A.I. Mastering VS Professional Engineer
Вставка
- Опубліковано 21 сер 2024
- ► Hire me to work on your music: whiteseastudio...
►Thomann affiliate: whiteseastudio...
►Sweetwater affiliate: whiteseastudio...
►Plugin Boutique Affiliate: whiteseastudio...
►Become a member: / @whiteseastudio
Wanna see me take another shot at this mastering job in 2023? Then check out this (members only) video: ua-cam.com/video/MPeppo7vU2Q/v-deo.html
I engineered for 15-years before “Mastering” meant anything other than getting a piece of music from a tape to an acetate for pressing vinyl records.
Later it was expanded to mean getting all the codes and everything into the digital files to be sent to the CD manufacturer. But at that time it was discovered that the recordings could be “enhanced” at the same time.
Then, vinyl and CDs became only a tiny fraction of music recordings consumed by the public, so only the “enhancing” part was left remaining.
However, the loudness wars and streaming services came along and introduced some of their own new requirements, which is exactly what White Sea Studio works with. And that’s valid and has to be done.
…but the “enhancements”? Are they necessary? Maybe. Maybe not.
Are they subjective? Always.
Today’s digital technology enables anyone with only introductory level skills to produce mixes that sound as good as the average PROFESSIONAL mix from back in the 60’s, 70’s, and 80’s.
A lot of the best sounding records EVER MADE were made during those years, but only by the best people. The AVERAGE record back then sounded pretty weak by today’s standards.
Since it’s now so easy to get a “decent” mix, and todays greatest are producing records that often sound SO incredible, the “enhancing” that these mastering engineers are doing IS needed.
But that alone doesn’t equate to “Mastering” if you’re not also correctly providing all the different masters for all the different audio platforms and formats.
So… perhaps what we need is a NEW TERM altogether.
FINISHING ENGINEER.
If so many “mixers” deliver masters that still require so much more subjective audio work, then this new stage of work needs to be identified and NAMED.
RECORDING Engineer - Records the musicians.
MIXING Engineer - Mixes the multi-track recordings.
FINISHING Engineer - Does all these audible adjustments to the stereo mix to improve it.
MASTERING Engineer - Performs all the technical tasks needed to prepare the finished sound recording for all the various end-listener formats.
Recorded by:
Mixed by:
Finished by:
Mastered by:
Not hard. 👍👍
Oh yeah the Bandlab feature has been there for years. They have 4 modes to choose from for your master so I don’t know which one they used. I use it when I want to quickly master something after production. It’s good but nowhere better than ozone 10 or an actual mastering engineer imo.
When I started mixing, the mastering engineer was the individual that took my mix and cut the acetate master for the pressing plant. Their job was [as far as I was concerned] to get the mix that I made and the client approved, as accurately as possible on the released vinyl disk. Their job was NOT to make creative changes. These daze, mastering has become some kind of voodoo that is supposed to make a supplied mix better but all that is necessary is to comply with the necessary release standards for distribution. Mastering is NOT magic it's technology and therefore the AI apocalypse is upon us. When was the last time you talked to a telephone switchboard operator?
EXACTLY.
I engineered for 15-years before “Mastering” meant anything other than getting a piece of music from a tape to an acetate.
Later it was expanded to mean getting all the codes and everything into the digital files to be sent to the CD manufacturer. But at that time it was discovered that the recordings could be “enhanced” at the same time.
Then, vinyl and CDs became only a tiny fraction of music recordings consumed by the public, so only the “enhancing” part was left remaining.
However, the loudness wars and streaming services came along and introduced some of their own new requirements, which is exactly what White Sea Studio works with. And that’s valid and has to be done.
…but the “enhancements”? Are they necessary? Maybe. Maybe not.
Are they subjective? Always.
Today’s digital technology enables anyone with only introductory level skills to produce mixes that sound as good as the average PROFESSIONAL mix from back in the 60’s, 70’s, and 80’s.
A lot of the best sounding records EVER MADE were made during those years, but only by the best people. The AVERAGE record back then sounded pretty weak by today’s standards.
Since it’s now so easy to get a “decent” mix, and todays greatest are producing records that often sound SO incredible, the “enhancing” that these mastering engineers are doing IS needed.
But that alone doesn’t equate to “Mastering” if you’re not also correctly providing all the different masters for all the different audio platforms and formats.
So… perhaps what we need is a NEW TERM altogether.
FINISHING ENGINEER.
If so many “mixers” deliver masters that still require so much more subjective audio work, then this new stage of work needs to be identified and NAMED.
RECORDING Engineer - Records the musicians.
MIXING Engineer - Mixes the multi-track recordings.
FINISHING Engineer - Does all these audible adjustments to the stereo mix to improve it.
MASTERING Engineer - Performs all the technical tasks needed to prepare the finished sound recording for all the various end-listener formats.
Recorded by:
Mixed by:
Finished by:
Mastered by:
Not hard. 👍👍
@@danepaulstewart8464 I have, for decades now, done it ALL. Mostly 'cause I couldn't, didn't want to, or just too stubborn to, hire a "mastering" engineer. When I've had a client insist, I found the final product to be different but not better, than the master I supplied. Then again, I do have a lot more training and experience than most.
it's so strange. If you focus on the snare than the premaster sounds so much better than all the other versions.
Thanks again for this video. It's true what you said. We're confusing taste with quality. So what the modern taste is, becomes quality. Like this loudness war. More loud is better... What is a big mistake in my opinion. I guess an AI for music mastering is trained with the sound of "successful" music. But it's true that successful music sounds in general better? I think not.
Everyone wants to sound like everyone, when all sound the same it gets really easy to be unique 😊😊
Hahaha! That is the best thumbnail you've ever published. Totally unexpected on my home page.
In my opinion: ai is already good enough - and it will get better. Enjoy being human while you still can, because we are well on the way of changing the meaning of being human (Zizek). It will be delayed for about 5-10 years due to the collapse of globalization in its current post Bretton Woods form (Zeihan). Music was the best thing that ever happened to me.
Yours is the only masterful mastering, yours has both presence and serious body with both having their own space in the master.
Wonderful! Let's stay in touch.
I actually like the way the AI mastering sounds. I think the added midrange makes the low-end sound tighter. I just don’t get the whole -17 LUFS with zero head room. That sounds more like a gained up mix. Yours is a least hitting the appropriate levels. It’s only draw back is it kinda just sounds like a louder version of the mix. That’s an issue with mastering your own mixes. How do you improve on something that is supposedly as good as you could get it already? 😮
Great video. I’ve used Bandlab to master mixes for my band. One thing area it falls short is that it outputs at 16bit regardless of whether your mixes are 24bit or not. So the depth is compromised. Additionally, a professional master would definitely be the route we would take if we wanted to master for vinyl.
Do you really hear a difference between 16 and 24 bit .wav? I can hear .mp3 compression, though it becomes hard at 320kbps, depending on the mix. FLAC I can already not distinguish, and from what I know 24bit only allows for quieter sounds to be encoded (like up to minus 100 something dB) and that really shouldn't matter because 16bit is already enough.
I personally preferred your master. I feel like the other one sounded like what Ozone loads as a starting point with master assistant, but then you don't have the option to hone in those settings to serve the song. Not saying it was bad, but I agree that some of the choices it made didn't serve this song best. That being said if budget is an issue then AI mastering is a great tool. The great thing about services like this is they really do help people release music. 👍
I bought the Ozone 10 elements for $30. Just ran it as is. It chose a preset configuration curve for me. Played around with other presets and settled for one.The results were as good as any « real » Mastering I could get. In a minute. And you can easily propose 3 or 4 versions to your client for not much hassle. None of my mix Masterings sounded tamed or over limited like our friend did.
Here is the order of mastering:
1st place = real engineer
2nd = Aria or LANDR for electronic music
3rd Waves automated mastering is very very good.
My tips would be to provide Waves/LANDR with reference track to get what you really want.
LANDR/Waves is like sugary cake. Aria is like a wholesome organic meal.. they have their moments. Band lab is not very good (although it is free), Cloudbounce is pretty good and you can download a desktop app and provid reference files. Ozone 11 assistant is too bright as a starting point so Ozone 10 wins on that front although clarity is an excellent tool
Instagram is REALLY bad. That is actually an audible difference that everyone can tell. UA-cam is good enough and while there is a difference, it's similar to the difference between high quality mp3 and wav - there is a difference but most can't tell in a comparison and without a comparison nobody knows whether it's mp3 or wav just from listening to the track. *edit* by the way: If ANYONE knows what settings one should use for instagram (in terms of loudness and peaks) - do tell! It sometimes completely messes up very fast transients. It sounds like it's going through a phaser - (no matter where I peak and no matter whether it's TP or regular P btw.)
The thing about professional mastering is that I can barely hear the difference between it and my mixes. You can make fun of me all you want for that, but it's the way it is. So if mastering only cost 80-100 dollars, I would say that's ok, and just get it done. But it's a ton of money. For those of us who have been listening and playing to music for 30 years but can't really appreciate the difference, it's only the rich people who are going to throw down 1000 dollars for a human, when we can throw down a few bucks for AI. At the end of the day, the Rolling Stones mono version of Satisfaction from 1965 sounds like crap from a technical standpoint, but still gets the blood going as a tune. That's the most important thing: the tune, not the absolute pristine nature of the hihat at 15k. So I'd spend my 1000 dollars on a great guitar before a mastering engineer. Sorry Wytse! You're still going to have a great career as a mixing engineer and advice-giver. AI can't do those things.
I've tried bandlab masters and they always sound sharp, harsh to me for some reason, except the fire one and that sounds dull haha. Maybe I should mix the two together....
From a subjective perspective, I prefer your mix, regardless of the faults you've identified (tastes change over time). However I'm really impressed by the AI service.
The hardest thing is to master really loud without damaging the music. -14lufs. -12 is waaaaaaay more easier
im honestly confused by the loudness test, only because any time i've ever ran a Bandlab master through Loudness Penalty it has shown it as being turned down anywhere from 3 to 5db?
the vocals are more up front in the bandlab master over yt which i like
Ok I am not in the best listening environment ( apple earbuds) but sadly I find the bandlab master much punchier but why on earth have it at -17LUFS as you said that doesn't make any sense for streaming
You've squash it. Expander with short release helps.
Thank you. This is such an important and honest post. For me, a robot lacking emotional or aesthetic judgement, or any preference for a single instrument level won the day using play-safe algorithms - and if its levels had been calibrated for Spotify, it would have been hard to beat. For me, the human master sounded too processed and lacked depth - it seemed to be fighting against itself. It was a great test and raises many questions. Being from an older generation I have witnessed so many classic albums ruined by dire mastering. Mastering is an incredibly important stage of a production and its critical to get it right. The robots are going to provide excellent points of reference - but in this area (unlike computerized chess), perhaps we can do better than machines - I hope so.
It seems like someone doesn't understand what a robot actually is. 🤣🤣🤣No, a "robot" didn't master the track.
So A.I. Mastering isn't a robot? 😆What would you like to call a set of algorithms applying themselves remotely to a soundtrack? A poached egg?@@Jazzguitar00
TBH, and I am sorry to say so, but #1 for me is Bandlab, Premaster as #2. The main reason is openness and not hearing the compression on the snare and on the voice as we do on your Mastering. It’s just an opinion of course.
4:00 Your mix sound alive and way more warm and natural. The other one sound thigt and sterile and cold. It's a matter of taste. i like it raw and natural.
The universal setting is best on BandLab mastering, imho.
Mind blown, you confused me while im sure that was not your intention lol Mastering is somekind of magic thats hard to put into words I guess.😁Facts vs feel, feel vs facts ahhhh??😊
I prefer the bandlab master for more perceived clarity and openness (basically more high frequencies) but the facts it’s only -17 lufs is crazy to me
I think it's better for the music, that it does not squeeze it that much
@ -14 lufs isn’t squeezing imo. My masters are about -7 lufs and I’m barely using compression/limiting on them
@@danzo9528 you are right. His master was good according to specifications, but he brickwalled the dynamics a little bit too hard, as he recognised it now. He knows what is the issue, for sure :)
I have seen many big names uploading very quite music and completely trusting UA-cam to manage the normalization. I don't know if you can extract that information from Spotify .
Its all described in TD1008 from the AES
In my MDR-7506, the White Sea Studio version sounds more muddy.
I always add a db of bass back into my ozone masters
Seriously... my preference goes to the unmasterded one... but again.. it's just a question of taste ;-)
Sound good to have different version
Actually I am shocked to say this, but I like the BL master better, oddly it seems more dynamic and punchy with more definition then your version. Louder is not better, too many songs are just mastered loud and have lost there dynamics and lack depth and detail, and to me just sound really bad when listened too on good gear. That's my two cents on this subject.
tbh that sloppy low end sounds like it hits harder b/c like you said... it's gritty
None A.I. can replace humans. Because inspiration and creativity have never come from intelligence anyway.
Premaster sounds best to my ears ;)
De jouwe smaakt beter. Maar dat is mijn smaak dan weer. Je hoort de ervaring met analog gear, klinkt minder zoutloos en meer peper? Keurig
Can technical oprimizations replace human emotions?
Loudness from production to mix to master can confuse the most experienced engineers nowadays.
Would be great to have a YT-short just to show your old master and new master in comparison
Watch the clip to zero method on UA-cam. No matter how experienced you are it will teach you the most transparent way of achieving loudness. It’s used in electronic styles but the method can be applied to any genre. I went from purist mixer, to modern loud mixer in literally 2 days
Cut out tinyness sound good mix
*** QUESTION: Should I get a Nuemann NDH 20 or Beyerdynamics DT 770 250 ohm with Scarlett Solo Gen 4 upgrade? (Or do you recommend something different?)
bandlab sounds as is supposed to be imo. Your one is not far from the premaster but too much low midrange decrease the overall quality and vibe and does some bury to overall mix...In other words hi end in Bandlab master is perfect, that's why it sounds more balanced
I wonder how well the bandlab file is mastered, in the end of the day, its a question of taste. Both are good, I like your low end work more.
AI got a little clarity & upfront on the vocals...similar to the pre-master version.
Sound good
As you demonstrated with the piano thing in the low end you wanted more of that in the end product and I think this is the reason why the craft of Mixing and Mastering will never be replaced by AI.
Never say never. Five years ago we thought stem separation would be impossible
After the stem separation by the same programs, giving you such options'd be quite easy.
(OFF TOPIC!)
Im scanning for sales and I come across devices that now offer
“AITO GAIN” as a new feature.
Dude. You shook up the industry with your rants!
Obviously, not too many companies cared about it in the past.
(New popular Audio interfaces now feature “auto gain”.) ?
Because before you they were like (auto gain for what? Scratch that idea.)
Criticizing illogical acts is a sign of upward intelligence!
Slow! But they’re finally catching up.
Get em rude!
I'm all for new tech esp as in the past I've had a Mastering engineer just set up his chain and didn't seem to take into account the dynamic contrast in my song. It leaves one asking "Why did I pay for this?"
Though I'm not ready to discard a human anytime soon!
Speaking of Lewitt. I am considering to get connect6 but i've heard the headphone output is heavy on a low end with some headphenes. Does any one know how would it work with ath50x?
Are you having any problems with reaper and your new mac
raw sounds the best by far. lesson of the day: don't master your tracks.
I like your vocals more than Bandlab.
Yep, your master killed the important and valuable transients for this track...
I think the audio comparison is moot. If someone wants to employ a human they will for many other reasons than the final audio alone. And if it's not worth a human master then it's not, it's either good enough or it isn't. Same reason as all technology shifts in history. The only people who argue over 1db here and there are other engineers, not the actual audience that will determine it's success.
Through the shit youtube audio the band lab sounds a little brighter
LOL, I got everything wrong in the Lewitt video.
I understand your position, but I could not hear the emphesis you intended me, the listener to hear in your mastered version. Honestly, I enjoyed the A.I version a lot.
first time hearing and the bandlab sounds the worst actually, the premaster sounds better then bandlab Master lol I still think yours sounds the best!
the good that if i make a stateent about my taste (per example dislike of linearphase stuff) that i always label it as oppinion. basically i do this every single time. good also that i sometimes even encourage people to have other taste than mine. (by saying: "if you have other taste, then all good").
That way i can say the things i find important, without negative sideffects ..
further, to imply the presence of factual things in Audio almost sounds a bit weird. I mean if you would have asked me, there is no study that i know of that prooves about what sounds good, or not..
Alltho im pretty shure some things might be preffered by a majority of peoples, how would this common taste be "a prooven right thing"..? you would have to connect alot of theories, like proove of swarmintelligence, and to make it actual proove you would need the proove of unintelligence of loners.. (that way you separating correlation from causation)-
So, i dunno what facts you are talking about then,, factual we could only call measurable things then. like lufs, readings, and so forth.
or perhaps things that not even the last person likes.. like in music playing the most out of key notes there can be. such a thing would be close to prooven bad, because not ebven one in a million people would like "out of key only" music
so, do you have a single example of something you have talked about, that is prooven in a study?. i mean scientifical proove?
((oh i remeber one, that most people preffere higher levels))
Or certain things wich almost nobody likes like say clipping the master 10 db,,lol,, i would even without proove say, well, almost no person in the world likes that sound.. Things like broken algorythims too, wich output a hefty cliped sound.. well, thats retty save to say its factually bad..
all im saying, : waht fact are you talking about? that separates you so strongly from people? i dunno what you mean really..
(probably ive not seen the studies about all those prooven facts.. so , enlighten me if you have studies on hand).
see, even what we apprechiate today a bit more, like say a lise master, would probably caused bad feelings around the early 2000s.
or loads of reverb, some like that, some are pretty bored from it.. im pretty shure dividation might be measurable if we would perform brainscans.. some would have a different "good thing". for some its a good thing to have extra-reverb for some its not.
or the amount of highend in a song..
the taste changes over time..
hey probably you didnt adress me, you just said "the people" talk about oppinions.. i just said what i said incase you ment me.
Look, all im saying: probably you are not unbiased.. of are you god, ?
and probably everyone has a slight amount of bias.. or i love to call it taste..
and you are not necessarly just free of it,. so,, you are similar like "the people".
NO AI IN MUSIC!!!!!!!!!!! ONLY OUR BRAIN AND HEART!!!!! EVER!!!!!
Paintings music movies. Everything will be made by ai in the future and the content will be produced specifically for each user account based on their preferences.
anyone can tell me what kind of loudness analyzer / software Wytse is using at 6:11 min? been looking for that kind of standalone for post comparing . any hints are highly appreciated :-)
Its part of the SWS extension for REAPER
TY Wytse 🖤@@Whiteseastudio
Respect for your genuine honesty when doing the LFUS comparison - brilliant 👍. The -14 LFUS debate has been raging on for years now - some producers say ignore it whilst others say stick to it. Be great to know your detailed view and analysis of it in maybe a future video (hint 🙂). BTW for me you proved AI can't 'feel' the emotion within a track hence why in this case it crushed the life out the transients at the most crucial moment of the track.
No one says to stick with it. I have not seen a single professional engineer say that they aim for anything as low as -14. Random UA-camrs? Sure, they say all kinds of stuff based on little bits they've read but it has no value in the real world.
@@Jazzguitar00 eh? That's not true to say 'nobody says stick to it' . Your comment doesn't offer anything here.
Massive push for Bandlab and the AI movmrent while us humans are forced to lower prices to make a living ...
Your opinion sounds a bit biased , you should have made a recent master instead of 3 years old master in your review
Confusion is the sparkling awareness of wisdom.
Other tinnyer
All studios is same
ahahahahah -17 LUFS wih true peak 0.0 db from an A.I. Mastering service is a shame!
It’s a shame that in todays day and age you have to explain and break down why certain decisions were made in the mixing and mastering processes because all people expect are very clean very loud mixes and masters this is the reason there is no character in music anymore, you literally have to explain concepts like “grit” and “dirty” to people and it’s annoying af I signed up for the Derek Ali engineears platform and was told by their verification team that I didn’t get an A level verification to allow me to be exposed to labels because my mixing wasn’t “clean” enough
There is a plugin that is the best at this and it is not osone
What's the plugin?
@@TheMrVogue not saying
AI better in my opinion
What matters most is which master sounds the best. Your master for my ears gets a prominent 3rd place. Bronze is nothing to be ashamed of. Think of it, as you were the chosen one to do a human master. Who cares that it lacks the "I" in the AI
Yuck, I do not like the AI one at all. Prefer the unmastered one to that. It's too tingy on the high end, and thin on the low, overcompressed as well.
I don't trust your mastering skills, you always make even sound like Sandstorms
Comment for da algorithm
Streak count: 180
0:36 If you're going to include that, please give a warning beforehand. If I was about to eat as I casually browsed UA-cam about music production, I would've stopped it for the entire day. OMG.
Wonderful! Let's stay in touch.