Updated mastering techniques 13 May 2024
HTML-код
- Опубликовано: 12 май 2024
- 📧 The BEST newsletter for working audio professionals who strive for more! Subscribe here - www.panoramamastering.com.au/...
🎹 Proud of your mix? AWESOME! Get a fresh set of ears on board for mastering and let’s work together! - panoramamastering.com.au
🛠️ Mastering Template for ProTools, Logic and Ableton- www.panoramamastering.com.au/...
--
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/Panorama_Mastering
Hey Nicholas. I love how you dissect audio like this! Great deep dive and synopsis into the clipping / oversampling topic from my video. I just linked to this video in the pinned comment so people can watch both. It's always a pleasure to see you analyze audio in RX Audio Editor. The workflow in Kotelnikov is also fun to see, so thanks for that. I'm working on our interview this week, so it'll be good timing to publish that soon :). Cheers mate!
Awesome; I'm excited for everyone to watch it!
Well, I know I forwarded that video to you, I wonder how many others did too LOL.
I certainly appreciate your comments, and it’s always good to see someone being open to updating their workflow without ego getting in the way. That transparency and openness more than anything really is an inspiration to the community at large.
Yeap! Thanks for pinging it across to me; there were a few others who did to! Hot topic!
Voxengo OVC-128. It oversamples 128 times and somehow is not destroying the CPU. Check it out, it claims to be the cleanest clipper.
Maybe they have a better calculator at the end.
@@americatunedright1211 It's not they, it's a one mad scientist from north of Russia 😂 And it has linear oversampling option in settings as I remember
Might be worth watching the video again
@@voinrima send link to article if any. 128 irrelevant imo, it’s the math coming down or kernel smoothing after output or ceiling whatever the order, geez now that I think about it ringing would be interesting to see how it folds back… you’re right this guy has to be groundbreaking with math skills to calculate an optimize cpu, kernel, or whatever mumble jumble needed. When I get the nerd bug I’ll look into that, thanks.
@@americatunedright1211 You can google it like "Voxengo OVC-128 User Guide", I am not sure if YT allows to send links. Also I've just checked in my last project, it overloads about 0.5 due to HF content, and also does not have the linear oversampling.
Nice - I saw Drew's video & found I could replicate that behaviour too (though I had to push the 2 bus a bit harder than normal). Thanks for going deep on it!
In regard to clipping introducing overs when down sampling. I believe this occurs where the clipped portion of the wave occurs between two sample points at the native sample rate. As the wave form is reconstructed you’re essentially reconstructed the non-clipped waveform. I get around this issue by setting a TP limiter to the same ceiling as my clipper and the result is fairly transparent.
Another issue of clipping not talked about on the 2bus is intermodulation distortion resulting from the more complex frequency content resulting in a different harmonic output than per bus clipping would. Based on your findings, I think intermodulation distortion might be playing a significant role in these overs as well.
On second thought, higher frequencies causing this issue is more indicative of sample to waveform reconstruction, as higher frequencies require more samples per second to accurately represent their frequency and would be subject to a higher error rate of becoming “non hard clipped” during downsampling due to the resolution of samples per cycle per length of the transient.
Would love to see you do a breakdown Kotelnikov and how you use it
Will do !
Will do !
Well said cause we need it
Well said cause we need it
yes please!
You saw that vid too! 🤣🤣🤣 I saw that, I was like 🤯🤯🤯
the free version is good but I love the Inertia and FDR on the paid version. I have strong feelings that you'd go crazy about it once you really try and learn every single thing about the Inertia mode. FDR saved me countless of time from destroying the mix with Multiband comp. Best compressor ever.
do you still recommend having the 17% knee standard clip on the mix buss?
Thanks for this Nicholas! I saw that video as well and was REALLY curious on your take on it. I immediately started testing as well. Always appreciate your videos. 🙏🏼
My pleasure! Thanks for watching!
Thank you for all these gems Nicholas 🤝
yaaayyy, Video under 10 Minutes .. thanks for the Update!
Have you tried TDR ULTRASONIC right before the clipper?
no! but you're not the first person to mention this;
Would really love one of your fine vids on that one where you analyse different uses of it;)) Thanks always for your work🙏
I use this one as well , great tool, I always export the master at x32 because it feels the purpose of my track pretty well in between aliasing content but how my ear are satisfied with the results . I know perfection is not always a good thing and if it sounds good enough to me I believe it sounds good enough to others without being to surgical but that my own take on it . I don’t know who rules the whole thing but we all know about the vast amount of conversations we could have around any sound topic probably for endless hours even sometimes forgetting ourselves because too passionate by what we do 😊
That's how we roll.
Clipping:
How would TDK Ultrasonic play into that?
Could it have an additional bonus of keeping reflections out of the game?'
Where in the chain would it make sense?
Not sure; I'd say BEFORE the clipper;
So glad you did this cause I just watched his video yesterday and I usually use standard clip in soft clip pro mode on my drums and I was like am I doing this wrong by putting it on 8x oversampling
You're welcome!
The Kotelnikov GE has frequency-dependent compression as well...
I know. But I very very rarely ever come across a project where that's an issue/necessity for me;
💯💯💯
I know you love your specific clipper, but if diferant clippers use diferant OS LP filters (and some will do), this effect may be minimised by a diferant choice when it comes to what OS filter they decided to use.
It might be worth while doing a shoot out to see if diferant clippers cause this to happen more or less, and see if there's a benafit in shopping around a bit. I know when it comes to saturation the type and curve etc. of the lpf can have quite profound effects on a number of variables, and saturation is not much diferant to clipping mathematically (hard clipping just essentually being one of the more extreme forms of saturation there is, but being a form of saturation none the less).
I'm now wondering what happens if we add another clipper with no oversampling after the already clipped signal?
Dude, what about just using a true peak limiter after your clipper to stop the overshoots with over-sampling? generally very small sporadic Gr that isnt particularly unmusical?
5:15 at one frequency does this over sample peak increase happen?
Hello, useful information, I'm from Russia, I love watching your videos, please make a video about dithering, when to use it and what settings to set in izotop
Great suggestion! Will do.
Would be really interested in your kotelnikov presets, im a (fairly new) email subscriber, how to find it? Thanks a lot!
Ping me an email and I'll link you ;)
I need to read about oversampling. Did not understand it yet.
Basically when plugins oversample they need to downsample again after they do their thing, back down to the project sample rate. They have to apply a filter before doing this to avoid Aliasing distortion which would completely negate the oversampling process. Its this filter on the end that causes the phase shift because its a min phase EQ. If they added a Linear phase filter you wouldnt get that but you would get prerining. Honestly its an option that it would be nice to toggle
Hmm I cant find the presets for the Kotelnikov, anyone an idea?
It’s another reason for 96k mixing. The up down sampling quality relies on great math smoothing of filter kernels and if that fails being in 96k really helps before the cliff. The results in the end benefit greatly because of mathematics in the digital realm, none of this matters in the analog domain.
A good lesson for you today, not so much the subject but the humbling, now you bettered yourself for it. It’s how I learned that I know shit😅
I usually work in 88.2k for everything. The original idea was to "force" oversampling. Since then I've found other benefits such as a removal of EQ cramping, automatically changing my recording sample rate, etc.
Do you see any benefit of 96k vs 88.2k? My thought was that 88.2 would give the same benefits but save on processing power and storage space over time.
@@alexarmstrong449 I used to do 88.2 around 2012 but things are much more digital that 96k became standard for me for a plethora of reasons, I don’t do 32bit which helps with large files (96k 24bit). But I use 44.1 deliveries for bass driving music like hiphop or even just mix at 44.1 for those types and stay caution which plugins I use, if I time stretch I’ll up and down sample or if I ever own seroto pro pitch n time it won’t be necessary but that plugin on its own is damn near $900 bucks!😬. Plus HD these days are cheap and fast.
Idk 0.5-1.5 db more on peaks that hit the limiter for a millisecond is not that of a deal
@@americatunedright1211 interesting. Could you elaborate on situations where you would use 44.1k mixing?
@@voinrima for mastering it's not a huge deal, I'd agree. The limiter will handle that usually. But where it really adds up is in mixing. If you're running hard clippers to take out micro-transients in your mix, clipping individual drums for example, or perc parts, or arp synths, then those all produce overshoots when using oversampling and they will all stack up. It's possible to get overshoots of up to 9% of the jump in amplitude on a transient, which is quite a lot. In fact, downsampling in any plugin for any reason at any rate to any lower rate produces overshoots. Any time you band limit the HF of a signal, you create overshoots. This is why the iZo RX SRC downsampling algorithm has a "post limiter" in it.
Funny how much compromises we ready to make for loudness sake. Reduce top end sparkle (filtering) or accept introducing more distortion during clipping (no oversampling) to gain 1 little db at best. When is it too much "material hurting" for decibel economy ?
I think you didn't see the original video he's referring to. You're assuming that introducing a couple dB of gentle hard clipping is producing audible distortion and that oversampling "fixes this". In most cases it's not. The clipping harmonics are getting masked by the transient, which contains all frequencies and is very loud. Or if you do end up getting a touch of aliasing, it's adding a slight bit of density to the transient which increases perceived loudness.
@@warpacademy Well I did see that video, and it was clear that the distortion was audible when no oversampling, the test he conducted showed it, and the material became gritty. That's when he concluded that he uses 4x oversampling, and beyond that there was no real gain compared to cpu stress. Now if it wasn't the case, why would he filter high frequencies to still use oversampling with no effect on peak value instead of simply no oversampling ?
This deserves a video of it's own; I've been diving into a lot more A/B's and there are few things I'm doing "less heavy handed" these days that are producing great results with less comprimise and also still finalising LOUD
@@bakerlefdaoui6801 sorry for the confusion. I meant this video: ruclips.net/video/5sAm7McrkA0/видео.html
TDR ULTRASONIC
why not just adding one clipper without oversampling after the one with oversampling
This is exactly what most clippers do. It's the parameter called "Ceiling" in Standard Clip and K-Clip. The issue with that is: Why are we using hard clipping? To create headroom. What's the problem with pushing clippers too hard? They create distortion. So when you create overshoots, and then just seek to deal with that with a second, non-oversampled clipper, you now create extra distortion to deal with the increase in peak level. IMO it's better (with this type of clipping) to not create the overshoots in the first place. Unless you really need oversampling to deal with audible aliasing, such as with soft clipping, or pushing a hard clipper fairly hard.
May as well not oversample in the first place, because you're going to be introducing (as you can see here by the overs created) an additional 1dB of hard clipping after the fact; may have went with 1x to begin with;
@@panorama_mastering exactly!
@@warpacademy ah okay I get your point. I need to investigate it more.
Clippers suck theres always distortion
Same with compressors, limiters, gates, multiband compressors and saturators.
@@panorama_mastering not like clippers though clippers make it sound like the clicking and scratching of a record player