Each time you run a signal through a limiter that has modes (such as Ozone and Pro-L2) it imparts a certain character to the signal of that mode. So having the signal run through that mode multiple times you are imparting that modes character into the signal multiple times on top of each other, that is why it won't null. Just listen (compare the original signal) to how the modes change the character of the original signal.
I just watched the video, and the comparison here doesn't really make sense because we're talking about different types of filters. A minimum-phase filter requires less time for processing compared to a linear-phase filter. Additionally, the processing curve in a linear-phase filter is designed so that the so-called filter delay occurs both before and after the actual signal, resulting in what’s known as pre-ringing. In contrast, a minimum-phase filter only produces post-ringing. Therefore, we are dealing with two different filter latencies, which also depend on the specific filters used for the processing. Performing an exact null test would require knowledge of the filter length and the precise timing of the signal's processing. This would prove to be extremely challenging since we didn’t program the plugin ourselves. Furthermore, I’d like to point out that oversampling does not inherently destroy the sound. The perceived sound issues often stem from what happens before the oversampling process. In the analog world, we don’t need oversampling because we’re not limited by the Nyquist frequency. Effectively, if we experience distorted sound due to oversampling, it’s usually because the prior signal processing was flawed. The oversampling process (which includes upsampling, processing, and downsampling) simply mirrors what is standard in the analog domain - working with significantly more sample points. When the signal is subsequently downsampled to the original sample rate, the average value between the sample points is usually calculated. Depending on the changes made to the signal, rounding errors can occur, making the signal appear more stepped. The key here is how the programmers designed the window size for filter calculation. Generally, there are two possible causes for audible artifacts: 1. The signal was poorly processed before oversampling, making artifacts more apparent at higher resolution. 2. The calculation of the average level over a certain time period introduces rounding errors, which make the artifacts audible.
Hey Nicholas. Super interesting discussion here. I love the explorative, testing-based approach you take and your commitment to using your ears first, and analyzers second. Your comments on linear phase aliasing LPFs is on point. I did an extensive project this year with Sound Theory on their Kraftur plugin. Their designer has his PHD in theoretical physics and really goes deep into things. He chose to design Kraftur's crossovers with minimum-phase filters, which I questioned, and he really took me to school! Our discussion was fascinating and got me to rethink many of the ways I've been working. His comments really centered around transient smearing, like you illuminated here. He also noted that you can't actually hear the dreaded "phase smearing" at the crossover points with the way they designed the crossovers. Interesting stuff. Thanks for leading the conversation as we head into 2025. I hope you have a wonderful time at NAMM!
Got a top tip for you, don't bother with any of it!! just release unmastered. Make your mix as loud as a referenced master with nothing on the output and I promise you it will get to a better place faster. I've been releasing unmastered professionally for nearly 10 years now.
Glad to see someone else publicly embracing stem separating in mastering. I felt a little dirty about using it 😂 but I always ensure I use linear phase processing when tweaking. So why the hell not? It’s the future baby!
@@Durkhead There’s no rule to say “you must have a separate person do your mastering”. But there’s many reasons you might choose to do so. For example, the mixing engineer may have less objectivity having worked on the track for too long. He may be an excellent mixer but not so much at mastering. Having a 3rd, 4th, 5th pair of ears can help ensure the final result is more objectivity pleasing and translating well across different systems. Personally I do both more often than not for the convenience of the client.
Minimum phase filters while over sampling is just like adding a low pass yourself at nyquist and still is a form of time dilation though, using minimum on the entire frequency maybe less harmful on a 2-track it can be possibly harmful if you end up doing a stem mastering job. That's why I master at 96 kHz and don't have to worry about filter cramping or compression artifacts. When it comes to processing that emulates the analog domain by introducing 2nd, 3rd, 5th harmonics, etc, I highly recommend oversampling, but only per plugin instance since intermodulation can occur if you have reflections above nyquist that are also then processed by analogue emulation. In a mastering context whether you use oversampling or not, or Linear phase or not has everything to do with the workflow you have due to the tools you've purchased. I still think a System 6000 with MD5 and Brickwall2 are the best solution for a mastering studio and only the true peak limiter in Brickwall2 does oversampling. So meh, maybe it's all overrated.
I love the way you teach man; your enthusiasm is contagious and makes me excited to get in the DAW. You're a great teacher and the way you describe/explain things with an open mind so we can follow your train of thought, is a type of teaching that really works for me. You discover things and share them, and we learn not only these things but the right attitude for moving forwards and continuing to grow. Thankyou for the great vids, u deserve a sick 2025! 🎺🦒
I’m no seasoned professional but I never even considered stem splitting to master. I’ve had a couple clients send me stems to master but I never thought to separate them myself. Thats huge!
About stem EQing, i've been quite experimenting with it lately with great results and to my actual impression and understanding it always works best with linear phase eq (as the stems are always a bit overlapping and using non-linear phase eq induces phase shifts between the stems). Curious to read your impressions and experience on the topic! Cheers and thanks for sharing ;)
Really great tips! Only thing I'd disagree with is the choice of AI stem split software. Have you tested RX vs others? I did a quick comparison with a mate. I gave him my stem splits from SpectraLayers and he instantly ditched RX after hearing the stems lol. I couldn't believe how bad RX was. Can't be 100% sure which version he was using, so it could've been a poor comparison, but it's one of those processes that's improving tonnes so might be worth investigating.
There was a massive update to the quality of stem splits in RX this year for version 11. I agree, before it was really bad. Since version 11 it’s among the best I’ve heard and I stick to it, since the workflow is great.
@@simon_dupp Ah that makes sense then! I guess they're roughly about the same with each improvement. Wouldn't surprise me if they used the same algorithms on the backend.
@@PrincipalAudio If you compare spectral layers 10 to 11 there's also a huge difference in quality. They cracked the non-destructive method of splitting with SL 11 and the quality jumped up massively. I'm pretty sure RX uses very similar tech now too. I've never compared them, but seen as they have also got non-destructive splitting now, I assume it would be the same once put back together. If there's a negative difference and you are trying to use the split track on it's own it's problematic, however if you are recombining it, any difference between algorythems is negligible providing they are non destructive like all the latest ones. 😉✌️
@@LUSH_AUS there's also spectral layers from steinberg, which has non-destructive stem splitting like RX does. If you use cubase or neuendo it seamlessly integrates as it's an. ARA plug and they have built their DAWs specifically to work with it as if it was part of the DAW it's self, it's really neat. If you use a dif DAW I'd test both to see what one you prefer, as I've never compared them personally so couldn't give an overall recommendation, but if you are already in the steinberg ecosystem it just makes sence, just like if you are already in the Isotopes one and have RX already and don't use Cubase or Nuendo like Nick, the one he's using makes sence. 😉
thanks for all your reseach ! good work ! So you still prefere true peak instead of oversampling on the master limiter in 25, right? Interesting, because my mastering engineer told me that he always use oversampling on the fabfilter L2 instead of true peak because it sounds more "open" in his opinion.
I would love to see a video from you about various stereo wideners and techniques. The pros and cons and nerdy details wold be great as you deliver really great in depth information. Mainly interested in newer tools like the mastering widener in Acustica Audio Wine and Leapwing Audio StageOneV2.
I'm so glad I watched this video right until the end so I could hear you say "Ass over tits!" Also great to hear about the evolution of clipper first in chain vs clipper after EQ. Now I remember your past video where Ian Stewart had talked about this, shifting sample points, and your response was to demonstrate how the EQ you were using was not shifting phase very substantially. Of course a HPF is an extreme example. So even with the less extreme example of, perhaps, some gentle broad Q bell shaping, do you still EQ pre-clipper? Why? Why not? Finally, it is worth noting about clipper design (and most saturators) is that they will typically have an HPF built into the clipper already to remove low frequencies that can make for muddy sounding saturation. Gold Clip, for example, has a 15 Hz HPF I believe. Sometimes this function will simply be referred to as DC Offset Filtering and can be toggled on and off. I didn't know that until I started interviewing guys like Ryan Schwabe. Cheers mate!
You should try out TDR limiter when it goes on sale. It has the true peak limiter separate to the limiter as an inbuilt process, and you can put a clipper wherever you want in the chain.
That AI stem splitting is actually done through spectral analysis useually afaik. It's super cool stuff, and I had a bunch of fun playing about with it in the spectral layers .ARA plugin. The fact you can now split it non-destructively is super cool and only really been able to be done in the last year or two! The previous couple of version of that tech it would deteriorate the recombined stems, which made it hard to justify using, but now it can be split without any degredation to overall quality it's super useful for small tweaks on indavidual elements.
@panorama_mastering I'll have a dig about m8. I'm mostly going off of steinberg info on it as they have it as part of their spectral layers program, which only deals with spectral processing, and when you seperate each track it pulls out all the harmonics via AI spectrally And that side of RX appears to work very similar. I havn't gone too deep into it myself either tbf, but will see what info steinberg has on it, as they seem to be pioneering it, or at least up there with some others at the forefront. 😉
Have you done that 180-degree phase shift into mid/side encoding thing in another video? Because I know I’ve seen that phenomenon before-and pretty recently too-since I knew exactly what was going to happen. I’m a real “lose it if I don’t use it” type of person, so it must have been fairly fresh in my mind. 😂 Actually, now that I think about it, a lot of these things seem familiar. 😊 You must be recapping some stuff, eh? Either way, it’s probably good to hear it more than once. I’ve learned so much from this channel over the years, and so much of it I’ve incorporated into my own workflow and mindset. You have definitely made me a much better ME. So thank you! Happy New Year 🎉-here’s to a great 2025!
Hey David, Yes, I've showed that excerpt before on the channel! And also the filters and clippers one is something I've shown here before too! (I reference the video where I explore it) I'm happy to hear these videos have been helping others! Happy new year to you too!
Interesting. While TP limiting is nice I feel like it’s driving the limiter too hard when certain genre’s (or people) asking for extremely loud masters in the -6/-5 lufsi range (it actually hits like -3 in the loudest parts). Of course you can control the low end that would typically hit the limiter the hardest to be able to control TPs somewhat, but that takes away from the impact. And TP limiting asks the limiter to work too hard and introduces more distortion that what can pass… what are your thoughts on TP limiting when extreme loudness is requested? Squash some dynamic peaks with saturation to be able to use TP limiting, or forego TP limiting entirely and allow the ISPs to provide some distortion to “mimic” the lost transients. Just curious about your practices
I agree, I feel more freedom in my masters when oversampling at 16x or 32x without TP limiting, especially if you got a lot of layers TP squash a little too much, sometimes for artistic reasons I kinda like how it makes the low-end "like trying to scape but can't type shi'"
NIcholas, I know you are a detail-oriented person but I have a question that might make for a good video in the future. i would like to know your opinion on this: If listeners don't sweat the details, why should we as mixing and mastering engineers sweat the details. I have the opinion that details really matter and every 1/2% of improvement takes away a barrier to listeners so that they can just enjoy the music, but i feel like you might have a stronger argument or opinion on this matter. Especially in terms of the end-listeners vs. your clients.
Oversampling is ineffective: raise the sample rate of the session to avoid pitfalls from stacked AA filters. 50dB is not a null at all, it's still 40dB over the noise floor! AI is frowned on because it sounds like *** farts, it completely destroys the timbre and dynamic envelope leaving nothing buy grey mush, why are you adding grey mush to a mix?
@@redcollard3586 while it may not be perfect it sums back together perfectly. So you’re actually not adding anything by using stem separation. At least with good modern algorithms like RX and spectralayers
oversampling in RX is kinda trash, on the other hand 16x, 32x oversampling in Fab Filter sounds more open and transparent in my opinion, but you gotta be careful with the transients and really test your ears.
Dont you use clipper plugins? If you do then what is the point of true peak limiting? Using the clipper is the same as letting the transients fly past zero
1:54 Here's another video that might give you some insight into why your oversampling is having some effects on your audio: th-cam.com/video/tMzQVOfNVbo/w-d-xo.htmlsi=4VdDNZTFYVvVslUR&t=1510
You've taught me so much Nicholas and perhaps all I can teach you in return is how to say aliasing. So at the risk of sounding patronising, can you say the word alias as in email alias? If so great then I'm assuming you can also say ing as in assuming? Cool then just string the two together. Let me know if that works or to just f*ck off lol
guilty of stem splitting for more than half of the projects for the past 2 years. 💀 If Split EQ can't do it, it's Ozone mothership splitting, if that didnt work as well, Spectralayers all the way. Now Spectralayers 11 you can change or layer the snare and never get caught. Working with low to mid tier in the industry, especially young clients, you gotta do what makes the kids happy, mastering is magical for them and you gotta deliver. Some straight from live recording to mastering, no mixing 🤣🤣All because they know they'll get good result from me. Others dont have grammy mixer to get the track that can truly benefit a standard 1db here and there mastering.
The more convoluted their language, the more they hope to seem credible, as if drowning you in complexity somehow compensates for the fact that there’s no real substance to their ideas.
Did you know you coud achive same or even better results only using a sinlge unit of FF L2? Obviously, true peak deactivated.Thats how I do it, it is actually in the manual 🤓....... I dont like Izotope at all man..!! Good contentc tho!!
AI Stem Mastering? Are you serious? A pile of artifacts! I have been mastering for almost 29 years. Almost 1000 albums mastered and If i heard linear phase artifacts once or twice, that was more than I can remember! BTW I use PSP Neon HR .
It's not by choice or by default, it's just a tool to use and explore for one-off situations. The stems all sum back to a null, and it's not done at the detriment of the record.
New mastering rules? Rules are rigid, mastering is not. Rules are not meant to be broken. Sometimes, breaking rules gets you to a new place you've never been before.
Have you seen how many “updates” there are on this channel? Of course it’s not rigid. You keep learning and adapting along the way. This video is just one instance of adaptation in a whole series of past and future ones😊
I agree broskii, but hey I do believe for some people it might work some of this rules for new genres or experimental techniques, but ngl I love oversampling my stuff I believe it sounds more open, and it makes a good combo when I'm saturating in soft clip pro.
Exactly! That's what my whole career follows. Learn, Unlearn, relearn, rinse repeat. This verbiage and the way I present these ideas in the video isn't intended to lead people towards a DO THIS ONLY ALWAYS AND NOTHING ELSE. It's an open discussion with case-study and experience behind each item. And as always, when I udpate/change things I will flag that here on the channel as I have in the past.
Thanks man, you are great! Wishing you all the best for 2025!
Appreciated! Same to you!
Each time you run a signal through a limiter that has modes (such as Ozone and Pro-L2) it imparts a certain character to the signal of that mode. So having the signal run through that mode multiple times you are imparting that modes character into the signal multiple times on top of each other, that is why it won't null. Just listen (compare the original signal) to how the modes change the character of the original signal.
Exactly
I just watched the video, and the comparison here doesn't really make sense because we're talking about different types of filters. A minimum-phase filter requires less time for processing compared to a linear-phase filter. Additionally, the processing curve in a linear-phase filter is designed so that the so-called filter delay occurs both before and after the actual signal, resulting in what’s known as pre-ringing. In contrast, a minimum-phase filter only produces post-ringing.
Therefore, we are dealing with two different filter latencies, which also depend on the specific filters used for the processing. Performing an exact null test would require knowledge of the filter length and the precise timing of the signal's processing. This would prove to be extremely challenging since we didn’t program the plugin ourselves.
Furthermore, I’d like to point out that oversampling does not inherently destroy the sound. The perceived sound issues often stem from what happens before the oversampling process. In the analog world, we don’t need oversampling because we’re not limited by the Nyquist frequency. Effectively, if we experience distorted sound due to oversampling, it’s usually because the prior signal processing was flawed.
The oversampling process (which includes upsampling, processing, and downsampling) simply mirrors what is standard in the analog domain - working with significantly more sample points. When the signal is subsequently downsampled to the original sample rate, the average value between the sample points is usually calculated. Depending on the changes made to the signal, rounding errors can occur, making the signal appear more stepped.
The key here is how the programmers designed the window size for filter calculation. Generally, there are two possible causes for audible artifacts:
1. The signal was poorly processed before oversampling, making artifacts more apparent at higher resolution.
2. The calculation of the average level over a certain time period introduces rounding errors, which make the artifacts audible.
This is the type of analysis I was looking for🔥👏🏽 thank you!
Compliments of the new year Nick :)
Likewise and to you too my man!
Hey Nicholas. Super interesting discussion here. I love the explorative, testing-based approach you take and your commitment to using your ears first, and analyzers second.
Your comments on linear phase aliasing LPFs is on point. I did an extensive project this year with Sound Theory on their Kraftur plugin. Their designer has his PHD in theoretical physics and really goes deep into things. He chose to design Kraftur's crossovers with minimum-phase filters, which I questioned, and he really took me to school! Our discussion was fascinating and got me to rethink many of the ways I've been working.
His comments really centered around transient smearing, like you illuminated here. He also noted that you can't actually hear the dreaded "phase smearing" at the crossover points with the way they designed the crossovers. Interesting stuff.
Thanks for leading the conversation as we head into 2025. I hope you have a wonderful time at NAMM!
Thanks mate! Appreciate your input! Any links to the project you did on the Kraftur plugin I can look over?
Bro, the stems split technique blows my mind... How the hell i didn´t think of this hahahah. Thanks man !
Got a top tip for you, don't bother with any of it!!
just release unmastered. Make your mix as loud as a referenced master with nothing on the output and I promise you it will get to a better place faster.
I've been releasing unmastered professionally for nearly 10 years now.
Glad to see someone else publicly embracing stem separating in mastering. I felt a little dirty about using it 😂 but I always ensure I use linear phase processing when tweaking. So why the hell not? It’s the future baby!
bang on!
@@panorama_masteringso why do we need a separate person to master when the same person can do the mixing and mastering?
@@Durkhead There’s no rule to say “you must have a separate person do your mastering”. But there’s many reasons you might choose to do so. For example, the mixing engineer may have less objectivity having worked on the track for too long. He may be an excellent mixer but not so much at mastering. Having a 3rd, 4th, 5th pair of ears can help ensure the final result is more objectivity pleasing and translating well across different systems.
Personally I do both more often than not for the convenience of the client.
Minimum phase filters while over sampling is just like adding a low pass yourself at nyquist and still is a form of time dilation though, using minimum on the entire frequency maybe less harmful on a 2-track it can be possibly harmful if you end up doing a stem mastering job. That's why I master at 96 kHz and don't have to worry about filter cramping or compression artifacts.
When it comes to processing that emulates the analog domain by introducing 2nd, 3rd, 5th harmonics, etc, I highly recommend oversampling, but only per plugin instance since intermodulation can occur if you have reflections above nyquist that are also then processed by analogue emulation.
In a mastering context whether you use oversampling or not, or Linear phase or not has everything to do with the workflow you have due to the tools you've purchased. I still think a System 6000 with MD5 and Brickwall2 are the best solution for a mastering studio and only the true peak limiter in Brickwall2 does oversampling. So meh, maybe it's all overrated.
I love the way you teach man; your enthusiasm is contagious and makes me excited to get in the DAW. You're a great teacher and the way you describe/explain things with an open mind so we can follow your train of thought, is a type of teaching that really works for me.
You discover things and share them, and we learn not only these things but the right attitude for moving forwards and continuing to grow. Thankyou for the great vids, u deserve a sick 2025! 🎺🦒
Thanks man! That's super humbling! I'm glad these have connected with you!
You too have an amazing 2025!
I’m no seasoned professional but I never even considered stem splitting to master. I’ve had a couple clients send me stems to master but I never thought to separate them myself. Thats huge!
Enjoy!
Lets go!! thanks for all the indepth info you gave us last year, looking forward to this years videos!. Best wishes by the way!!
My pleasure! Excited for a new big year!
Try PRO-L2 (true peak off) into the TDR Limiter 6 GE but just the last module on which is the true peak limiter.
Happy New Year sir. Looking forward to learning from you again this year and elevate my skills.
HNY to you too! Excited to share more!
Thank you for saving me literally hours of researching random things I’ve also been hearing lately. This is gold.
You're welcome! Glad it was helpful!
Really interesting ! Thanks a lot !
About stem EQing, i've been quite experimenting with it lately with great results and to my actual impression and understanding it always works best with linear phase eq (as the stems are always a bit overlapping and using non-linear phase eq induces phase shifts between the stems).
Curious to read your impressions and experience on the topic!
Cheers and thanks for sharing ;)
Spot on! Using linear phase EQ on split stems keeps the resuming of the waveforms most accurate!
Really great tips! Only thing I'd disagree with is the choice of AI stem split software. Have you tested RX vs others? I did a quick comparison with a mate. I gave him my stem splits from SpectraLayers and he instantly ditched RX after hearing the stems lol. I couldn't believe how bad RX was. Can't be 100% sure which version he was using, so it could've been a poor comparison, but it's one of those processes that's improving tonnes so might be worth investigating.
There was a massive update to the quality of stem splits in RX this year for version 11. I agree, before it was really bad. Since version 11 it’s among the best I’ve heard and I stick to it, since the workflow is great.
Thanks man!
Yeah I've got RX11 too which serves well enough for me.
@@simon_dupp Ah that makes sense then! I guess they're roughly about the same with each improvement. Wouldn't surprise me if they used the same algorithms on the backend.
@@PrincipalAudio If you compare spectral layers 10 to 11 there's also a huge difference in quality. They cracked the non-destructive method of splitting with SL 11 and the quality jumped up massively. I'm pretty sure RX uses very similar tech now too. I've never compared them, but seen as they have also got non-destructive splitting now, I assume it would be the same once put back together.
If there's a negative difference and you are trying to use the split track on it's own it's problematic, however if you are recombining it, any difference between algorythems is negligible providing they are non destructive like all the latest ones. 😉✌️
@@DaftyBoi412 I use spectralayers. It’s the best I’ve heard imo!
Great video and Happy New Year mate
Thanks mate and to you too!
Clippers are great but they are real beasts...
Great video!
Thank you very much!!!!!!!!!
Great 2025 for you and all yours.
YES! And likewise back to you!
What are you using for A.I. Stem splitting?
iZotope RX11
@@panorama_mastering Thank you for replying!!
Which stem splitter are you using to achieve the perfect null for the stems?
I use Studio One and the stem separating is brand new and according to reviews not the best. 😎
iZotope RX
@@LUSH_AUS there's also spectral layers from steinberg, which has non-destructive stem splitting like RX does. If you use cubase or neuendo it seamlessly integrates as it's an. ARA plug and they have built their DAWs specifically to work with it as if it was part of the DAW it's self, it's really neat.
If you use a dif DAW I'd test both to see what one you prefer, as I've never compared them personally so couldn't give an overall recommendation, but if you are already in the steinberg ecosystem it just makes sence, just like if you are already in the Isotopes one and have RX already and don't use Cubase or Nuendo like Nick, the one he's using makes sence. 😉
thanks for all your reseach ! good work !
So you still prefere true peak instead of oversampling on the master limiter in 25, right?
Interesting, because my mastering engineer told me that he always use oversampling on the fabfilter L2 instead of true peak because it sounds more "open" in his opinion.
I can confirm, it sounds more open, sometimes getting too technical with audio engineering limits your ears.
Correct! TP > ovesampling for myself.
Excellent work. Thank you for all you do Bro
Thanks m808!
I would love to see a video from you about various stereo wideners and techniques. The pros and cons and nerdy details wold be great as you deliver really great in depth information. Mainly interested in newer tools like the mastering widener in Acustica Audio Wine and Leapwing Audio StageOneV2.
Good idea! lets get to it!
12:11 because of the data lost in the split and subsequent recombination of the stems... thats why
🔥 Happy New Year
Happy new year to you too!
I'm so glad I watched this video right until the end so I could hear you say "Ass over tits!" Also great to hear about the evolution of clipper first in chain vs clipper after EQ.
Now I remember your past video where Ian Stewart had talked about this, shifting sample points, and your response was to demonstrate how the EQ you were using was not shifting phase very substantially. Of course a HPF is an extreme example. So even with the less extreme example of, perhaps, some gentle broad Q bell shaping, do you still EQ pre-clipper? Why? Why not?
Finally, it is worth noting about clipper design (and most saturators) is that they will typically have an HPF built into the clipper already to remove low frequencies that can make for muddy sounding saturation. Gold Clip, for example, has a 15 Hz HPF I believe. Sometimes this function will simply be referred to as DC Offset Filtering and can be toggled on and off. I didn't know that until I started interviewing guys like Ryan Schwabe.
Cheers mate!
Thanks for watching mate! Have you got a link to the interview with RS?
You should try out TDR limiter when it goes on sale. It has the true peak limiter separate to the limiter as an inbuilt process, and you can put a clipper wherever you want in the chain.
I use it too, as far as my knowledge and experience goes it is a brilliant plugin. 😎
@kadiummusic I love it, took a lot of manual reading to get results, but it’s really great
I bought it during Black Friday, this plugin is really nice. The HF limiter, in relative is amazing 🤩
You boys try limitless DMG
Is this TDR Limiter related to the Limiter N⁰6? Because I was unabled to find anything about it in the TDR website.
Beautiful. Have a great 2025!
You too! Happy New Year!
That AI stem splitting is actually done through spectral analysis useually afaik. It's super cool stuff, and I had a bunch of fun playing about with it in the spectral layers .ARA plugin.
The fact you can now split it non-destructively is super cool and only really been able to be done in the last year or two! The previous couple of version of that tech it would deteriorate the recombined stems, which made it hard to justify using, but now it can be split without any degredation to overall quality it's super useful for small tweaks on indavidual elements.
Interesting, never looked into HOW it's done before.
Do you know of any resources I can read over?
@panorama_mastering I'll have a dig about m8. I'm mostly going off of steinberg info on it as they have it as part of their spectral layers program, which only deals with spectral processing, and when you seperate each track it pulls out all the harmonics via AI spectrally And that side of RX appears to work very similar.
I havn't gone too deep into it myself either tbf, but will see what info steinberg has on it, as they seem to be pioneering it, or at least up there with some others at the forefront. 😉
Have you done that 180-degree phase shift into mid/side encoding thing in another video? Because I know I’ve seen that phenomenon before-and pretty recently too-since I knew exactly what was going to happen. I’m a real “lose it if I don’t use it” type of person, so it must have been fairly fresh in my mind. 😂
Actually, now that I think about it, a lot of these things seem familiar. 😊 You must be recapping some stuff, eh? Either way, it’s probably good to hear it more than once. I’ve learned so much from this channel over the years, and so much of it I’ve incorporated into my own workflow and mindset. You have definitely made me a much better ME.
So thank you! Happy New Year 🎉-here’s to a great 2025!
Hey David,
Yes, I've showed that excerpt before on the channel!
And also the filters and clippers one is something I've shown here before too! (I reference the video where I explore it)
I'm happy to hear these videos have been helping others!
Happy new year to you too!
@ Yeah. I remember now. I just wasn’t sure at first. And I need to comment something so. lol. 😂
Interesting. While TP limiting is nice I feel like it’s driving the limiter too hard when certain genre’s (or people) asking for extremely loud masters in the -6/-5 lufsi range (it actually hits like -3 in the loudest parts). Of course you can control the low end that would typically hit the limiter the hardest to be able to control TPs somewhat, but that takes away from the impact. And TP limiting asks the limiter to work too hard and introduces more distortion that what can pass… what are your thoughts on TP limiting when extreme loudness is requested? Squash some dynamic peaks with saturation to be able to use TP limiting, or forego TP limiting entirely and allow the ISPs to provide some distortion to “mimic” the lost transients. Just curious about your practices
I agree, I feel more freedom in my masters when oversampling at 16x or 32x without TP limiting, especially if you got a lot of layers TP squash a little too much, sometimes for artistic reasons I kinda like how it makes the low-end "like trying to scape but can't type shi'"
Does having 2 IRC's active in this process effect the sound at all? If so what do you do about it?
NIcholas, I know you are a detail-oriented person but I have a question that might make for a good video in the future. i would like to know your opinion on this: If listeners don't sweat the details, why should we as mixing and mastering engineers sweat the details. I have the opinion that details really matter and every 1/2% of improvement takes away a barrier to listeners so that they can just enjoy the music, but i feel like you might have a stronger argument or opinion on this matter. Especially in terms of the end-listeners vs. your clients.
Defintiely. Good Question! Consider it done !
Thanks a lot ❤
My pleasure!
PRONOUNCED - ay-le-ay-sing
Oversampling is ineffective: raise the sample rate of the session to avoid pitfalls from stacked AA filters. 50dB is not a null at all, it's still 40dB over the noise floor! AI is frowned on because it sounds like *** farts, it completely destroys the timbre and dynamic envelope leaving nothing buy grey mush, why are you adding grey mush to a mix?
@@redcollard3586 while it may not be perfect it sums back together perfectly. So you’re actually not adding anything by using stem separation. At least with good modern algorithms like RX and spectralayers
oversampling in RX is kinda trash, on the other hand 16x, 32x oversampling in Fab Filter sounds more open and transparent in my opinion, but you gotta be careful with the transients and really test your ears.
Definitely; ultimately, we should aim to avoid un-necessary processing or filtering that has no intent for the given result!
We really overthink audio huh. lol
Keep pushing every day!
@@panorama_mastering What is the average time it takes you to master a song?
Shoshin - forever student
@@cainomusic Just doesn't seem very musical.
30 minutes to 90 minutes
Great video
Cheers mate!
Quick question: Does the 'True Peak Only ' final Maximizer have to be in the same IRC Mode?
Why limit yourself? Mix and Master Unlimited
This shift should have happened back when the loudness war ended
ain't gon' lie at some point in the future music will be limitless and 32 bit as standard, just imagine that for a moment.
@@shakysavy7980 that’s the future I want to live in…
Yeah... I waited for this 🎉
I hope it delivered :)
@panorama_mastering absolutely! 🙏🏼
Dont you use clipper plugins? If you do then what is the point of true peak limiting? Using the clipper is the same as letting the transients fly past zero
It's all about ballistics and distortions. Same result different effects. I prefer TP limiters to avoid distortions at the very end.
@panorama_mastering so to avoid distortions you use something that adds distortion, makes sense
What happens if I have 5 tracks running into a bus and I clip each of the 5 tracks individually and then use an eq on the bus? Am I going to hell?
@@yurikalashnikov2460 you’re buying a one way ticket to peak city 👀
@@yurikalashnikov2460 peak control should always happen after phase distortion if you want to benefit from the reduction in crest factor
Then youv just emulated a clipping plugin thats all
1:54 Here's another video that might give you some insight into why your oversampling is having some effects on your audio: th-cam.com/video/tMzQVOfNVbo/w-d-xo.htmlsi=4VdDNZTFYVvVslUR&t=1510
Amazing thanks for sharing! I'll give it a good look over!
@ you know it’s gonna be good when it’s narrated by Dan Worrall.
You've taught me so much Nicholas and perhaps all I can teach you in return is how to say aliasing. So at the risk of sounding patronising, can you say the word alias as in email alias? If so great then I'm assuming you can also say ing as in assuming? Cool then just string the two together. Let me know if that works or to just f*ck off lol
gosh i love u
IN 2025 I WANT TO SCORE YOU 80% ..NICE START
guilty of stem splitting for more than half of the projects for the past 2 years. 💀 If Split EQ can't do it, it's Ozone mothership splitting, if that didnt work as well, Spectralayers all the way. Now Spectralayers 11 you can change or layer the snare and never get caught. Working with low to mid tier in the industry, especially young clients, you gotta do what makes the kids happy, mastering is magical for them and you gotta deliver. Some straight from live recording to mastering, no mixing 🤣🤣All because they know they'll get good result from me. Others dont have grammy mixer to get the track that can truly benefit a standard 1db here and there mastering.
The more convoluted their language, the more they hope to seem credible, as if drowning you in complexity somehow compensates for the fact that there’s no real substance to their ideas.
are you saying he's overcomplicating what he's explaining? If so, skill issue on your part.
Did you know you coud achive same or even better results only using a sinlge unit of FF L2? Obviously, true peak deactivated.Thats how I do it, it is actually in the manual 🤓....... I dont like Izotope at all man..!! Good contentc tho!!
AI Stem Mastering? Are you serious? A pile of artifacts! I have been mastering for almost 29 years. Almost 1000 albums mastered and If i heard linear phase artifacts once or twice, that was more than I can remember! BTW I use PSP Neon HR .
It's not by choice or by default, it's just a tool to use and explore for one-off situations. The stems all sum back to a null, and it's not done at the detriment of the record.
New mastering rules? Rules are rigid, mastering is not. Rules are not meant to be broken. Sometimes, breaking rules gets you to a new place you've never been before.
Have you seen how many “updates” there are on this channel? Of course it’s not rigid. You keep learning and adapting along the way. This video is just one instance of adaptation in a whole series of past and future ones😊
@@SonicWizards Then why use the words "New mastering rules (!) I'm following in 2025"? Words matter, and it gives the wrong impression to newbies.
I agree broskii, but hey I do believe for some people it might work some of this rules for new genres or experimental techniques, but ngl I love oversampling my stuff I believe it sounds more open, and it makes a good combo when I'm saturating in soft clip pro.
Exactly! That's what my whole career follows. Learn, Unlearn, relearn, rinse repeat.
This verbiage and the way I present these ideas in the video isn't intended to lead people towards a DO THIS ONLY ALWAYS AND NOTHING ELSE.
It's an open discussion with case-study and experience behind each item.
And as always, when I udpate/change things I will flag that here on the channel as I have in the past.
This is what makes you the Michael Jordan of mastering. It's in the details
Thanks amte! Appreciate that!