To quote the great JPEGMAFIA: "why would I pay someone $300 to make my song sound like shit when I'm perfectly capable of making it sound like shit on my own"
I agree... this comment section is filled with anecdotes like "I don't know anything about AI so I tried using it, and it made crappy-sounding music." If I said "I don't know anything about multiband compression, so I tried using a multiband compressor and it made my music sound like crap," the same people would have critical notes.
@@stickyfox I actually tried to get set up to use the Comfy AI interface and got lost with all the virtual Linux machine stuff. The base "make this sound like this" seems to be hit or miss but I'd love to dig in with more control for sure. Just gotta learn how to code, I guess LOL.
@@cutaia If the kernel is the same on those then he was correct to cut them out before the study. If not, then perhaps he should have set up a submatrix instead of just discarding them outright . I'll also be blunt and say that no one really believes AI is competing in this space with humans. It is more apt to say that people are using AI where they otherwise would not have hired a human anyways. AI (not just in this instance) is much like Harbor Freight in that regard.
@@cutaia I'm going to bow out here. I was just happy to see someone on youtube even approaching the level of rigor I used to use when I was a professional production biochemist. I'm not sure if I stepped into something you have against this guy because to be honest this is the first time he's been suggested to me. Take care, and feel free to attempt to replicate his results.
@@cutaiano, the method isn't "obviously flawed". The results presented do not cover the eliminated mastering tools. His personal bias does not affect the presented data. All that step you're shouting about makes the study less exhaustive. That's it. It's not poor methodology because it was not a part of the method. What it is a reduction in scope. Given what absolute gibberish machine learning can come up with, it's an entirely unsurprising event. This limitation of scope does not affect the results that were presented - and the results show that the best AI mastering tool does a comparable job to a highly regarded expert on a time constraint.
63 yo recording/mixing engineer here…. Done tons of records in all kinds of genres. Just a reminder that initially mastering was a necessary step in the production chain. Mainly it was to make sure the final mix would be printable on a vinyl record. It was a very precise job to do and it was a very restrictive process as the margins of error were pretty tight. Those days are gone and mastering is certainly not a necessity anymore. As long as your levels match the levels required by either Apple, Spotify, or any other platform you will be fine. Those levels are easily attainable with any plugin that do just that. Couple of years ago I got Ozone 10 for under $100. Tried it just for the fun of it. Used the presets and voilà… Very impressive. My clients are pretty happy. Unless you produce a very prolific artist in the higher spheres of the music industry, spending money on a mastering for a digital only release doesn’t make sense anymore.
Well said! I make music for fun. For me, mastering is adding that little touch at the end. Just a bit of EQ and light compression adds that final touch. There is no need to spend a bunch of time and money on that when some free VSTs do the job. In general I prefer hardware. So I plan on moving my mastering to hardware.
@@jennoscura2381 Funny because my path is the other way around… I have all the real gear… Tubetech, Elysia Alpha, Neve Eq, Manley, API, etc, name it I have got it… And honestly, since my clients always come back for « can you +1 dB this and that, lower - 1dB this and that » or just a bit of this and that and that’s final I promise… I resorted to all in the box without them knowing it and they are happier because « yes I can do all the little useless tweaks » that will make you happy… It’s possible that if I were to compare the same processing paths with hardware I would feel the difference… But I am done with that because it’s so much time consuming that it’s not worth it. But for you, an independent producer who works at his pace for his own projects, it’s pretty fun turning the buttons for real and smell the warming tubes… 👍🏼
@craigwillms61 I agree that a competent engineer is the key factor. My point is more about what tools are involved in the process. I am no longer convinced that analog gear is superior to top digital tools for mastering purposes. I suppose it is a very subjective debate... Cheers !
@@robertl.6919 I reread what you said and deleted my comment. Also, every time I click reply I tell myself I've got to stop doing this, no one cares what I think. I'm still working on that, obviously
@@craigwillms61 😁 Haha ! Well you could also say « who cares about what anybody else says in these forums »… it’s our choice to comment in these forums. At some point, all opinions are as valuable as anyone else’s unless a well known expert joins the conversation but that rarely happens. We talk, share opinions and experiences. That’s all there is to it.
3:25 about a decade ago I was friends with this metal band that made great music, they saved up a bunch of money to get their album mastered and chose someone who specialized in COUNTRY MUSIC, because they were a "big name". it came out sounding terrible. picking the right mastering person matters.
@@andreviana4418 Nah, that was all Lars and James. They're the ones who kept turning down Jason's bass tracks. Much like Death Magnetic, it arrived to the mastering engineer pre-fucked.
And give them credit once you do... Friend of mine mastered the demo of a three piece from the uk, then they got signed, released an album which went nr. 1 I think, with some of those tracks on it as they were on demo, without giving any credit to my mate. Contracts matter... Band not at fault though btw. Blame Jamie... xx
I love the "putting on boxing gloves to build Lego" analogy for the obsession people have with plaintext prompting interfaces. Prompting in plain English is just the worst way to accomplish basically any computing task and also takes a lot more processing power to work than a traditional UI.
@@jeffkeeling8917 I, too, was one of the 472 and would love to know which was which. I'm kind of surprised the average ratings were all relatively tight. I had one 10, one 8, one 7, and the rest were all quite poor (1-3).
I mastered my new album myself, and tested a few of these AI mastering plugings / websites as I was curious. Once volume matched, literally every AI mastered track was worse than my mixdown.
Yeah I suspect all a lot of them do is make shit louder. Well, we all have a brickwall limiter somewhere in our toolkit (99% of the time theres one in your daws bucket of default plugins) so we can do that ourselves, and usually wreck all the transients in the process.
I had isse with mastering engineer as well... Where we just wasn't vibing... And paying different people to see we are "matching" sound wise just became so expensive that I'm now happy with whatever commes out at the end out of my Daw 😅
I used Ozone quite a lot in the past, and my mixes were super loud and brickwalled. But when I used it I caught myself mixing "for mastering", in other words I was making final mixes with Ozone on master to make the result transferrable (headphones, car stereo, phone loudspeaker) and louder, and not to make mix better. Now I use EQ + Multiband Comp + Limiter on master. I don't think my music is louder now, but it sounds tighter and I have better control over mix (I still think I could use new Ozone features, or at least it's "maximizer", or whatever it called nowadays, though)
Fun trick to get your band some more attention at shows: make a instrumental, sort of 'chill' or 'minimalisted' mix of your set, and not the songs in their entirety, just like intro verse chorus outro, or verse bridge chorus, or intro bridge outro, or intro solo. Have the club sprinkle those tracks in with the house music, but only loud enough that you can hear it behind conversation while not being attention grabbing. By the time you go on stage, people will be used to hearing your songs but wont be able to place them ;)
If this happened to me I'd wonder who the ***sholes are playing music so low I can hardly hear it. That's what Christian Soccer moms do when driving. You will not win anyone over this way. Music is meant to be heard, not played so quietly you "can't place it"
Mastering is like putting varnish on an oil painting. If you use the wrong one or apply it incorrectly the painting is ruined. Same with using AI. Very nice and informative video, keep it up benn!
@@henryglennon3864 I've heard the analogy "sonic varnish" before refering to layers of subtle saturation, but it also works really well when describing what a mastering engineer should be doing most of the time too (not nessisarilly adding layers of subtle saturation, but just adding the sonic varnish to a track, in a more generalised way). :)
Good mastering is also concerned with the presentation or to continue your analogy "the frame" and environment in which the format of music is going to be experienced. i.e. vinyl mastering is a whole different realm vs streaming, or for that matter vs cinema and vs radio.
@@rottingsunrecords666and you both just made me double take then correct the spelling and realise no, not Dillinjah, Dillinger .... That would have been crazy😂
@@tezeta3725 yes! one of my local venues has old concert posters all over the walls and i was floored when i saw one with Dillinger and Flash Bulb. i think it was from 2006
On the subject of brickwalling -- Prince's Purple Rain got a remaster from him a year or so before he died that was released to cash in on his passing and... wow. Let's just say it is so bad that WB lets you stream the original CD master from 1984 *and* the new one on Spotify. As in, it is so horrendous and unbalanced that they just threw their hands up and went "okay, here's both of them. Pick your poison. Brickwalling, or overly quiet 80s mastering for CD."
"believing in yourself just a little bit, and not taking shit so seriously" IMHO is the real takeaway from this video. Shot for your work bro. Im always happy to see another down to earth video from you.
“Repurposed open source software without attribute” is a perfect description of the vast majority of “ai” these days. Awesome video, thanks for the education
Great job corralling 472 people into providing data for your test, along with all the work behind a double blind study. Your data set is huge and thus a favorable P value making it statistically significant. I was involved in audio quality analysis in the early days of creating Handsfree for the car and just getting 27 subjects was work. I was reduced to taking a serving cart between cubical offering doughnuts for a 3 minute “your opinion counts” test. I had the double bind sample sets ready to go, and though they were part of my team I still got many “not today” skips.
It's a shame that all we got from the data was a stack order of the results - not a reveal which sample was which, not a distribution of votes for each of the masters, not an anonymised result set. I was one of the people that took part in the study and I feel like the data I provided wasn't used well - or at least hidden from the people put in the effort. Benn portrays himself as a data scientist, he should know how important and interesting these things are and how underwhelming the results are...
@@ileutur6863It really didn't. The vast majority of albums are not pushing the sort of loudness they were in the late 00s. I don't think I've heard any album worse than Metallica's Death Magnetic. Five Star Hotel literally won the loudness wars with their album Grey Data, but I'd actually say it's a good master even though it's +4 (yes, plus four) LUFS.
@@GuyGamer1 I mean a lot of genres still are from -8 to -4 LUFS, I'd say that's quite loud. Isn't that quite much the same as the late 00s? The difference just is that the engineers can make that same loudness sound better.
Aww, no side by side comparisons of tbe different masters? Im sure you had your reasons, but if run time was a concern i know id keep watching a longer video for that. On another note, great timing! I just finished mastering for a client's ep that i had recorded and mixed and i just wanna put out there to any amatuer/semipro engineers to not be afraid of mastering your own mixes! I dont market myself as a mastering engineer but i always bring it up with clients before a project and include the $50 per song i charge for mastering in the initial estimate while offering to help them find a more experienced mastering engineer should they wanna go that route. Im usually working with independent artists who arent really expecting to make much money on their work so they usually opt to have me do the master and i much prefer that to them sending it to the cheapest person they could find or an ai service. My main mastering-your-own-mix takeaways are as follows: 1) take at least a couple days between mixing and mastering so you can hear it with some fresh ears 2)make extra double sure that you and your client are happy with the mix before you start mastering 3) get familiar with some of the routine technical stuff like knowing the prefered file format/resolution of cd's streaming etc, when to and not to use a dither, and what how things like leaving a db of true peak headroom or not can effect the sound when a file gets comoressed for streaming. These things tend to be more of a straightforward right or wrong answer depending on how the music will be distributed and to a lesser extent the genre of music. Dan worral's dither video is pretty much all you need on that particular topic. 4)respect your mix and dont send something back that sounds completely different than the mix the client already approved of 5)dont be so respectful that youre affraid to do anything. Just keep the untreated mix handy so you can check in on how youve altered it now and again (with levels matched). I did find myself using more subtractive eq and a lil soothe on my master to tame some hi-mid harshness and boomy lows which next time around im gonna try and be more mindful about managing at the mix level, but in this case me and the client both wanted the project done sooner than later and the problems were relatively mild so i could manage them at the master without making everything sound weird so i rolled with it and now im done and everyones perfectly happy with the results :)
I thought we would get side to side comparisons too. I would like to know his reasoning behind this decision. Benn's insights are worth every thought he has put into them.
Having no side by sides is bad. So he is telling us what to think instead of letting us decide for ourselves. I don’t think this guy is trustworthy after the poly end+ review. Once a shill always a shill
@@Wild_D I would say you should expand on your statement. But it's overflowing with ignorance, so please don't. If you have followed this channel for any amount of time, you would realize that from all music (gear) channels Benns is the least "shill". Saying that he left out the comparison to take away our decision and calling it bad is just unreasonable.
Maybe a 2nd channel video? TBH I don't have the time to do that comparison myself right now, so I probably would've ducked out early from this vid, if it was backloaded with 12 tracks. Plus, it would be better to randomly switch between versions. So... IDK
His study isn't really good if he doesn't have graphs, plots etc. to show the results. Just ranking the results or saying what each scored isn't good enough for someone that is not him to form an opinion on. What if the data was heavily skewed by a portion of people rating things really highly? E.g. lowest score as a 7, whereas other people were rating best as a 10 and worst as a 0.
@@vampires-from-mars I don’t want to believe that it’s the case but I mean look how far AI has come in literally just 2 years. As it gets tuned with a plethora of information, it will only get better and better. Maybe it’ll even be a good thing and help us expand the frontier of music, letting us move past our current limits the way synthesizers did, the way autotune did and the way DAWs helped millions without any formal musical training make some incredible beats with nothing but a laptop. Either way though, real music will always triumph
@@armisphereThe secret AI companies really don't want to tell you is that the "plethora of information" was already included in the first models you saw. ChatGPT was trained a full 2 years before they ever offered a way to opt your data out, and there doesn't seem to be much left beyond what they've already scraped that can be trained on, especially compared to the exponentially inflating data requirements of more advanced LLMs, hence the mafia-esque data sharing deals they keep cutting with publishers. The future "innovation" of LLMs is likely to be quiet implementation into everyday workflow that it is actually suited for, rather than the big and loud "AI POWERED" buzz nonsense that keeps being shoveled down all our throats. This is especially true in the arts where opinions are *sharply* divided and smaller products are liable to commit marketing suicide by associating themselves with AI.
Ozone 11 works beautifully if you implement it as you explain in 19:33 If anyone uses it as a one and done, they totally don't get that the AI assistant is only supposed to be a starting point. You are encouraged to go through the mastering chain, audition changes, tweak settings and compare with reference tracks to get the most mileage. When I can afford it, I would use a mastering engineer on an EP, but if I'm just making singles, Ozone 11 does what it needs to do.
Honestly, I know absolutely nothing about mastering, zero, really. Not proud of it, but there it is. I filter, EQ and volume control my tracks as I go, so that the end results sounds good to me, and if it doesn't, I go back to the track and fix things. I'm only an after-hours hobbyist, what do I care. But once, just once, I put Ozone (10, not 11) on my delicate, silky ambient track, choosing the least objectionable, gentlest settings it offers. And - it brickwalled my ambient track. Bloody brickwalled it. At that moment I wished there was a way to kick a piece of software in the gut.
@@SongOfItself Change the settings using a fresh audio file that is only mixed? I don't understand what you mean even after researching what brickwalled means in audio Could you explain pls? sorry if it's something simple
@@SongOfItself No shame in this my friend. If you don't know and don't feel like romanticizing perfection in music, you can 100% get away with just using the AI feature. Provided the mix is good. No amount of mastering fixes a bad mix. If you just write for yourself and post things as a hobby, I think you could do as you are doing tbh. Now, if your goal is to get better and release radio ready songs, then invest that time in learning the why and most important, training your ears.
What all the kids don't know: "Mastering" used to be the thing you may try, if you did anything in Audioengineering before. For 20 Years. I will never forget when I was in the "Mastering Cave" from EMI Elektrola back in the late 90ies for the first time, meeting "God and his machines" Literally a Roomunder the basement with an older man and expensive Stuff on every wall up to the ceiling. You gave him your DAT, he listend to your Song..."Boy, wtf were you thinking...." 1 Hour later I almost did not recognize this thing as my recording anymore.... "Now the stations might play it" Of course a lot happend...but a one button template thing will always sound like a one button template thing.
except none of these are a template. That's not how the tech works. We've had that before. These suck in a different way. Although the best one scored within the margin of error of Ed the Soundman, so make of that what you will.
I really adore that you hit upon the nuance here, there's many, many reasons to be pissed about AI from the attribution to the so-called "strip mining of human creativity". There are existential and legal questions and sometimes the whole thing makes you want to throw up your hands in despair. but dialectically - holy shit these things are so fun to play with. If you have the mentality that you're just going to screw around in comfyUI for a while, or even see what kind of sounds can come out of the TTS models when you purposely hand them really weird samples of non-voices or something - when you allow yourself to play with it like a toy - I think you're doing it right. These things can make shitty replacements of us and that's a little worrying? but they can also do stuff that's fundamentally new, and that's exciting and I just want to poke it and see what it does.
I just want to say that I always love your use of voice actors for small character bits in your video. They really add a lot of quality, and I think a lot of other artists would just throw an AI voice on it for a cheap bit. Greatly appreciated
I always thought mastering was a scam. It's basically giving someone your track to implement their taste on the final finish, like when oil paintings put on that gloss mixture to make the colours shine and pop (can't remember what it's called). This you can do and will certainly have already done if you are mixing the track. True mastering is a thing of the past as you said and it's all in the mix, always. If your mix is lacking punch, depth and space no singular plugin or algorithm will fix the problems in arrangement, sound selection and bus mixing within the track. It's really frustrating to see how so many people fall for the idea of mastering as this unreachable "enhancement" just because they don't know how simple balancing could repair over 80% of the problems. Thanks for the vid
I'd like to hear a comparison of the Californication song mastered properly vs the version that was put out to the public. PS: subscription models are the biggest scam to hit capitalism since rent. Such things make capitalism guilty of being the exact way that it accuses communism/socialism of being; the whole "Under [X economic system] you will own nothing".
I’m very glad this channel found me. I am so out of practice with recording these days, often doing live sound stuff more than anything else, but I yearn to start making music again & your little bit about having a bit of confidence in ourselves & reminding us not to take it too seriously is something I really needed to hear from a person who is as experienced as yourself. You’re doing a world of good, man. Thanks for helping me gradually find that spark again.
really love when ducking & side-chaining is used stylistically to make it sound like a drum is causing a void in the mix. tickles my brain in a good way
How helpful is this if all the people are audiohpines such as yourself? Keep in mind the majority of people who hear your music do not have a pros ears. They don't hear the majority of subtly we do. This is kinda useless when you aren't usually mastering a song for other industry pros, but rather the general public. I assure you that 400-4000 random spotify users would provide the most inconsistent useless results from this study.
@@SinewaveSinatra Well, some people need a one button solution and some would like 100 sliders. Depends where your focus is... But shiddy mastering is worse than no mastering as it just makes things louder, kills dynamic and is frankly unrespectful to the music. But sure, McDonalds sells more burgers than the local Mom&Pop burger joint as well.
Benn you are insane. I clicked on this video for the main topic that would've reaffirmed my assumptions but walked away with tons of production recommendations i would've had to search for eons to find Your work is invaluable, thank you ❤
I love that Benn did this! I recently covered installation and use of Matchering on my channel. Taking things a bit further, I used my favorite Matchering outputs as reference to adjust effects on my master chain of my mix. I think this hybrid approach can give anyone the confidence they need to get a great master on their tracks, and yeah, online mastering is a colossal scam!
As a person who makes techno, I have to do everything myself. I’m not a mastering engineer and while I tried for several years, I sucked and could never produce music that sounded good once I tried to export and share my music. I use Izotope Neutron equalizers on my group tracks and master chain, as I hate the Neutron Mix Assist. I use Tonal Balance to help get my tracks balanced. Then I use Ozone 11 and it does some additional eq filtering, stereo widening, and maximizing. I will compare my tracks to professional tracks and mine now sound pretty good. I’m happy with my Izotope purchase.
I work with AI and implement AI driven processes for companies and I can see that the biggest issue is that very quickly you fall into the fallacy that AI is going to do ALL work for you. You start looking for that recipe where you don't have to lift a finger to make money. But really - if there was such a machine that generated infinite value without any work, why would that machine need you for anything? It's like this: AI needs to be your sidekick - something that helps you over the bumps you couldn't get over without it, something that perhaps caters a valuable perspective when you missed one. But its your experience, emotions and art that needs to be the seed that develops. Don't kill your art with AI - find ways to make it into a supporting tool, but don't let it "run everything".
One of these days, you should try to get at Doug Van Sloun at Focus Mastering in Omaha. Dude has mastered Cursive, Bright Eyes, etc. Probably one of my favorite mastering engineers who cared more about harmonic content, musicality and system compatibility than loudness during the loudness war.
Месяц назад
The DETAIL you go into here is phenomenal. GREAT for people at all levels!
While I agree, FL Cloud (which is basically Image-Line's Splice, but intergrated directly into FL, so the barrier to entry is reduced), also has AI mastering. I've used it a few times. Since it's included, I can't call it a waste of money, but I know I get better results with just EQ and a limiter.
Not only you are super talented musician and producer I used to love to listen, but somehow you managed to become absolutely brilliant video content creator. Educational, scientific, and funny, what else is there to wish for! (a new Flashbulb album?)
Dear Ben, I'm just over a minute in and all my nerd-senses are tingling. Double blind study you say? Oh boy I am so stoked. You are killing it dude. Thanks.
Good points...but working on a desktop 8 track, 5 bucks for a decent tuneup on an mp3 for a guitar/vocal demo that needs turned in asap for a pitch deadline is pretty handy
for years I've been using my Compounder on drum-group channels, basses, vocals and even on the master output on a few occasions. (as a live-sound-engineer) I still have all my analogue gear, but as everything has been turning digital, I don't get to use it quite as often. happy to learn that there is still a lot to be done with it in my home studio!
Ha ha, just commented myself. Bought the Vintage Warmer when it came out. Until today part of my productions (though I now only do my own music). Cheers from Germany!
You’re not just a doer but also a teacher and I really appreciate the explanation of order of operations. I don’t think anything will ever get that human touch quit like a human.
Benn when mastering and mixing I know you mentioned going back and forth when needed. (No eq on the master bus part) But my question is do you mix with some mastering plug-ins loaded? Or do you turn them all off to get a good mix first? Should the “mix almost stand on its own” type of philosophy. Thanks for the awesome video!
Funnily enough, matchering does not even use any machine learning (there is a link to a paper on how it works in the git repo, it's in russian but google translate does a good job here). It only uses some somewhat basic dsp processing combined in a clever, easy to use way.
i love that matchering took this video as an advertisement and is proudly displaying it on its github page as 'ranked 3 out of 12 just behind 2 human mastered mixes' lol
I was one of the 472 and was enjoyably geeky in my ratings process. I pulled all the tracks into a DAW, volume matched them and A/B'ed against each other through some ATC SCM monitors. Obviously I have no idea how my top picks line up with the overall results but I will say that my top two felt head and shoulders above the rest and the difference between top two and bottom two was staggering. I have never had the opportunity to compare masters in this way before, it was a very useful and gently mind blowing experience, well worth the hour or so I put into it. I have always used Dominique Brethes at Flow Mastering for albums I've worked on, as he came recommended and when I lived in London I could cycle over to his place and do any final tweaks/notes with him. He does remastering stuff for Blue Note and hillariously hasn't updated his website the whole time I've know him... it is like a virtual time capsule of legacy web nostalgia! An artist I worked with had their next album mastered at Abbey Road, paid x5 more and felt that they had wasted quite a lot of ££.
Interesting video, as always. Still the question for me is: which would get the most plays? Paid master, AI master or I do it myself with Ozone? I've done then each several times now and in terms of listeners...it makes no difference! The pro mastered do sound a little better ... but not enough that would change one persons mind on how much they like the track. Actually, most listeners can't hear the difference anyway. I do agree that the advantage of doing it myself (as well as cost, which is important when you aren't earning anything from it) is that if I hear problems at mastering, I go back to the mix and try fix there. And even rearrange and swap out sounds etc.
all I did some years (ages) ago was put the "broadcast" preset in the multiband compressor in adobe audition, and 90% of the time the clients were happy. Perks of working for a small local studio with mostly punk rock bands lol
Hot diggity, Mr Jordan! I do wonder which tracks I was upvoting from your double-blind tests. I've just installed Matchering and I am very impressed with the ease and speed of use. Choose your reference track accordingly. Thank you so much for bringing this all to light. Cheers!
@@mo2cubing I should add that installing from the command prompt took a remarkably long time to process and following that I was a bit trepidatious at the multiple restarts to complete the installation. Maybe there was excessive server demand and you were 500’d (server error)? Try again in 12-24 hours? All up the installation process took about 15-20 minutes for me. Good luck!
Thanks Ben, this was helpful. Getting ready to throw up my first album of electronic stuff, and I’ve been paralyzed by what to do in the mastering phase. You rock mang!❤
just tried out Matchering on a few mixes, it seemed to make them sound a bit thin and removed a lot of low end. I compared it to a quick master I did with some Gullfoss, Unisum, UAD Ampex for light saturation and Ozone Maximizer as the limiter, and I liked my master a lot better. obviously those aren't free (or cheap) plugins, so it seems like Matchering could be a good option for anyone who just needs a free, quick master. but if you want better results and have the money, I would still invest in mastering plugins and/or hire a (human) mastering engineer. good to see this is basically what the blind study confirmed too. maybe AI won't take our jobs.. yet
Thats the thing is that a lot of times you basically need to watch the low end really heavily when mastering, so an ai doing it I would say would make things too flat and not have the liveliness of someone "spicing it to taste" so to speak.
I haven't been to your channel since people were giving you a hard time over your gear reviews. You seem a lot happier with the content you're making now, and it makes me happy to see!
I wish I had a way of knowing which ones I rated high/low. I thought some of the masters were absolutely awful but I have no idea if I went with the general consensus.
Same! I specifically recall one of them being extremely loud and muddy, wouldn't be surprised if that were the one he described coming in last, I certainly rated it pretty poorly, a 2 I think.
CHANNEV by analog obsession is an amazing Neve channel strip AND it's free (with optional donations) analog obsession also has an equally great low end expander called LOVEND! my favorite plugin by them though is probably Re-Life, I honestly don't know what it does specifically but it can really reshape the sound of your instrument with tighter lows and sparkly highs and is also extremely user-friendly!
Ozone has been the best for me for getting a reference of what could be done to a mix with references used and then using each module to see what I should fix in my mix down then toggle it off and go fix the mix where its needed
I feel like it’s tough to fully believe the outcome considering the mastering websites weren’t even included in the poll and the raw data is not made available. I want to believe it, but why can’t all 12 versions and the raw data be made available so people can really trust the results?
100% this - not sharing this data feels lazy at best and malicious at worst. Plus people who took part in the study (and others that would want to replicate it on themselves) don't hace a chance to learn about their own preferences.
I like the breadth of the subjects that you cover on your channel and not only that the depth You're also go into it to a level and explain things in a way that makes them understandable to everyone which really shows true mastery thank you!
you should not apply any technique by deffalut and on all things. Highpassing that low usually does-can do a lot of damage, depending on the song of course. So if you have a DnB song with lots of energy in the subbass you can destroy a lot more by highpassing it since the highpass filter introduces a phase shift which, (again depending on the mix) can completely alter the low end if its fundamentals reach very low. Its safer to use low shelving filters and thin the critical regions out a bit rather than simply highpassing everything. and if you have to highpass better to do it on an individual channel manner and not on the sum before mastering. Thats only my opinion of course
@saardean4481 Thanks for the reply. For mastering I only use a linear phase 20 Hz high pass for the mastering plugin input. My (noob) thinking is that this gets rid of inaudible sub frequencies and any DC offset from analog recording, to give the mastering plugins a better mix to work on. Just not sure about it.
@@Goettelthis causes more harm than good usually. Either a phase shift, or pre ringing or something else. And people do this thinking it saves headroom without checking their meters. I’ve never seen/heard this shave off more than 1dB without audibly compromising the sub. So as the other commenter said, rather stick to shelves. Even dynamic EQ shelves, or multiband with slow attack and release. Also, it’s usually best to use zero latency/natural phase on low end stuff. Linear phase causes pre ringing in the low end most frequently. But of course your ears + a reference track trump all the theory any way🤷🏻♂️
This has always seemed weird to me. Is there even significant energy down there? Is it significant energy that isn't supposed to be there? What channel is it coming from? How did it get there? Can you eliminate it at the source instead of applying a band-aid? It makes me think of tracks that have audible mains hum. You can put a 50/60Hz notch in there, but there are going to be harmonics that you have to notch out too. And before long, you're comb-filtering your track just to deal with a problem you could've fixed by... tapping the cord from your guitar a foot away from the power cable or something like that.
I got Ozone for $99 on sale and it has saved me so much time. I used to spend days even weeks mastering my music, now this gets me 80% there with the AI suggestions and I can tweak from there. And MAN can you tweak stuff. I'd use the phrase "Game Changer" unironically.
ozone is all i use. i mostly just make music for funsies and am not trying to make marketable material for an income, but i have been producing in one form or another since the late 90s. ive had professional mastering done, and have also just thrown a compressor on the whole mix and hit the "mastering" preset, and have tried just about everything in between. ozone just checks off all the right boxes for me. cost is right, and its highly tweakable, but doesnt take a phd in music to understand. i had a cracked copy initially to try it out, but promptly bought it after literally one track being run through it, and this was with a preset and before i knew how to actually use it. its deep enough that it will reward you spending time to learn it.
Yeah, I always got good results from Ozone. The presets are usually a good start, then you can tweak the Hell out of them. $99 is money well spent... :)
And 90% of people then listen to streaming and over compressed music on an mobile phone with earbuds where you earn 0.00000000001 cent per play. But I get the point.
Which actually makes it more important. It's not too hard to make a track sound good on a high end playback system in a good listening environment, but it takes a real master of their craft with years and years of experience (and very accurate monitoring) to make it also sound halfway decent when someone plays a RUclips upload with 2 or 3 generations of lossy compression through their cell phone speaker at the other end of the bus.
3:15 while yes a bad master can ruin a good song in my option it’s significantly easier to learn the fundamentals of a good master than it is to learn any other part of the music making process. If you are a good mix engineer you absolutely can master your own music.
When someone cares about my music, I'll pay for mastering. For now, Logic's AI mastering does me pretty well. I used to try doing it myself, but I like being lazy and I was never that good anyway. Biggest complaint with Logic's mastering is everything is smoothed out too much. Still sounds good enough for most ears though.
There was a sale a couple of years ago and I mistakenly got elements as opposed to the standard version, and since I have it, might as well use it. I usually end up doing one of three things with it. Sometimes I run it and genuinely feel it's good enough after doing various reference tests and just roll with it. This usually works best on really simple tracks. Other times I will still use it but run some plugins before it just to change the character of it slightly, like a saturator or similar things. The third option, and probably the best use for it, I'll run it, listen to what it's doing, and then sort of use that as a loose guide to use what few mastering plugins I do have to get to a similar place but with my own tweaks on it. Regardless of which I do, I always turn off the elements imager. Even the free version of ozone's imager gives you slightly more control than the single slider elements gives you, and I rarely use imagers to begin with.
Mastering engineer removing original comment: Sorry, guys… I may well have recorded, performed on, mixed or mastered ≈ 2k5 (have no idea) published tracks over 35 years, and trained seven engineers (five kept at it). This summer I was offered a college level teaching position, but declined for health reasons. Youth isn’t eternal after all. That doesn’t mean much - who knew, right…? I could be as silly as my clients could be - although there’s a serious discrepancy to that logic. By all means - follow your own paths, I won’t be in the way. Aggression & hyper-competitiveness is quite abundant without me contributing to more of it. Make music - that may help…?. 🎶💛
I think there's a bit more nuance than that. I'd argue that not adjusting for loudness is actually the right way to go, but everyone is releasing music for different contexts and that matters too. On one hand loudness is an extremely important factor of a master. You said it yourself that anything louder will sound better. So generally the louder someone can pump a track without losing perceived quality, ie loss of clarity due to overcompression, the better. The prevalence of automatic volume normalisation on modern listening platforms is a huge mark in favour of not caring about loudness. If you only care about Spotify and Apple Music then just hit -14 LUFS and call it a day. But many platforms don't have this feature, and many users of the platforms that do have it will just turn it off. There's also the consideration that if you're making music that you're intending to play/have played in a club/party/festival context then you don't want your music to sound quiet compared to every other track. On a similar note, even if the DJ level matches between songs, a less compressed song played loud sounds totally different to a very compressed song sound quiet. One will just sound odd in the context of a set filled with the other. The biggest issue with Benn's test is that he only did one single test, rather than a multitude of songs across a multitude of different genres. IMO only using a single song invalidates any conclusion that can be made from the test.
@@LunicussOfficial Your argument doesn't make sense. On the one hand you say its better to not adjust for loudness when mastering. But then right after that you acknowledge the importance of the music having been able to match radio or dj's other songs. Sure a less compressed song sounds different, but this difference is not the same across the board. "Over compression" is exactly that, it's too much compression. If your song is loud, or a part of your song, you don't need to turn it down. If anything, you turn one fader up and it's really loud compared to the others, that means the rest are almost certainly too soft. What you mention is exactly the need for mastering. I understand that too much compression sounds like shit, sure as it's manipulating and squashing the sound down. But you don't need to do that. You should be able to adjust the volume somewhere else besides the mastering stage. There would be a few places you could increase the lower ones.
@@LunicussOfficial I think you missed my point. As professionals, when we _compare_ mastered and unmastered versions of the same track, we are NOT comparing loudness. That’s the end result, and relates to the target platform or media, their guidelines or how they can be… exploited. As an example, Jordan played bits of audio - unmastered and mastered - where the mastered ones sounded “louder” on playback. Sure, fine when said track is competing with others out there. In the studio, any tonal, phase issues, clipping or dynamic wrongdoings will go unnoticed if the master appears louder than the mix - when _checking the integrity_ of your result. Simply getting “impressed” by your own processing doesn’t cut it, as you risk ruining your client’s hard work. The client - and future listeners - will hopefully be impressed. That happens “out there”, not in the hopefully (!) controlled environment of the mastering suite. Jordan mentions many of the same issues. Don’t mistake “nuances” for taste or aesthetics - surely important for the artists and their listeners. The technical aspects and limitations however, are the same across genres and taste. The simple difference between a calibrated monitor & room, and a speaker, is that the monitor reveals what a typical speaker wouldn’t. A good master will translate well to a phone, a mono BT speaker, earbuds, a full blown car stereo… That’s not about loudness of the master itself. The encoding used by a distributor, or limitations of physical media and playback equipment come into play (bad pun), and are known by the mastering engineer. Silly analogy; people wouldn’t like their taxi-driver to “hate them” or drive recklessly to “compete” with an Uber? Hopefully, (s)he’ll treat any paying passenger with respect, knowledge beyond the GPS, and without judgement on where you want to go. ‘Nuff nuance. Make music… 👍
Benn, Thank you. I am not in a position to reward you financially at this time but I do really appreciate the integrity and good will of your channel. Would love to see more of your playing too!
Landr has been useful for my band. Just gives an idea of how mixes might come out. Before sending the songs to someone who is good at mastering. Also helps demos sound a little more clear. The professional definitely sounded better than Landr. Imma make my band watch this video though. I bet they will enjoy learning some of this.
Playing piano and trumpet and flute and guitar and having been in high school and junior college band/jazz-rock ensemble and taken electronic music classes (1978) and having had a buddy go to the Berkley School in Boston and now 45 years experience in the high-performance computing industry, I appreciate everything in this video, and I am not at all surprised that AI did not come out on top...
Most of A.I. mastering for me was just exporting it, shooting it through, tweaking the original and exporting it again. After a certain point it just became like another mastering VST, one in a big chain. Lots of stuff didn't really benefit from it, like the raw stuff I'd make in Audacity or field record, but the electronic stuff I'd do turned out alright. Most of it was just, like, a motivator to finish the thing. It's a fast thing that makes the last part of the process less of a nervous uncertain tweak of a knob and more of a climactic loading screen that just shouts "You did it!"
Half of this video felt like a big ad. The title of this video implied that the emphasis would be on the experiment and industry trends when it was mostly just a tutorial.
To quote the great JPEGMAFIA: "why would I pay someone $300 to make my song sound like shit when I'm perfectly capable of making it sound like shit on my own"
he said ass not shit 🤓☝
For FREE even!
Add reverb to the master and Turn off the limiter
not an artist but I feel like mastering is a total scam, Mozart and Beethoven had none of this bs and they are the most famous artists of all time
@@ClaimClam They didn't have mastering because they weren't putting out albums, dumbass. They weren't invented for another 200 years.
FINALLY A DOUBLE BLIND STUDY. No one outside of the hard sciences seem to understand how important this can be for something like this. Thanks!
I agree... this comment section is filled with anecdotes like "I don't know anything about AI so I tried using it, and it made crappy-sounding music."
If I said "I don't know anything about multiband compression, so I tried using a multiband compressor and it made my music sound like crap," the same people would have critical notes.
@@stickyfox I actually tried to get set up to use the Comfy AI interface and got lost with all the virtual Linux machine stuff. The base "make this sound like this" seems to be hit or miss but I'd love to dig in with more control for sure. Just gotta learn how to code, I guess LOL.
@@cutaia If the kernel is the same on those then he was correct to cut them out before the study. If not, then perhaps he should have set up a submatrix instead of just discarding them outright . I'll also be blunt and say that no one really believes AI is competing in this space with humans. It is more apt to say that people are using AI where they otherwise would not have hired a human anyways. AI (not just in this instance) is much like Harbor Freight in that regard.
@@cutaia I'm going to bow out here. I was just happy to see someone on youtube even approaching the level of rigor I used to use when I was a professional production biochemist. I'm not sure if I stepped into something you have against this guy because to be honest this is the first time he's been suggested to me. Take care, and feel free to attempt to replicate his results.
@@cutaiano, the method isn't "obviously flawed". The results presented do not cover the eliminated mastering tools. His personal bias does not affect the presented data. All that step you're shouting about makes the study less exhaustive. That's it. It's not poor methodology because it was not a part of the method. What it is a reduction in scope. Given what absolute gibberish machine learning can come up with, it's an entirely unsurprising event.
This limitation of scope does not affect the results that were presented - and the results show that the best AI mastering tool does a comparable job to a highly regarded expert on a time constraint.
63 yo recording/mixing engineer here…. Done tons of records in all kinds of genres. Just a reminder that initially mastering was a necessary step in the production chain. Mainly it was to make sure the final mix would be printable on a vinyl record. It was a very precise job to do and it was a very restrictive process as the margins of error were pretty tight. Those days are gone and mastering is certainly not a necessity anymore. As long as your levels match the levels required by either Apple, Spotify, or any other platform you will be fine. Those levels are easily attainable with any plugin that do just that. Couple of years ago I got Ozone 10 for under $100. Tried it just for the fun of it. Used the presets and voilà… Very impressive. My clients are pretty happy. Unless you produce a very prolific artist in the higher spheres of the music industry, spending money on a mastering for a digital only release doesn’t make sense anymore.
Well said! I make music for fun. For me, mastering is adding that little touch at the end. Just a bit of EQ and light compression adds that final touch. There is no need to spend a bunch of time and money on that when some free VSTs do the job. In general I prefer hardware. So I plan on moving my mastering to hardware.
@@jennoscura2381 Funny because my path is the other way around… I have all the real gear… Tubetech, Elysia Alpha, Neve Eq, Manley, API, etc, name it I have got it… And honestly, since my clients always come back for « can you +1 dB this and that, lower - 1dB this and that » or just a bit of this and that and that’s final I promise… I resorted to all in the box without them knowing it and they are happier because « yes I can do all the little useless tweaks » that will make you happy… It’s possible that if I were to compare the same processing paths with hardware I would feel the difference… But I am done with that because it’s so much time consuming that it’s not worth it. But for you, an independent producer who works at his pace for his own projects, it’s pretty fun turning the buttons for real and smell the warming tubes… 👍🏼
@craigwillms61 I agree that a competent engineer is the key factor. My point is more about what tools are involved in the process.
I am no longer convinced that analog gear is superior to top digital tools for mastering purposes. I suppose it is a very subjective debate...
Cheers !
@@robertl.6919 I reread what you said and deleted my comment. Also, every time I click reply I tell myself I've got to stop doing this, no one cares what I think. I'm still working on that, obviously
@@craigwillms61 😁 Haha ! Well you could also say « who cares about what anybody else says in these forums »… it’s our choice to comment in these forums. At some point, all opinions are as valuable as anyone else’s unless a well known expert joins the conversation but that rarely happens.
We talk, share opinions and experiences. That’s all there is to it.
3:25 about a decade ago I was friends with this metal band that made great music, they saved up a bunch of money to get their album mastered and chose someone who specialized in COUNTRY MUSIC, because they were a "big name". it came out sounding terrible. picking the right mastering person matters.
and justice for all ?🤔
@@andreviana4418 Nah, that was all Lars and James. They're the ones who kept turning down Jason's bass tracks.
Much like Death Magnetic, it arrived to the mastering engineer pre-fucked.
And give them credit once you do... Friend of mine mastered the demo of a three piece from the uk, then they got signed, released an album which went nr. 1 I think, with some of those tracks on it as they were on demo, without giving any credit to my mate. Contracts matter... Band not at fault though btw. Blame Jamie... xx
bit gossipy at the end there but just HAD to get their name in somehow...
@@post5230 😉
I love the "putting on boxing gloves to build Lego" analogy for the obsession people have with plaintext prompting interfaces. Prompting in plain English is just the worst way to accomplish basically any computing task and also takes a lot more processing power to work than a traditional UI.
SPOILER ALERT (actual scores w/o context):
6.4/10 - Max Hosinger
6.1/10 - Ed The Soundman
5.9/10 - Matchering 2.0 (run locally)
5.8/10 - Ozone + Neutron (guided by Benn)
4.9/10 - Kits.ai (running Matchering w/ custom presets)
4.8/10 - Platinum ComPounder (hardware only)
3.8/10 - Ozone (best AI recommendation)
Disqualified due to objectively bad mix or errors such as severe clipping/distortion:
Landr, Bandlab, Waves, Virtu, Mixea
Any chance of you publishing which one was which? I was one of the 472 and I'd sure like to know if I can trust my ears.
@@jeffkeeling8917 I, too, was one of the 472 and would love to know which was which. I'm kind of surprised the average ratings were all relatively tight. I had one 10, one 8, one 7, and the rest were all quite poor (1-3).
@@Potatographical That's kind of how I had it...a 10, an 8, a 5 and then like you 1~2 points for the rest.
Please can we just get the standard deviation of the scores. I’m sure we might find out that this scores are not statistically different
So #1 was 8.5% better than matchering and #2 was 3% better. I share the intuition that some of these differences may not be reliably different
I mastered my new album myself, and tested a few of these AI mastering plugings / websites as I was curious. Once volume matched, literally every AI mastered track was worse than my mixdown.
Yeah I suspect all a lot of them do is make shit louder. Well, we all have a brickwall limiter somewhere in our toolkit (99% of the time theres one in your daws bucket of default plugins) so we can do that ourselves, and usually wreck all the transients in the process.
Same here. And I'm far away from considering myself a professional
I had isse with mastering engineer as well... Where we just wasn't vibing... And paying different people to see we are "matching" sound wise just became so expensive that I'm now happy with whatever commes out at the end out of my Daw 😅
AI cannot sum 2+2.....
I used Ozone quite a lot in the past, and my mixes were super loud and brickwalled. But when I used it I caught myself mixing "for mastering", in other words I was making final mixes with Ozone on master to make the result transferrable (headphones, car stereo, phone loudspeaker) and louder, and not to make mix better. Now I use EQ + Multiband Comp + Limiter on master. I don't think my music is louder now, but it sounds tighter and I have better control over mix
(I still think I could use new Ozone features, or at least it's "maximizer", or whatever it called nowadays, though)
Max Honsinger here. Thanks for the great review Benn! Good to know Ai isn't coming for my job.... yet.....
great job max! I doubt the silicon will be stealing your office anytime soon
You mean, rebranded faster automation inclinations?
AI would need training data of mixed/unmixed songs. I think Some companies are searching mastering experts right now.
@@Solunex ha yea they contact me asking me to train the systems. Like, you want me to train the machine that replaces me?! For a few $? Seriously?!
@@mixgenie and still. Can't replace you xD
“could we get a vibraphone playing a whole-tone scale, up? thanks”
these are the jokes i come here for, my man.
03:35
Editing is great as well. "Mastering" caught me off guard 😂
Fun trick to get your band some more attention at shows: make a instrumental, sort of 'chill' or 'minimalisted' mix of your set, and not the songs in their entirety, just like intro verse chorus outro, or verse bridge chorus, or intro bridge outro, or intro solo. Have the club sprinkle those tracks in with the house music, but only loud enough that you can hear it behind conversation while not being attention grabbing. By the time you go on stage, people will be used to hearing your songs but wont be able to place them ;)
Nope just make your selfs sound good and look good on camera and sing pop songs as remix or covers and every 5th video do a original song.
@HOLLASOUNDSMUSIC boy, are you barking up the wrong tree 🤣🤣🤣
If this happened to me I'd wonder who the ***sholes are playing music so low I can hardly hear it. That's what Christian Soccer moms do when driving. You will not win anyone over this way. Music is meant to be heard, not played so quietly you "can't place it"
Mastering is like putting varnish on an oil painting. If you use the wrong one or apply it incorrectly the painting is ruined. Same with using AI. Very nice and informative video, keep it up benn!
Good choice of analogy.
And if someone doesn't like the painting to start with, no amount of varnish will persuade them to like it.
@@OrangeNashAnd if the painter isn't skilled, no amount of expensive paint will make their picture look good.
@@henryglennon3864 I've heard the analogy "sonic varnish" before refering to layers of subtle saturation, but it also works really well when describing what a mastering engineer should be doing most of the time too (not nessisarilly adding layers of subtle saturation, but just adding the sonic varnish to a track, in a more generalised way). :)
Good mastering is also concerned with the presentation or to continue your analogy "the frame" and environment in which the format of music is going to be experienced.
i.e. vinyl mastering is a whole different realm vs streaming, or for that matter vs cinema and vs radio.
Hey, Benn. The author of Matchering here. Thank you for your awesome review! Greetings from Russia. 👋
Greetings from Mytischi, Homie! Just was going to mention you here but found this comment.
@@rpocc Hey, SSSR Labs friends 🤩
greets from Kursk then =)
can you link the hosted version plz
@@zloidooraque0 Greets, bro, take care 🙏
The fact you mastered that dillinger EP has floored me 25s in to the video
Same, that’s crazy I had to do a double take on that one
@@rottingsunrecords666and you both just made me double take then correct the spelling and realise no, not Dillinjah, Dillinger .... That would have been crazy😂
He toured with them back in the early 2000s too I think!
@@tezeta3725 yes! one of my local venues has old concert posters all over the walls and i was floored when i saw one with Dillinger and Flash Bulb. i think it was from 2006
dillinger escape plan?
On the subject of brickwalling -- Prince's Purple Rain got a remaster from him a year or so before he died that was released to cash in on his passing and... wow. Let's just say it is so bad that WB lets you stream the original CD master from 1984 *and* the new one on Spotify. As in, it is so horrendous and unbalanced that they just threw their hands up and went "okay, here's both of them. Pick your poison. Brickwalling, or overly quiet 80s mastering for CD."
"believing in yourself just a little bit, and not taking shit so seriously"
IMHO is the real takeaway from this video.
Shot for your work bro. Im always happy to see another down to earth video from you.
“Repurposed open source software without attribute” is a perfect description of the vast majority of “ai” these days. Awesome video, thanks for the education
Great job corralling 472 people into providing data for your test, along with all the work behind a double blind study. Your data set is huge and thus a favorable P value making it statistically significant. I was involved in audio quality analysis in the early days of creating Handsfree for the car and just getting 27 subjects was work. I was reduced to taking a serving cart between cubical offering doughnuts for a 3 minute “your opinion counts” test. I had the double bind sample sets ready to go, and though they were part of my team I still got many “not today” skips.
It's a shame that all we got from the data was a stack order of the results - not a reveal which sample was which, not a distribution of votes for each of the masters, not an anonymised result set. I was one of the people that took part in the study and I feel like the data I provided wasn't used well - or at least hidden from the people put in the effort.
Benn portrays himself as a data scientist, he should know how important and interesting these things are and how underwhelming the results are...
From the outside looking in, the fact that the “Loudness Wars” is still going on is nuts.
The loudness wars ended. Loudness won.
I don’t think the loudness war is near as bad as in the early 2000s
@@ileutur6863It really didn't. The vast majority of albums are not pushing the sort of loudness they were in the late 00s. I don't think I've heard any album worse than Metallica's Death Magnetic.
Five Star Hotel literally won the loudness wars with their album Grey Data, but I'd actually say it's a good master even though it's +4 (yes, plus four) LUFS.
@@GuyGamer1 I mean a lot of genres still are from -8 to -4 LUFS, I'd say that's quite loud. Isn't that quite much the same as the late 00s? The difference just is that the engineers can make that same loudness sound better.
@@ileutur6863 WHAT?
Aww, no side by side comparisons of tbe different masters? Im sure you had your reasons, but if run time was a concern i know id keep watching a longer video for that.
On another note, great timing! I just finished mastering for a client's ep that i had recorded and mixed and i just wanna put out there to any amatuer/semipro engineers to not be afraid of mastering your own mixes! I dont market myself as a mastering engineer but i always bring it up with clients before a project and include the $50 per song i charge for mastering in the initial estimate while offering to help them find a more experienced mastering engineer should they wanna go that route. Im usually working with independent artists who arent really expecting to make much money on their work so they usually opt to have me do the master and i much prefer that to them sending it to the cheapest person they could find or an ai service. My main mastering-your-own-mix takeaways are as follows:
1) take at least a couple days between mixing and mastering so you can hear it with some fresh ears
2)make extra double sure that you and your client are happy with the mix before you start mastering
3) get familiar with some of the routine technical stuff like knowing the prefered file format/resolution of cd's streaming etc, when to and not to use a dither, and what how things like leaving a db of true peak headroom or not can effect the sound when a file gets comoressed for streaming. These things tend to be more of a straightforward right or wrong answer depending on how the music will be distributed and to a lesser extent the genre of music. Dan worral's dither video is pretty much all you need on that particular topic.
4)respect your mix and dont send something back that sounds completely different than the mix the client already approved of
5)dont be so respectful that youre affraid to do anything. Just keep the untreated mix handy so you can check in on how youve altered it now and again (with levels matched).
I did find myself using more subtractive eq and a lil soothe on my master to tame some hi-mid harshness and boomy lows which next time around im gonna try and be more mindful about managing at the mix level, but in this case me and the client both wanted the project done sooner than later and the problems were relatively mild so i could manage them at the master without making everything sound weird so i rolled with it and now im done and everyones perfectly happy with the results :)
I thought we would get side to side comparisons too. I would like to know his reasoning behind this decision. Benn's insights are worth every thought he has put into them.
Having no side by sides is bad. So he is telling us what to think instead of letting us decide for ourselves. I don’t think this guy is trustworthy after the poly end+ review. Once a shill always a shill
@@Wild_D I would say you should expand on your statement. But it's overflowing with ignorance, so please don't. If you have followed this channel for any amount of time, you would realize that from all music (gear) channels Benns is the least "shill". Saying that he left out the comparison to take away our decision and calling it bad is just unreasonable.
Maybe a 2nd channel video? TBH I don't have the time to do that comparison myself right now, so I probably would've ducked out early from this vid, if it was backloaded with 12 tracks. Plus, it would be better to randomly switch between versions. So... IDK
His study isn't really good if he doesn't have graphs, plots etc. to show the results. Just ranking the results or saying what each scored isn't good enough for someone that is not him to form an opinion on. What if the data was heavily skewed by a portion of people rating things really highly? E.g. lowest score as a 7, whereas other people were rating best as a 10 and worst as a 0.
Once again, it shows that AI is best used as a tool, assisting a real human instead of replacing them entirely
…for now
@@armisphere I'll believe it when I see it
@@vampires-from-mars I don’t want to believe that it’s the case but I mean look how far AI has come in literally just 2 years. As it gets tuned with a plethora of information, it will only get better and better.
Maybe it’ll even be a good thing and help us expand the frontier of music, letting us move past our current limits the way synthesizers did, the way autotune did and the way DAWs helped millions without any formal musical training make some incredible beats with nothing but a laptop.
Either way though, real music will always triumph
@@armisphereThe secret AI companies really don't want to tell you is that the "plethora of information" was already included in the first models you saw. ChatGPT was trained a full 2 years before they ever offered a way to opt your data out, and there doesn't seem to be much left beyond what they've already scraped that can be trained on, especially compared to the exponentially inflating data requirements of more advanced LLMs, hence the mafia-esque data sharing deals they keep cutting with publishers.
The future "innovation" of LLMs is likely to be quiet implementation into everyday workflow that it is actually suited for, rather than the big and loud "AI POWERED" buzz nonsense that keeps being shoveled down all our throats. This is especially true in the arts where opinions are *sharply* divided and smaller products are liable to commit marketing suicide by associating themselves with AI.
@armisphere the improvement of AI at this point does not seem to be exponential
Ozone 11 works beautifully if you implement it as you explain in 19:33
If anyone uses it as a one and done, they totally don't get that the AI assistant is only supposed to be a starting point. You are encouraged to go through the mastering chain, audition changes, tweak settings and compare with reference tracks to get the most mileage. When I can afford it, I would use a mastering engineer on an EP, but if I'm just making singles, Ozone 11 does what it needs to do.
This is what I do!
Honestly, I know absolutely nothing about mastering, zero, really. Not proud of it, but there it is. I filter, EQ and volume control my tracks as I go, so that the end results sounds good to me, and if it doesn't, I go back to the track and fix things. I'm only an after-hours hobbyist, what do I care. But once, just once, I put Ozone (10, not 11) on my delicate, silky ambient track, choosing the least objectionable, gentlest settings it offers. And - it brickwalled my ambient track. Bloody brickwalled it. At that moment I wished there was a way to kick a piece of software in the gut.
...and yes, I kn ow it's supposed to be a starting point only, but where exactly do I go starting from a brickwalled track?
@@SongOfItself Change the settings using a fresh audio file that is only mixed?
I don't understand what you mean even after researching what brickwalled means in audio
Could you explain pls? sorry if it's something simple
@@SongOfItself No shame in this my friend. If you don't know and don't feel like romanticizing perfection in music, you can 100% get away with just using the AI feature. Provided the mix is good. No amount of mastering fixes a bad mix.
If you just write for yourself and post things as a hobby, I think you could do as you are doing tbh.
Now, if your goal is to get better and release radio ready songs, then invest that time in learning the why and most important, training your ears.
What all the kids don't know: "Mastering" used to be the thing you may try, if you did anything in Audioengineering before. For 20 Years. I will never forget when I was in the "Mastering Cave" from EMI Elektrola back in the late 90ies for the first time, meeting "God and his machines" Literally a Roomunder the basement with an older man and expensive Stuff on every wall up to the ceiling. You gave him your DAT, he listend to your Song..."Boy, wtf were you thinking...." 1 Hour later I almost did not recognize this thing as my recording anymore.... "Now the stations might play it"
Of course a lot happend...but a one button template thing will always sound like a one button template thing.
That's kind of like listening to Rush demo tapes and the final particularly for Territories, Big Money and others on that album.
except none of these are a template. That's not how the tech works. We've had that before. These suck in a different way.
Although the best one scored within the margin of error of Ed the Soundman, so make of that what you will.
Coulda used this video 2 dozen albums ago.
ahahaha fairz
Daft Punks RAM is so beautiful mixed and mastered....its just unbelievable. Glad to see it used as an example
Wow D.E.P is a hell of a band to master, That's really awesome Benn!
I really adore that you hit upon the nuance here, there's many, many reasons to be pissed about AI from the attribution to the so-called "strip mining of human creativity". There are existential and legal questions and sometimes the whole thing makes you want to throw up your hands in despair.
but dialectically - holy shit these things are so fun to play with. If you have the mentality that you're just going to screw around in comfyUI for a while, or even see what kind of sounds can come out of the TTS models when you purposely hand them really weird samples of non-voices or something - when you allow yourself to play with it like a toy - I think you're doing it right. These things can make shitty replacements of us and that's a little worrying? but they can also do stuff that's fundamentally new, and that's exciting and I just want to poke it and see what it does.
100% agree. Same for image and video.
Superb video Benn! Shared on Sonicstate today 👏
I just want to say that I always love your use of voice actors for small character bits in your video. They really add a lot of quality, and I think a lot of other artists would just throw an AI voice on it for a cheap bit. Greatly appreciated
distrokid has some new dumb AI "mastering" charging 10$ to ruin the tracks by just making it loud and distorted.
I always thought mastering was a scam. It's basically giving someone your track to implement their taste on the final finish, like when oil paintings put on that gloss mixture to make the colours shine and pop (can't remember what it's called). This you can do and will certainly have already done if you are mixing the track. True mastering is a thing of the past as you said and it's all in the mix, always. If your mix is lacking punch, depth and space no singular plugin or algorithm will fix the problems in arrangement, sound selection and bus mixing within the track. It's really frustrating to see how so many people fall for the idea of mastering as this unreachable "enhancement" just because they don't know how simple balancing could repair over 80% of the problems. Thanks for the vid
I'd like to hear a comparison of the Californication song mastered properly vs the version that was put out to the public.
PS: subscription models are the biggest scam to hit capitalism since rent. Such things make capitalism guilty of being the exact way that it accuses communism/socialism of being; the whole "Under [X economic system] you will own nothing".
I’m very glad this channel found me. I am so out of practice with recording these days, often doing live sound stuff more than anything else, but I yearn to start making music again & your little bit about having a bit of confidence in ourselves & reminding us not to take it too seriously is something I really needed to hear from a person who is as experienced as yourself. You’re doing a world of good, man. Thanks for helping me gradually find that spark again.
The best part of all this is I don’t have any money to give anyone. Someone has already claimed all my money. Problem solved.
Or you could just not have any music
really love when ducking & side-chaining is used stylistically to make it sound like a drum is causing a void in the mix. tickles my brain in a good way
Everything Newfangled is just amazing...Eventide quality meets modern UI...funny-face-reviewer would love it.
love Newfangled too, elevate is superb.
Ditto for this - great company. Their Generate synth is also the bizznizz :-)
How helpful is this if all the people are audiohpines such as yourself? Keep in mind the majority of people who hear your music do not have a pros ears. They don't hear the majority of subtly we do. This is kinda useless when you aren't usually mastering a song for other industry pros, but rather the general public. I assure you that 400-4000 random spotify users would provide the most inconsistent useless results from this study.
@@SinewaveSinatra Well, some people need a one button solution and some would like 100 sliders.
Depends where your focus is...
But shiddy mastering is worse than no mastering as it just makes things louder, kills dynamic and is frankly unrespectful to the music.
But sure, McDonalds sells more burgers than the local Mom&Pop burger joint as well.
I just realized your voice sounds a lot like that of Not Just Bikes. Another top pedant in their field!
The shocked face thumbnail part felt personal 😭🤣
Great video as usual! 🙌🏻
Benn you are insane. I clicked on this video for the main topic that would've reaffirmed my assumptions but walked away with tons of production recommendations i would've had to search for eons to find
Your work is invaluable, thank you ❤
I love that Benn did this! I recently covered installation and use of Matchering on my channel. Taking things a bit further, I used my favorite Matchering outputs as reference to adjust effects on my master chain of my mix. I think this hybrid approach can give anyone the confidence they need to get a great master on their tracks, and yeah, online mastering is a colossal scam!
As a person who makes techno, I have to do everything myself. I’m not a mastering engineer and while I tried for several years, I sucked and could never produce music that sounded good once I tried to export and share my music. I use Izotope Neutron equalizers on my group tracks and master chain, as I hate the Neutron Mix Assist. I use Tonal Balance to help get my tracks balanced. Then I use Ozone 11 and it does some additional eq filtering, stereo widening, and maximizing. I will compare my tracks to professional tracks and mine now sound pretty good. I’m happy with my Izotope purchase.
Good video, always like the insight on the tech side most youtubers miss
I work with AI and implement AI driven processes for companies and I can see that the biggest issue is that very quickly you fall into the fallacy that AI is going to do ALL work for you. You start looking for that recipe where you don't have to lift a finger to make money. But really - if there was such a machine that generated infinite value without any work, why would that machine need you for anything? It's like this: AI needs to be your sidekick - something that helps you over the bumps you couldn't get over without it, something that perhaps caters a valuable perspective when you missed one. But its your experience, emotions and art that needs to be the seed that develops. Don't kill your art with AI - find ways to make it into a supporting tool, but don't let it "run everything".
One of these days, you should try to get at Doug Van Sloun at Focus Mastering in Omaha. Dude has mastered Cursive, Bright Eyes, etc. Probably one of my favorite mastering engineers who cared more about harmonic content, musicality and system compatibility than loudness during the loudness war.
The DETAIL you go into here is phenomenal. GREAT for people at all levels!
While I agree, FL Cloud (which is basically Image-Line's Splice, but intergrated directly into FL, so the barrier to entry is reduced), also has AI mastering. I've used it a few times. Since it's included, I can't call it a waste of money, but I know I get better results with just EQ and a limiter.
That just makes it a waste of time, which still isn't great
Not only you are super talented musician and producer I used to love to listen, but somehow you managed to become absolutely brilliant video content creator. Educational, scientific, and funny, what else is there to wish for! (a new Flashbulb album?)
22:05 - I literally laughed out loud. I don't know why this tickled me so, but I wasn't expecting it, and you got me.
@Benn Jordan. So impressed people like you exist. Thank you so much for everything. Now looking for your website.
i have a professor from my uni who masters my music for free, im not excited to start having to pay so much money for mastering 😭
Thank god. "THIS IS ART THERE IS NO SUCH THING AS A PERFECT MASTER." Also I am a HUGE fan of Newfangled Elevate.
Dear Ben,
I'm just over a minute in and all my nerd-senses are tingling. Double blind study you say? Oh boy I am so stoked. You are killing it dude. Thanks.
Good points...but working on a desktop 8 track, 5 bucks for a decent tuneup on an mp3 for a guitar/vocal demo that needs turned in asap for a pitch deadline is pretty handy
I did NOT expect to see VintageWarmer darkening my doorstep at this late date.
Also all those green waveforms makes me miss CoolEdit :(
Loved cooledit. ❤
for years I've been using my Compounder on drum-group channels, basses, vocals and even on the master output on a few occasions. (as a live-sound-engineer) I still have all my analogue gear, but as everything has been turning digital, I don't get to use it quite as often. happy to learn that there is still a lot to be done with it in my home studio!
What a surprise! Thank you for your kind words about our company and for featuring our plugins.
Always had great products!
Ha ha, just commented myself. Bought the Vintage Warmer when it came out. Until today part of my productions (though I now only do my own music). Cheers from Germany!
You’re not just a doer but also a teacher and I really appreciate the explanation of order of operations. I don’t think anything will ever get that human touch quit like a human.
Benn when mastering and mixing I know you mentioned going back and forth when needed. (No eq on the master bus part)
But my question is do you mix with some mastering plug-ins loaded? Or do you turn them all off to get a good mix first? Should the “mix almost stand on its own” type of philosophy. Thanks for the awesome video!
Awesome. Thanks for making this Benn! Suuuuper helpful!!
Funnily enough, matchering does not even use any machine learning (there is a link to a paper on how it works in the git repo, it's in russian but google translate does a good job here). It only uses some somewhat basic dsp processing combined in a clever, easy to use way.
i love that matchering took this video as an advertisement and is proudly displaying it on its github page as 'ranked 3 out of 12 just behind 2 human mastered mixes' lol
I was one of the 472 and was enjoyably geeky in my ratings process. I pulled all the tracks into a DAW, volume matched them and A/B'ed against each other through some ATC SCM monitors. Obviously I have no idea how my top picks line up with the overall results but I will say that my top two felt head and shoulders above the rest and the difference between top two and bottom two was staggering. I have never had the opportunity to compare masters in this way before, it was a very useful and gently mind blowing experience, well worth the hour or so I put into it.
I have always used Dominique Brethes at Flow Mastering for albums I've worked on, as he came recommended and when I lived in London I could cycle over to his place and do any final tweaks/notes with him. He does remastering stuff for Blue Note and hillariously hasn't updated his website the whole time I've know him... it is like a virtual time capsule of legacy web nostalgia! An artist I worked with had their next album mastered at Abbey Road, paid x5 more and felt that they had wasted quite a lot of ££.
You mastered a Dillinger Escape Plan album? LEGEND
Interesting video, as always. Still the question for me is: which would get the most plays? Paid master, AI master or I do it myself with Ozone? I've done then each several times now and in terms of listeners...it makes no difference! The pro mastered do sound a little better ... but not enough that would change one persons mind on how much they like the track. Actually, most listeners can't hear the difference anyway.
I do agree that the advantage of doing it myself (as well as cost, which is important when you aren't earning anything from it) is that if I hear problems at mastering, I go back to the mix and try fix there. And even rearrange and swap out sounds etc.
Ayyye, shout out from Kalamazoo, Love randomly seeing Bell's IPAs in your vids!
all I did some years (ages) ago was put the "broadcast" preset in the multiband compressor in adobe audition, and 90% of the time the clients were happy. Perks of working for a small local studio with mostly punk rock bands lol
I’m interested in music & audio, but didn’t know much about mastering.i found this fascinating and educational. Thank you!
Hot diggity, Mr Jordan! I do wonder which tracks I was upvoting from your double-blind tests. I've just installed Matchering and I am very impressed with the ease and speed of use. Choose your reference track accordingly. Thank you so much for bringing this all to light. Cheers!
How did you get it to work? I've been trying for 3 hours to no avail. The IP website does not work at all.
@@mo2cubing I went to the Matchering website and followed the links to install the Docker tools. I’m on Windows 10. What OS version are you running?
@@mo2cubing I should add that installing from the command prompt took a remarkably long time to process and following that I was a bit trepidatious at the multiple restarts to complete the installation. Maybe there was excessive server demand and you were 500’d (server error)? Try again in 12-24 hours?
All up the installation process took about 15-20 minutes for me. Good luck!
Thanks Ben, this was helpful. Getting ready to throw up my first album of electronic stuff, and I’ve been paralyzed by what to do in the mastering phase. You rock mang!❤
just tried out Matchering on a few mixes, it seemed to make them sound a bit thin and removed a lot of low end. I compared it to a quick master I did with some Gullfoss, Unisum, UAD Ampex for light saturation and Ozone Maximizer as the limiter, and I liked my master a lot better. obviously those aren't free (or cheap) plugins, so it seems like Matchering could be a good option for anyone who just needs a free, quick master. but if you want better results and have the money, I would still invest in mastering plugins and/or hire a (human) mastering engineer. good to see this is basically what the blind study confirmed too. maybe AI won't take our jobs.. yet
Thats the thing is that a lot of times you basically need to watch the low end really heavily when mastering, so an ai doing it I would say would make things too flat and not have the liveliness of someone "spicing it to taste" so to speak.
I haven't been to your channel since people were giving you a hard time over your gear reviews. You seem a lot happier with the content you're making now, and it makes me happy to see!
I wish I had a way of knowing which ones I rated high/low. I thought some of the masters were absolutely awful but I have no idea if I went with the general consensus.
Same! I specifically recall one of them being extremely loud and muddy, wouldn't be surprised if that were the one he described coming in last, I certainly rated it pretty poorly, a 2 I think.
CHANNEV by analog obsession is an amazing Neve channel strip AND it's free (with optional donations)
analog obsession also has an equally great low end expander called LOVEND! my favorite plugin by them though is probably Re-Life, I honestly don't know what it does specifically but it can really reshape the sound of your instrument with tighter lows and sparkly highs and is also extremely user-friendly!
Ozone has been the best for me for getting a reference of what could be done to a mix with references used and then using each module to see what I should fix in my mix down then toggle it off and go fix the mix where its needed
Benn, you guidance is very important and appreciated.
So frustrating that this video is so long but contain ZERO examples with the actual audio that is being talked about
But he shows disabled and enabled, would that not be audio examples?
Yeah. You could only hear a difference with headphones tho. Very short
@@Solunex i mean it still counts xD but fair enough lol
@@Solunexi mean obviously, especially with producer headphones, because normal headphones wont be very obvious
Given how bad copyright trolling for music is on yt, it's probably for the best
I feel like it’s tough to fully believe the outcome considering the mastering websites weren’t even included in the poll and the raw data is not made available. I want to believe it, but why can’t all 12 versions and the raw data be made available so people can really trust the results?
Exactly. I need to hear it first to believe the final result.
100% this - not sharing this data feels lazy at best and malicious at worst. Plus people who took part in the study (and others that would want to replicate it on themselves) don't hace a chance to learn about their own preferences.
I like the breadth of the subjects that you cover on your channel and not only that the depth You're also go into it to a level and explain things in a way that makes them understandable to everyone which really shows true mastery thank you!
Repeat noob question: should I cut everything below 20 Hz before feeding mastering plugins?
you should not apply any technique by deffalut and on all things.
Highpassing that low usually does-can do a lot of damage, depending on the song of course. So if you have a DnB song with lots of energy in the subbass you can destroy a lot more by highpassing it since the highpass filter introduces a phase shift which, (again depending on the mix) can completely alter the low end if its fundamentals reach very low.
Its safer to use low shelving filters and thin the critical regions out a bit rather than simply highpassing everything. and if you have to highpass better to do it on an individual channel manner and not on the sum before mastering.
Thats only my opinion of course
@saardean4481 Thanks for the reply. For mastering I only use a linear phase 20 Hz high pass for the mastering plugin input. My (noob) thinking is that this gets rid of inaudible sub frequencies and any DC offset from analog recording, to give the mastering plugins a better mix to work on. Just not sure about it.
@@Goettelthis causes more harm than good usually. Either a phase shift, or pre ringing or something else. And people do this thinking it saves headroom without checking their meters. I’ve never seen/heard this shave off more than 1dB without audibly compromising the sub. So as the other commenter said, rather stick to shelves. Even dynamic EQ shelves, or multiband with slow attack and release. Also, it’s usually best to use zero latency/natural phase on low end stuff. Linear phase causes pre ringing in the low end most frequently. But of course your ears + a reference track trump all the theory any way🤷🏻♂️
@@avationmusic Thanks for the reply, I'll experiment more.
This has always seemed weird to me. Is there even significant energy down there? Is it significant energy that isn't supposed to be there? What channel is it coming from? How did it get there? Can you eliminate it at the source instead of applying a band-aid?
It makes me think of tracks that have audible mains hum. You can put a 50/60Hz notch in there, but there are going to be harmonics that you have to notch out too. And before long, you're comb-filtering your track just to deal with a problem you could've fixed by... tapping the cord from your guitar a foot away from the power cable or something like that.
As a newer mixing engineer/artist this makes me so happy because it means I can spend more money on plugins
I got Ozone for $99 on sale and it has saved me so much time. I used to spend days even weeks mastering my music, now this gets me 80% there with the AI suggestions and I can tweak from there. And MAN can you tweak stuff. I'd use the phrase "Game Changer" unironically.
ozone is all i use. i mostly just make music for funsies and am not trying to make marketable material for an income, but i have been producing in one form or another since the late 90s. ive had professional mastering done, and have also just thrown a compressor on the whole mix and hit the "mastering" preset, and have tried just about everything in between. ozone just checks off all the right boxes for me. cost is right, and its highly tweakable, but doesnt take a phd in music to understand. i had a cracked copy initially to try it out, but promptly bought it after literally one track being run through it, and this was with a preset and before i knew how to actually use it. its deep enough that it will reward you spending time to learn it.
Yeah, I always got good results from Ozone. The presets are usually a good start, then you can tweak the Hell out of them. $99 is money well spent... :)
It "lifted the veil", yes?
Great video! Please make a part 2 video with audio comparisons of the masters and the exact reasons why each AI too failed?
And 90% of people then listen to streaming and over compressed music on an mobile phone with earbuds where you earn 0.00000000001 cent per play. But I get the point.
Which actually makes it more important. It's not too hard to make a track sound good on a high end playback system in a good listening environment, but it takes a real master of their craft with years and years of experience (and very accurate monitoring) to make it also sound halfway decent when someone plays a RUclips upload with 2 or 3 generations of lossy compression through their cell phone speaker at the other end of the bus.
Hey Benn. You're still the audio-tech-channel GOAT, in case you were wondering. Always setting the bar high, my friend.
Oh my god, I can almost feel the competence when you talk. F*cking love it!
3:15 while yes a bad master can ruin a good song in my option it’s significantly easier to learn the fundamentals of a good master than it is to learn any other part of the music making process. If you are a good mix engineer you absolutely can master your own music.
When someone cares about my music, I'll pay for mastering. For now, Logic's AI mastering does me pretty well. I used to try doing it myself, but I like being lazy and I was never that good anyway. Biggest complaint with Logic's mastering is everything is smoothed out too much. Still sounds good enough for most ears though.
There was a sale a couple of years ago and I mistakenly got elements as opposed to the standard version, and since I have it, might as well use it. I usually end up doing one of three things with it. Sometimes I run it and genuinely feel it's good enough after doing various reference tests and just roll with it. This usually works best on really simple tracks. Other times I will still use it but run some plugins before it just to change the character of it slightly, like a saturator or similar things. The third option, and probably the best use for it, I'll run it, listen to what it's doing, and then sort of use that as a loose guide to use what few mastering plugins I do have to get to a similar place but with my own tweaks on it.
Regardless of which I do, I always turn off the elements imager. Even the free version of ozone's imager gives you slightly more control than the single slider elements gives you, and I rarely use imagers to begin with.
Mastering engineer removing original comment: Sorry, guys…
I may well have recorded, performed on, mixed or mastered ≈ 2k5 (have no idea) published tracks over 35 years, and trained seven engineers (five kept at it). This summer I was offered a college level teaching position, but declined for health reasons. Youth isn’t eternal after all.
That doesn’t mean much - who knew, right…? I could be as silly as my clients could be - although there’s a serious discrepancy to that logic.
By all means - follow your own paths, I won’t be in the way. Aggression & hyper-competitiveness is quite abundant without me contributing to more of it.
Make music - that may help…?. 🎶💛
I think there's a bit more nuance than that. I'd argue that not adjusting for loudness is actually the right way to go, but everyone is releasing music for different contexts and that matters too.
On one hand loudness is an extremely important factor of a master. You said it yourself that anything louder will sound better. So generally the louder someone can pump a track without losing perceived quality, ie loss of clarity due to overcompression, the better.
The prevalence of automatic volume normalisation on modern listening platforms is a huge mark in favour of not caring about loudness. If you only care about Spotify and Apple Music then just hit -14 LUFS and call it a day. But many platforms don't have this feature, and many users of the platforms that do have it will just turn it off.
There's also the consideration that if you're making music that you're intending to play/have played in a club/party/festival context then you don't want your music to sound quiet compared to every other track. On a similar note, even if the DJ level matches between songs, a less compressed song played loud sounds totally different to a very compressed song sound quiet. One will just sound odd in the context of a set filled with the other.
The biggest issue with Benn's test is that he only did one single test, rather than a multitude of songs across a multitude of different genres. IMO only using a single song invalidates any conclusion that can be made from the test.
@@LunicussOfficial Your argument doesn't make sense. On the one hand you say its better to not adjust for loudness when mastering. But then right after that you acknowledge the importance of the music having been able to match radio or dj's other songs.
Sure a less compressed song sounds different, but this difference is not the same across the board.
"Over compression" is exactly that, it's too much compression. If your song is loud, or a part of your song, you don't need to turn it down. If anything, you turn one fader up and it's really loud compared to the others, that means the rest are almost certainly too soft.
What you mention is exactly the need for mastering. I understand that too much compression sounds like shit, sure as it's manipulating and squashing the sound down. But you don't need to do that. You should be able to adjust the volume somewhere else besides the mastering stage. There would be a few places you could increase the lower ones.
@@LunicussOfficial I think you missed my point. As professionals, when we _compare_ mastered and unmastered versions of the same track, we are NOT comparing loudness. That’s the end result, and relates to the target platform or media, their guidelines or how they can be… exploited.
As an example, Jordan played bits of audio - unmastered and mastered - where the mastered ones sounded “louder” on playback. Sure, fine when said track is competing with others out there.
In the studio, any tonal, phase issues, clipping or dynamic wrongdoings will go unnoticed if the master appears louder than the mix - when _checking the integrity_ of your result. Simply getting “impressed” by your own processing doesn’t cut it, as you risk ruining your client’s hard work.
The client - and future listeners - will hopefully be impressed. That happens “out there”, not in the hopefully (!) controlled environment of the mastering suite.
Jordan mentions many of the same issues. Don’t mistake “nuances” for taste or aesthetics - surely important for the artists and their listeners. The technical aspects and limitations however, are the same across genres and taste.
The simple difference between a calibrated monitor & room, and a speaker, is that the monitor reveals what a typical speaker wouldn’t. A good master will translate well to a phone, a mono BT speaker, earbuds, a full blown car stereo…
That’s not about loudness of the master itself. The encoding used by a distributor, or limitations of physical media and playback equipment come into play (bad pun), and are known by the mastering engineer.
Silly analogy; people wouldn’t like their taxi-driver to “hate them” or drive recklessly to “compete” with an Uber? Hopefully, (s)he’ll treat any paying passenger with respect, knowledge beyond the GPS, and without judgement on where you want to go.
‘Nuff nuance. Make music… 👍
@@louispconstant6624 I said "not adjusting for loudness" with respect to Benn's testing methodology, not mastering itself.
@@LunicussOfficial What they said^
How am I just now discovering your channel!!? This video is GREAT.
The Ozone is a great mastering tool. You don't really have to tweak it much after it analyzes the song if your mix is already good
Dayum, that Polyend sounds good! Those Plugin demos were also lush as hell. Great video!
Benn, Thank you. I am not in a position to reward you financially at this time but I do really appreciate the integrity and good will of your channel. Would love to see more of your playing too!
Everyone needs a Benn as a frenn ❤
7:00 Compiling a python program? '-'
skill issue if you can't compile python
@@Andres-Estrella It's less about whether you can, and more about whether you should.
lol non programmer people aren't gonna know it's an interpreted language!
Landr has been useful for my band. Just gives an idea of how mixes might come out. Before sending the songs to someone who is good at mastering. Also helps demos sound a little more clear. The professional definitely sounded better than Landr. Imma make my band watch this video though. I bet they will enjoy learning some of this.
So the free AI thing was almost as good as the very, very expensive mastering engineer with all the fancy analog gear...?
Playing piano and trumpet and flute and guitar and having been in high school and junior college band/jazz-rock ensemble and taken electronic music classes (1978) and having had a buddy go to the Berkley School in Boston and now 45 years experience in the high-performance computing industry, I appreciate everything in this video, and I am not at all surprised that AI did not come out on top...
Most of A.I. mastering for me was just exporting it, shooting it through, tweaking the original and exporting it again. After a certain point it just became like another mastering VST, one in a big chain. Lots of stuff didn't really benefit from it, like the raw stuff I'd make in Audacity or field record, but the electronic stuff I'd do turned out alright. Most of it was just, like, a motivator to finish the thing. It's a fast thing that makes the last part of the process less of a nervous uncertain tweak of a knob and more of a climactic loading screen that just shouts "You did it!"
of course ben is a professional engineer he is going to shit on AI mastering, he needs his job😂
Outstanding video! Very entertaining and informative simultaneously!!!
Half of this video felt like a big ad. The title of this video implied that the emphasis would be on the experiment and industry trends when it was mostly just a tutorial.