plugin. Stick this at the end of the tracks where metering is important. Take it off before mixdown.

@init
  // 32 bits
  //clip_to = 0.99999999976716935634613037109375;
  // 64 bits
  clip_to = 0.99999999999999999994578989137572;

@sample
  spl0 == 1 ? spl0 = clip_to : spl0 == -1 ? spl0 = -clip_to;
  spl1 == 1 ? spl1 = clip_to : spl1 == -1 ? spl1 = -clip_to;

Haven't tested it myself. Does it work?

nice ... what is it? does it lower the level so that reaper doesn`t show clipping, or does it change the clipping-indication border to above 0dbfs?
i`m not familar with implementation of js plugins ... what do i have to do with this code exactly?
thanks for trying to help!

[Dstruct]

Originally Posted by the brok

not at all: samplitude, logic, and some others are showing clippings at +0,0001dbfs, not at 0dbfs (in samplitude you even have an option to tell the metering when to show overs, this is what i requested here).

Why do they show clipping at +0.0001dBFS? Now that's plain wrong. Clipping occurs at any value >0dBFS. +0.00000000000001dBFS = clipping for example.


Can you show a screenshot from Samplitude with this option? I have no clue why someone would need that.

Also all the applications you mentioned might show clipping just if MORE than 1 sample (0dBFS) in series occur. Reaper already shows clipping if just 1 sample reaches 0dBFS.


Also for the audio inputs this "show clipping just for >0dBFS values" can't work at all. Reaper doesn't know values above 0dBFS. So 0dBFS=clip. Period.


The only option Justin could add is

Peak indicator: ___ 0dBFS samples in series



Otherwise a useless topic IMO.

[Fabian]

Originally Posted by the brok

not at all: samplitude, logic, and some others are showing clippings at +0,0001dbfs, not at 0dbfs

So those apps don't show clipping at +0,00001dBFS?

I think here is the heart of Dstructs argument, those apps are equally as wrong as you claim Reaper to be. Clipping indication would always be subject to (if nothing else) the numeric accuracy and so it's meaningless to talk about "above 0 dBFS", how much above? And thus, the only reasonable implementation is to show clipping *at* 0 dBFS, that's how I understand it.

Technically, in the analog real world and by definition you are correct, brok, but in the quantized digital world things are not so simple.

[Evan]

Originally Posted by Dstruct

Not really. Why?

1. Tracks can be sent directly to hardware outputs -> here we need the clipping indicator

2. You might want to render the output of a track to a fixed point format -> here we need the clipping indicator too


IMO there's no need to change anything.

Thank you very much.

[the brok]

Originally Posted by Dstruct

Why do they show clipping at +0.0001dBFS? Now that's plain wrong. Clipping occurs at any value >0dBFS. +0.00000000000001dBFS = clipping for example.

ahhh, sorry, my fault ... stupid me ...
see, i have an se plugin that allowes me to rise the gain in 0.0001db steps ... i found it somewhere on the net ... i quickly used this in rush to see when clipping indication actually occours in logic and ext, etc ... so, i raised the level from exact 0dbfs (no clipping indication) with the finest step in this plug to 0.0001dbfs, and the track showed clipping ...
i was just not thinking ... again, my fault, sorry for this misinterpretation ...
so, to get this corrected:

these apps don`t show cliping indication at 0dbfs but anything over 0dbfs, though i don`t know, how much over 0dbfs ...in good faith these apps are behaving correctly, i assume they show clipping at any value over 0dbfs...


however, with every brickwall limiter i have it`s the same:
they have a ceiling at 0db (be it useradjustable or fixed), and they don`t visually clip in any other app (as they don`t digitally clip in general), but in reaper. so you want to tell me all the plugs are wrong, but reaper is right?
and even if there are many ways to implement a clipping indicator, what good is a clipping indicator that indicates clipping when there is none, which is exactly what reaper does????
all the limiters are useless then, as one can`t see if clipping occours or not in reaper ... what good could this possibly be???

look, up to exact 0dbfs, there`s no clipping, correct? if this is so, why does reaper indicate clipping then, and gives you a false information about the current state?? and why are you guys still telling me that the way reaper behaves is correct???

just do the test with any 0db brickwall limiter in any other host, then in reaper, you`ll most likeley see that reaper shows clipping, whereas all other hosts won`t (and they`re right, as there is no clipping in the signal at exactly 0dbfs) ... now all others are wrong, and reaper is the only host that`s correct?

please forgive me thatone, if i can`t believe it ...
or, please, enlight me and explain why reapers behave (to show clipping when there actually is none) is correct, while other hosts (showing no clipping at 0dbfs, while the signal doesn`t clip anyway) are plain wrong??

[zildjohn01]

Originally Posted by the brok

nice ... what is it? does it lower the level so that reaper doesn`t show clipping, or does it change the clipping-indication border to above 0dbfs?
i`m not familar with implementation of js plugins ... what do i have to do with this code exactly?
thanks for trying to help!

I just threw it together. It lowers any sample at +0dB to 1 bit below +0dB. Effectively reducing your dynamic range by about 0.0000000000000000001% or so, but REAPER will no longer think it's clipping so it will leave the light alone.

To use it, put it in a text file with no extension in "*\REAPER\Effects\Utility". You should notice the pattern in that folder. Then in REAPER add it the same way you add a VST plugin