Is True Peak Limiting a Lie? What You Need to Know!
HTML-код
- Опубликовано: 7 сен 2024
- ⭐ STEM MASTERING MASTERCLASS - www.panoramama...
📧 The BEST newsletter for working audio professionals who strive for more! Subscribe here - www.panoramama...
🎹 Proud of your mix? AWESOME! Get a fresh set of ears on board for mastering and let’s work together! - panoramamaster...
🛠️ Mastering Template for ProTools, Logic and Ableton- www.panoramama...
--
Hello, I'm Nicholas Di Lorenzo, Studio Owner, Mixing and Mastering engineer at Panorama Studios.
I'm an Italian-Australian born and raised in Melbourne. I've been a creative professional for 10 years managing some pretty awesome projects for artists, labels and producers all around the globe.
What motivates and drives me?
My family,
Good food,
Great coffee.
You can find me on many platforms:
Instagram: / panorama_mastering
Facebook: / panoramamastering
Twitter: / panoramamasters
Kit: kit.co/Panoram...
This is absolutely brilliant. Game changing approach. Facts not opinions, I love it. Keep it up man
Thanks, will do!!!
@@panorama_mastering I saw your "a day in the life" video where you break down your schedule. I have no idea where you find the time but this is incredible. I'm doing my PhD and this is definitely research, depth, analysis at that level or higher all while holding it down in the lab. Thank you so much for this!
Appreciate your support and recognition of the detail I'm going to!
This is truely my "creative escape" where I get to tie in my worklife from the studio to feed my curiosities and professional development; all documented here on RUclips;
@@panorama_mastering Ditto...thanks for your work. Saved me pulling my hair out trying to find real info about this topic. What I want to see is a simulation of what the loudspeaker actually does when a "true peak" hits it. My guess is, no real difference.
@@panorama_mastering found your channel a few days ago, brilliant, thank you, cool approch to things, with deep explained topics. We needed that in today junktube ♥️
Refreshing
The amount of valuable info you put out here for free is just mind blowing.
thanks man!
@@panorama_mastering i dont understand what you said at the end if i do use true peak limiting do i need to limit more in order to get a higher volume output as if the true peak limiting was turned off ???
Solid video, very entertaining, I have learnt a lot here and gives a lot of thoughts. Not much advanced nitty gritty mastering content out there, so really appreciate the content! Literally confirmed my thoughts on TP and as to why I may prefer it off.
Would it be possible for you to explore more topics like this? I also appreciate your videos regarding marketing. So maybe you could do some sort of 'how to build your sound engineering business' type video. Would be cool to hear your expertise on how to get more clients if you already have a few and in general how scale your mixing/mastering business!
Thanks again Nicholas, look forward to watching the next one.
Thanks man!
I will certainly be doing more; I currently have 4 videos in production; a lot of the comment sections on the videos feed and fuel new curiosities for me; So the resulting video's tend to be on topics I have knowledge gaps in and need to learn more about!
I've done some fundamental work on "How to Actually get clients for mixing", the business stuff I tend to only explore if there are specific niched topics which peak my interest; only because I try to keep those VERY actionable; so people leave with some direction and not just "business hustle fluff"
Don't forget that oversampling ALWAYS requires filtering, and filtering ALWAYS messes with phase and/or volume. If you set a non-oversampled limiter up to brickwall at, say, -6 dBfs, then there will not be any peaks above -6 dBfs, full stop. (Except potential inter-sample peaks, but the limiter doesn't know that.) If you switch oversampling on in that limiter, the filters used in the downsampling stage will either create pass-band or stop-band ripples, i.e. volume instabilities, looks like a little sine wave going along the spectrum, up and down, up and down.
Filters with low or no pass-band ripples usually have a lot of stop-band ripples, and filters with low or no stop-band ripples usually have a lot of pass-band ripples. In some filters, the ripples can even be adjusted, so you can balance e.g. towards a steeper cutoff slope with more pass-band ripples, or towards a wider cutoff slope but with fewer stop-band ripples.
If a filter creates ripples in the pass-band, i.e. the already brickwall limited signal that goes into the downsampling filter, this means that the previously brickwalled target volume of -6 dBfs no longer applies, since the ripples raise (and lower, but that doesn't matter for peak brickwalling) the volume above that -6 dBfs level at certain spectral ranges. And if the filter is set to create no ripples in the pass-band, then it will create ripplies (i.e. "signal leftovers") in the stop-band, which should be "as completely quiet as possible".
Remaining signal in the stop-band means: things that shouldn't be there when the signal is decimated/sample-reduced again, so effectively - aliasing. Aliasing as in signal that wasn't there before, but now creates resonances and dips, throwing the spectrum and volume out of balance yet again.
So there it is. When oversampling, you have the choice between throwing the spectrum and volume of your already brickwalled signal off again by filter ripples or by aliasing.
Either way, what comes out of a limiter that SHOULD brickwall at -6 dBfs but oversamples the audio path, is NOT going to be brickwalled at -6 dBfs anymore.
Same goes for distortion or clipping processors.
A non-oversampled hard-clipper can reliably cut off any sample (!) value above a certain ceiling, but it can not detect and reduce inter-sample peaks. An oversampled hard-clipper can reliably detect and reduce inter-sample peaks, but the output volume after the down-sampling filter will be entirely random and not hard-clipped.
In the end, you either go for accurately brickwalled volume at the risk of missing inter-sample-peaks, or you go for no inter-sample peaks at the cost of losing brickwalled volume.
There's no inbetween.
Some people run oversampled anti-ISP limiters and clippers first, followed by non-oversampled brickwall limiters and/or clippers, that only shave the odd 1-2 dB or so off here and there. But non-oversampled always means: aliasing. As well as the (highly likely) danger of generating new inter-sample peaks from it.
🤷♀
AMAZING; great addition and commentary on the topic;
If you check out Paul Third's video on Oversampling he makes a great case for the benefit of minimising aliaising being a greater benefit to the sonics in comparison to the filtering needed to make those changes;
For myself; AFTER making this video; I was on a call with Bill Podolak(Product developer for iZotope Ozone); and we looked at how FF Pro-L2 graphed on phase; and can conclude the TP limiter uses the oversampled signal as a SC input to the main signal; so there's a good case for plugins oversampling internally (Especially limiters) the SC signal for greater accuracy in the ballistics of the processing;
Further more; Fabien brought this up with me too when I posed the question of simply UPSAMPLE EVERYTHING;
It's an interesting topic where the intersect of the theory in the DSP and the application have many consideration we must make;
Ultimately; it's one of many topics I hope to further explore!
Audio wise, use whatever you like. But be aware that PCM limiters have all odds pointing against them, except maybe digital nostalgia. Don't let the "TP on/off fallacy" distract you from the good stuff (like having full control over the audio peak, ending up with much less aliasing in the control signal, more "analogue" behaviour, all in all a much more natural sound). :)
One can also resample the sidechain only (pretty standard in today's most popular limiters), or resample the delta signal only (the audio "added" by the limiter). These resampling filters then won't ever affect the original signal, neither bandwidth nor phase.
And even if one would resample the audio itself, brute force, it would be easy to create perfectly flat linear phase low pass filters that have stopband ripple below system noise floor. It's trivial. BTW some filter types (Gaussian) have no ripple at all, which coincidently are optimal for DC applications (a sidechain has a DC output).
A limiter's sidechain is a highly nonlinear element. The rectification, threshold/ratio function, AR filtering, each of these multiply the input bandwidth several times. It means without special control, aliasing inside the sidechain will grow by *insane* amounts. This aliasing in the control signal produces somewhat erratic behaviour. Many learned to enjoy the sharpening effect of such limiting, though.
Old school PCM riding limiters/clipper were short cuts. Crude and cheap approximations to overshoot protection. They didn't ever really limit or control *audio* in the first place, all they care about is the PCM samples (like most audio editor's naive waveform representation). Mathematically, they openly break the sampling theorem.
@@TokyoDawnRecords Resampling the detector path is also a possibility, true, I forgot about that. But even if only the detector signal is oversampled, the calculated gain reduction will be applied to the audio path, which is going to introduce aliasing. And since the detector/sidechain signal goes through a filter, it would throw the detection off, meaning still no true "always below ceiling" brickwalling in the audio path.
I could be totally wrong, but to my understanding a Gauss filter's pass-band response isn't totally flat? So it wouldn't be completely ripple-free, meaning it would cause aliasing and (being a filter) throw off the peak volume, meaning it would also not guarantee "no peaks over the set ceiling". Gauss filters don't seem to be particularly steep, so its center frequency would have to be quite far inside the audible spectrum, creating not just a little "rounded knee" near Nyquist, but likely a rather significant one. I could be wrong about this though, it's been a while since I played around with digital filters.
But all filter-naming aside, in reality most new plugins these days seem to be JUCE based, made by coders who "learned it from that course", so I'm fairly certain they're just using the oversampling methods JUCE provides, which (the last time I tried a few years ago) don't provide fancy (or even optimal) filters but are geared towards simple use. On the road to making money, not everybody's interested in wasting weeks of development time on trial-and-error experiments, in the hope of implementing their own (rather) optimal method for something, if a ready-to-use building block that does a fair job is already available. Sadly.
I understood your last paragraph. Interesting, I never looked at it that way, but you're obviously completely right. But in the discrete/digital domain, the closest thing to "actual audio" like in the continuous/analog domain is an upsampled sample stream at "finer discreteness"? Which would mean all digital non-linear processing would be an approximation, since no "fineness of discreteness" (i.e. amount of up-sampling) could ever reach true continuous-ness?
@@1b2m Oh I agree, audio DSP is quite advanced, and it's not unusual to find all sorts of weird ideas or half baked implementations in the plugin market. It's still in development.
About the Gaussian filter specifically, consider an oversampled sidechain, where the input SRC filter (AC context) is a traditional SINC to capture the full input spectrum as precisely as possible. But the output of the SC (now a DC context) is SRC'ed via Gaussian, to deliver the smoothest possible envelope for that audio input. This is also quite useful in combination with the "max windowing" necessary for lookahead limiting. It's realistically possible to make this act completely flat, because every filter type works in the context that fits it best.
Of course this is highly specialized stuff, but the way educated developers chose and use the full range of filters available to them. Of course, brute force resampling, especially if not done correctly, will quickly show deficiencies in one way or another (e.g. bandwidth systematically reducing input to output, static aliasing, noise, phase distortion, etc), it's easy to mess up!
But all in all, we should value the opportunity - first time in history - to have moderately nonlinear digital devices act like optimal analogue devices. A limiter that is not true peak aware systematically/logically violates the sampling theorem. This is worth being questioned, loudly!
lmfao im going to bed
I would be super grateful for a realistic audio comparison example between no oversampling, oversampling and true peak limiting. Even better would be a blind test with _the great reveal_ afterwards .
These tests were showcasing the function of the limiter; the sound nulls -140dB below 0dBTP Full scale; which would be indistinguishable; otherwise;
The other-thing is; it's easier to set up a null test where I can calculate the positioning of sample points in a scaled sinc-wave over material;
@@panorama_mastering true enough. So there's not much of a "sound" here...
@@drrodopszin Correct, no sound; the ballistics(Att release lookahead) ALL ACT THE SAME; EXCEPT the way it measures/detectrs the peaks
Lol at first I was like " I can't even understand this guy he is crazy", then I play and pause a lot of times and investigate terms on Google, and OMG lol you are a genius.
Sweeeet
Yeah, same here. It’s too fast for my brain to process 😅 Reducing the playback speed to 0.75x made it easier. Anyway, great content!
So, the best way to test TP limiting against normal limiting would be to match gain reduction. Anyone thinking TP limiting sounds worse is most likely hearing the extra limiting happing. Correct?
Correct; not only gain match but also peak normalise the result!
Well seems that true peak does still sound worse. You get less gain. So same limiting effect but less increase in gain. It’s good for stuff pushing less loud material but when going loud, if this is correct and I understand it correctly, true peak still sucks
Dont get me wrong i just try to undersrand. If your ears telling you that it is worst than before all that overdampling process. We all think there will be the best natural sound when we keep it as it was sound designed and mixed, but it is just because of the two différents approch because with know technicaly that if we respect all this process, it cant be worst it is mathemathicly impossible because of the high oversampling.
Or maybe you talk only about the mastering end of the end of the process.
Production value on these vids is top notch
Thanks mate!
these sort of issues happen in all sorts of places in the digital domain... don't worry about oversampling, just work at 88.2, 96, 176 or 192kHz 😊
I'm not sure if this is in the video but this applies to 48 kilohertz sample rate and The oversampling required is less proportionally for higher sample rates.
What an absolutely brilliant video. Thank you so much for compiling this information. You've really helped me out here!
Nice video man! I've also asked Fabian several DSP questions in the past as well haha! He's such a nice and helpful human being.
The one question that is killing me after watching this video is: does the truepeak limiting mode in L2 also get rid of aliasing the same way as 8x oversampling? I'm wondering if there is a difference in the AA filters or if it's literally just a shortcut to 8X oversampling...
Nice job doing your research - wish there were more people out there putting in this much effort and research before making a video.
Might I make a kind suggestion to maybe add a references section to your description for future videos when you pull literature from the web so others can more easily find the same sources that you use in your videos? Cheers man!
Hey!
Yes he's incredibly helpful; FabFilter's truepeak mode oversamples the side-chain signal which triggers the gain reduction; not the signal itself; so if you don't want to oversample but have the same accuracy viola; the TP mode is for you; which would make sense when you're in a session at let's say 192k
YES! I must add reference sections; I get lazy on the publishing side of things; because I'm so fast paced between sessions at the studio and making these videos;
@@panorama_mastering One more question since I've gone down this rabbit hole many times before haha!
If I understand this correctly, the only difference between using a standard limiter vs true peak limiter when mastering our music is that we would be encoding intersample peaks if we use a standard limiter. Now, from what I understand, this isn't a problem if the DAC has the headroom to reconstruct the intersample peaks without clipping. Given that most DACS these days are pretty good, many can accommodate some significant intersample peaks without noticeable distortion.
From my blind shootouts I've conducted, I haven't been able to consistently identify a song that was encoded with large intersample peaks vs the same gainmatched song without ISPs. They sound essentially the same to me... but I do listen back with some very high-quality audio gear.
My thoughts are that the final sound of the master ultimately comes down to the DAC quality of the consumer, which will dictate if they will ever hear these ISPs being distorted or not. Given that most DACS nowadays are really good, I have my doubts that anyone would ever notice a significant difference between true peak limiting and standard (assuming levels are correctly gain matched). But if you happen to be listening to music on a Sony Walkman from the 90s, then maybe the true peak limiter would help keep those kicks and snares from getting crunchy.... thoughts?
I guess this all boils down to do you want to make your masters slightly louder with the risk of ISPs or quieter without much risk of ISPs...
One thing I clearly hear on normal limiters and true peak limiters, is that true peak limiters are more "agressive". I definitely hear they catch too much of a signal, same gain reduction applied. For super loud materials, there's no way I keep using them. All references masters I do have from other studios, don't use True Peaks as well. The amount of True Peak volume above 0 is incredible, lot of tracks peaks +2dB TP (sometimes even more). Not that I like it, but clearly these same tracks with TP Limiters sounds ass.
For streaming purposes, I use them, but I make sure it does not limit more than 0.2 gain reduction, or not even go through threshold, just a safe spot to follow streaming recommandations.
Beeing that said, very informative video, thank you! Wish we could get that much of infos in French, that was hard to follow at some points >< It's extremely technical! I still miss some vocabulary.
Best Audio Channel on RUclips!!!
Super humbling; thank you muchly!
It s difficult to understand ...but it seems pertinent !!!!! But what i like with you Nicholas ; very often you "ve mathematical approach !!! and for me it's not the best and easier way .But losing habits is the key to change and with a little luck , to evolve. . Cheers 😎🌍.... 🐌
Wow nice, this has been very informative! Thanks! Now it makes sense why I don't like the sound of TP limiting. It's actually just increasing the precision which results in more compression from the limiter.
EXACTLY!!!!
💣 i love it.
Hey good to see you here on RUclips! Killer video man, very informative and helpful for understanding what's going on when I'm limiting!
Hey Mate! Great to see you too! Just had to bump your name on INSTA, I remember you from my trip earlier this year in the states!!
I might be making another trip out to LA next year too!
@@panorama_mastering sweet man! I'll for sure be heading to LA for NAMM again this year it was 👌
Finally I understand why it is a good thing to do oversampling! Something wrong with 32x oversampling AND TP-Limiting at the same time?
It just becomes redundant. Introduce processing for a reason/cause.
That said some TP limiters will only oversample the sidechain signal NOT the processed
Great demo about this. I had to wonder if true peak limiting was a good thing. I've been using it, but with so many saying it's bad, makes you consider.
So it seems true peak limiting or 4-8X oversample gives similar results, and neither change the audio perceptibly.
I'll have to switch between true peak and oversample to see what happens. Right now I'm clipping into an SSL X-Limit. Other limiters I've got: LVC Limited-MAX, TDR Limiter 6 GE, Venomode Maximal 3, Nugen ISL-2, stock Studio One, Slate FG-X2 in their Complete bundle, D16 Frontier...
I always separate True Peak limiting into its own stage after my main limiter. Since it detects higher peaks than a standard limiter, it would cause my main limiter to overwork itself for the same result. As such, I've found it best to have it separately to catch the errant True Peaks here and there, while my main limiter does the actual limiting.
Great tip! I actually was introduced to this method by another person too; and with maximizer; (ozone) a seperate instance of limiting with 0thresh and TP engaged sounds better!
are truepeaks even something worth worrying about if you're working with an agressive genre like dubstep and doing your final stage of limiting with a clipper instead of a limited. I mean if something gets past the clipper and... clips... well then it's just doing what it's supposed to do.
Depends what you're trying to achieve; the harmonics distortions of hard-clipping; stylistically is GREAT in dubstep it adds a punchy edge to the material;
The knowledge here-in this video is about understanding more comprehensively the toolsets we have at our fingertips to more comprehensively verse ourselves with out tools/application.
@@panorama_mastering thanks, love learning more about tools, but I don't really think I'll need this one ^^
Nico I remember years ago listening to Steve Duda telling Deadmou5 to have the limiter output set to -3db because of intersample peaking. I'm guessing this is what he was talking about? Thanks for the vid!
I'm surprised that you didn't switch to Reaper yet, since Reaper has built in ReSampling for plugins and the FX chain that can function on any plugin, and you can go up to X32...and many more amazing options to optimize Reaper for Mastering application...
When I was a mastering assitant, I cut my teeth capturing, sequencing, topping and tailing masters in reaper!
When I opened the studio; I had the same set-up; with Pro-Tools as the pitch and Reaper as the catch;
But I streamlines everything into Pro-Tools, for no particular reason!
Looking forward to this
thanks mate!
i just bought brainworx true peak and have other limiters including sonible...even from the first few minutes of your thoughtful and detailed presentation style (first time seeing your channel) ...subbed! 👍🏻
Great to hear! Welcome! Don't be a stranger in the comments; OR afraid to call me out if you have a differing opinion; this is a platform for learning!
Thxs for sharing your knowledge with us, Pure gold ❤
Glad you enjoyed it! My pleasure!
Interestingly when i have my Pro L2 set to 32x oversampling it still is different to 32x oversampling + true peak on. Does it oversample twice?
This is absolutely gold. Thank you so much for everything you do.
I have one remaining question - can it be said that the True Peak Button is roughly the same as the 8x oversampling setting? So in the end it wouldn't really matter (assuming I want 8x oversampling on this particular track) if I turn on True Peak OR put the oversampling at 8x, right?
My pleasure!
After completing this video; I was on a call with the product manager for iZotope ozone; and he showed me in plug-in doctor; that the TP mode in FF Pro-L2 is actually creating a SC @ 8x oversampling; so it detects the same signal as would an 8x oversampled one would; but the TP still processes at the native sample rate;
That said; I would opt for 8x oversampling because it also helps alleviate the effects of aliaising distoriton;
Thank you for all of your super informative videos, and the enthusiasm with which you share this cutting-edge information :). My only question is with regards to using the Ozone Maximizer, I don't believe we're able to turn on oversampling without engaging the true peak, which engages oversampling when it's turned on. Since this is the case would you recommend using a different limiter in order to oversample without engaging true peak, or is the Ozone Maximizer doing something different under the hood where it's true peak isn't affecting things in the same way? Thanks so much!
absolute great, i thought everytime to not use to much oversampling.
Great Video(As usual)
so basically true peak limiting is the same as limiting with upsampling?
The ballistics work EXACTLY the same, but there's oversampling mixed in for more accurate sample measurement;
Great video, you're really passionate about mastering we should work one day, I'll make you my mastering engineer. Also I've been oversampling my limiter to x16 on Reaper since forever.
Sounds like a plan! Keen to hear your mixes :)
There’s a matter of fact that DACs can handle overshoots over 0 with no problem, with no distortion. DACs are designed with some headroom above 0dBFS.
Is this the case of all DAC’s, what about after lossy encoding? How is that measurably handled? I am asking this because I am not expertly verses on the DAC side especially in cosumer electronics, but it’s a worthwhile dive for another video!!!
This is great info. Thanks for the research and video.
I’ve been doing experiments with multi-band metering and noticed my true peak levels were off compared to the broadband. After your video I think the intersample peaks may be the issue.
Wow this was mind blowing, thanks for sharing this info with us.
Glad you enjoyed it! My pleasure!
Outstanding work and much appreciated.
Glad you enjoyed it!
Thanks man, much appreciated!
Just want to make sure I understand something you said at the end. The best thing is probably to use a normal limiter with over sampling rather than True Peak. Is that right?
Bang on!
@@panorama_mastering Got it! Thanks😊
This was gold :) thank you so much :)
You're so welcome!
Awesome content!!!! Thank you so much!
Glad you liked it! Thanks for watching!
Great Video!
Wow, you're technical
The topic is technical; I'm not the most well versed engineer out there; a lot of this stuff I have to heavily research, re-read, annotate, and it takes some time to sink in; some of the people mentioned in this video are the real "technicians" who are comprehensively versed in DSP;
Hello, first of all thanks for your study, for people who are not engineers like me it is difficult to understand a little ... but I think I'm getting your idea, I would be eternally grateful if you solve my doubts as a novice.
1.- according to your study, it is more convenient in Fab Filter Pro L2 to deactivate ''True Peak limiting'', leave it deactivated and activate ''Oversampling'' to x8.
If I can opt for x32 is it much better?
2.- when I am mastering a disc of my music, I put a L2 and at the end of the chain a '' Youlean loudness '', and in '' True Peak max '' it marks me in red according to my '' output '' in fabfilter L2, I explain me .... The peaks stop appearing in red only when put in my Fabfilter L2, in ''output'' at -1db.
3.- after doing that I have another problem... my numbers in the ''master channel'' don't reach 0-db... but my limiter solves the peaks... but at the same time it doesn't let me reach -0db in the master... this confuses me a little, since I need to upload my music to Spotify... I'm sorry if I'm confusing, I'm from latinamerica.
4.-
if i have all my songs separately, mastered and i want to join them all in a daw to join them and sound one after another, should i put a limited? or should i just put them all, export in wav and ready ... please this topic also confuses me a lot ....
this is highly enlightening thaks a lot
My pleasure!
the basic, use it since yearssssss, why people still use the bad representation of zero order hold. Of course true peak don't get a sound just increase your precision.
At sametime not all oversampling are the same in limiter. With some in *4 output -0,3 db you set a true peak meter after you will never clip, and some at same setting still clipping.
So spend time to learn and measure your stuff. So with basic limiter no oversampling or true peak a -1 db output like apple ask is good, but if you get good limiter you can win 0,7db. Wav master at -0,3 db convert in MP3 320 you get less than 50 true peak clipping, in most commercial track you can get 20 000
BANG ON!
Very helpful, excellent.
Awesome analysis
Thank you!
Great content mate 🙏🏻
Great video. Wouldn’t this underread issue also apply to compression plugins? Unless they oversample?
Sorry my expression but this info is fucking awesome!!! - Thanks for that amazing research!
My absolute pleasure! Thank you for watching!
So you’re simply saying turning on true peak and over sampling is best right??
For most circumstances it will cover you; yes!
Did any of the developers talk about peak detection in the input signal vs the sidechain? I'm curious if they had any comments about dealing with the ringing from the interpolation filter.
Funnily enough, I was on a call with Bill Podolak not too long after the production of this video and this point came up in discussion;
Currently on Plugin Doctor; FabFilter Pro-L2 linear anlysis/phase shows that it's most likely "the sidechain", but I'm yet to look into it FURTHER!
Very interesting, I imagine this was quite time consuming to put together so thank you. I think this maybe this is one of those occasions that makes sense in theory but not in practice, because I'm pretty sure that most of us (I assume it's not just me😂) can easily hear that it sounds worse when TP is engaged, than when it's turned off. It's about the easiest test you can do, just toggle it on and off and you can clearly hear the difference.
"It's about the easiest test you can do, just toggle it on and off and you can clearly hear the difference." is fundamentally incorrect;
The ballistics of the limiter are operating in the same manner; but the measurement/detection is altered;
A TRUE test matches the measurement/detection like in said tests and normalises the peaks against one another...
When you monitor a digital limiter with overshoots of up to 3dB and doing the equivalent less gain reduction; the difference will be night and day;
But if you compensate the threshold and post processing normalise the peaks; you'll find the ballistics will be shooting in the same manner;
Also; as always NoQualms, muchly appreciate the support and love you show for these videos as always! So big thanks for sticking along this video; it was a passion project to pull this together!
John Doe; - That's super fair; I can respect this positioning; and it's definitely why I make these video's it's to get us thinking and being considerate of our applications; BUT ULTIMATELY not to tell people how to "suck-eggs" but give them the tools to make their own educated decisions based on their experiences; so much respect for following along on the channel; I'm hoping these videos and (More like these to come!!) continue to help the channel grow and reach/help more proffesionals;
Engaging TP makes the limiter work harder without increasing loudness, so this is not surprising. The problem with letting the overs through is that you have zero control over how well the listener’s playback system handles overs. Potentially it can be a lot uglier than the additional gain reduction of TP (or oversampled) mode
A Maeland, that’s a vaid point I agree with; but measuably, how well do DAC’s on the most popular consumer electronics perform?
So.... should I bother with True peak limiting, or just say screw it and leave it off like 99% of the top 40 mixes? haha (awesome video btw, subbing now)
Depends!
@@panorama_mastering lol, what about the whole -1 db of headroom thing? I've also noticed 0% of top mixes doing that
thank you so much man.
You're welcome!
So, with the Pro L2, is there any advantage to using 16 or 32 x oversampling? (I generally leave it on 32 as my M1 seems okay with that CPU load-wise) How does TP and oversampling interact? If I'm oversampling at 32 is there any point in also having TP on?
So after all of this knowledge you've gained, will you use TP limiting on your next masters or you decide to stick to your previous way of working without using TP limiting?
Good question; funnily enough I've always used 8x oversampling in FF Pro-L2; so in practice I've always been reducing these undereads (Even if I haven't been conscious of it)
So with FF Pro-L2 I WON'T change the way I'm working or turn TP limiting on;
However in iZotope Ozone Maximizer; I will turn TruePeak limiting ON because as confirmed with their engineers; it enables the 4x oversampling which helps reduce those over-reads;
And I use both limiters for different use cases;
@@panorama_mastering finally we know about oversampling in ozone. why don't they tell in the manual? Luckily I turned TP on anyways. Thank you for your work even though I don't get the math haha
@@panorama_mastering Thank you for the info Nicholas! When you say Maximizer's TP button enables 4x, are you saying that it is doing this to create more ISPs to show up in the meters? I would have thought that a better "brickwall solution" like the TP button would be non-oversampled since oversampling creates ISPs...how does that work?
I am struggling to understand. Does this mean TP Limiting is actually more accurate and a good thing?
When I test the same production of mine on TP meters and sites there is a warning when I leave it off and the TP Is almost 0.5 higher.
"Does this mean TP Limiting is actually more accurate and a good thing? " Absolutely correct!
@@panorama_mastering Thank you for your insight. I appreciate your dillient research.
Thanks for the science it was awesome!
Glad you enjoyed it! My pleasure!
True that !!! Done the experiment before
Nice!
mind blowing! what a best seller
Yeap! Same, the more peeled back the curtains the more I dug into the techy side of this and there’s so much more than meets the eye
How about less is more with fewer tracks in the mix? True peak limiter works better?
Thanks for your video.
What about the limiters that do not offer oversampling only TP limiting?
- Ozone Maximizer + Vintage Limiter
- TDR Limiter 6GE
Would you still set your ceiling at -0.2 dB TP?
In TDR Limiter 6GE you can deactivate TP Limiting and set the Processing Quality to Insane.
@@HeartleaderMusic
Thanks, I might give it a try and see how it fares against TP.
Proved it myself
Since you're referring to Pro-L for the most part, why no mention of 16x or 32x oversampling?
Thanks!
Welcome!
So basically true peak limiting sounds worse because it does more gain reduction than non peak true limiting, with the same threshold setting. right?
TP limiting can have more gain reduction because of it's accuracy; that results in it's "sound" even though the ballistics of attack, release, lookahead all function the exact same;
Ohhhhhhh that ISSS what's happening... You see I theorized this last week in a much less technical way, but because I often don't turn on true peak until I've already dialed it the limiter I always turned it back off immediately because of what it did to the transient information, plus... I haven't been aiming for Apple Digital Masters anyways 😳
yeap ! It's a very funny user-workflow error inherent int he design of just having a "button" which you switch on and off;
if when it was enabled the gain was offset by the difference of maximum under-read between current PCM encoded data and given upsampling size that would solve this;
I wonder if they thought of that or made a conscious decision not to dummy proof it, send this to the Fabfilter folks
Ha YES! I shal;
So, i always used true peak limiter, should i stop? i didnt understanding correctly, causes all music we download on traxsource or beatport, when you run trough the youlean loudness meter, they are clipped above the 0 db... all music, all artists all clipped, i dont know what to do anymore, causes i think i am wrong mastering in wav wihout clipping, i always target bellow the 0 db..
how about 8x oversampling with the True Peak limiter like Pro-L2, which has the option to do both. Is there any benefit to it compared to the non-TP limiter with equal amount of OS ?
Measurably; we may be talking within tenths if not hundredths of a dB of tolerance if you're both TP limiting engaged and 8x oversampling;
My bet is stick to 8x oversampling in FF Pro-L2; you know the max overshoot will be 0.168, so set your ceiling to 0.2dB and you're completely safe!
@@panorama_mastering What in your opinion is the ideal true peak level considering encoding to MP3 and for streaming platforms?
Spotify reccomend up to -2.0dB true peak. I have used Ozone's codec to see how this affects the final reading and it seems that up to 1dB can be added during encoding which can result in unwanted clipping.
i love u so much bro, thanks very much
You're welcome!
On the Ozone Maximizers i still don’t like ultimately what the True Peak button does to the sound. Even when dialed back. Do you think it’s worth the hassle of pulling up something like DDMF Metaplugin to oversample it? If so, what’s the multiplier you recommend?
Sessions at 44.1/48k (8x oversampling)
Sessions at 96k (4x oversampling)
Both will bring you within a 0.169dB of tollerance with under-reads which is exceptional!
Well, I don't hear any distortion with intersample clipping or mushed transients with true peak limiting... It's exactly the same for me.
Very informative!
My pleasure!
Every analog device can handle few dbs over zero..so is TP really needed ?
Maybe, maybe not; I'm not exceptionally well versed in how cheaper devices handle overs at the DA stage; other than I've "experienced" them not handling significant overs very well; (Subjective not objective) This is only my experience...
TP limiting is not needed for the sake of "avoiding" true peaks over 0; but more-so for better and more accurate measurements for a limiter to operate by;
Wow :)
Thanks mate!
but true peak uses less or more cpu than just 8x?
Hope you'll at least get Pro-L3 for free, these lessons are pure gold
Haha thanks mate!
Great. Video. So what happened when I clip a signal, with a clipper, then I have only one peak = true peak . Or I am wrong ? Thanks
Definitely not wrong; and this is a TECHNIQUE I use in mastering with overly transient/pokey mixes;
The theoretical "true peak" is the point in which the reconstructed wave-form reads higher than the adjacent sample points;
Great info but in real life you won't get a loud master using true peak. It's sucks all the life out of a track if you push a track hard into a TP brick wall limiter.
True Peak sucks transients and punch, as well as oversampling in PRO L2 for example..
Rephase that statement; True Peak is a more accurate form of limiting that controls overshoots.
I'll stick with a clipper 😉
Ain't nothing wrong with that at all!
@@panorama_mastering Sure. I forgot to thanks for your very interesting video, btw!
“While we gain knowledge we also gain ignorance” or indifferent? How did you say it?
I think that’s actually true! 😂 haters gonna hate
works gj
thanks mate!
Pro L 2 has the most audible sound quality difference when engaging oversampling than any other plugin I know.
Nice work. BUT, could you please give people a little more time to consume your waterfall of words? Faster than a spanish grandmother on Amphetamine.
Maybe! I try; honestly! ....
@@panorama_mastering Now i am really sorry about my colorful description. I actually really enjoyed your video. 100 Likes!
@@cemski haha no need to apologize at all!
This feedback keeps me honest and allows me to tune my messaging and help these videos connect/help others BETTER
hahsdhahah good
thanks mate!
Eight minutes of word salad. Thanks.
This is all not relevant and doesn't help you to build a great song! 🚧 Forget all this nifty stuff and concentrate on song creating. Otherwise you waste your lifetime for nuts!
I never said this will help you build a great song; my work as a mastering engineer is one of my life's passions; that why I take learning comprehensively my tools-sets so seriously;
@@panorama_mastering Does it help you to know your profile of your tires when you came into a aqua planing situation? So, are you constantly driven all your tracks or master volume at 0db? 24bit or 108db should the minimum of your projects you work with. So you can be constantly below about -6db without quality issues. Take it easy and good luck! 👍
Roger is waffling so much non sense haha 😮💨 Thank you Nic for the Video 🙏🏽😂
Thanks for clearing that out 🫶
No worries!
There is far to many technical BS videos about music on RUclips stop wasting time trying to learn how to micro manage every tiny little issue that may not even be an issue after you've mixed your track, none of this crap even mattered 20 years ago ferry corstens track " out of the blue" wasn't even saturated or compressed and still went on to blow the roof off clubs and hit number 1 in the dance charts, as long as you have the very basic of plugins instead of wasting money of big expensive plugins that are not needed at all, then have a good arrangement good leveling have a clean low boomy sub, and crank up the mid range your 99% done.
Music has now become all about sound quality and not about making a great track which is why there is a huge avalanche of shit being produced compared to the 90s.
Engineers have always been concerned about the sound quality. This is my profession and in my best interest to sweat the small details so creatives, artists and producers have the freedom to create at will, it's my job to make that all fit together.
Understanding concepts like this is very important!
@@panorama_mastering In my case, turning true peak ""off sounds cleaner/clearer to my ears. I have A/B'ed this with Ozone, X-Limit, and also Pro L2. I do use oversampling whenever possible and the true peak feature still clouds/muffles the topend of the audio.