Preach it Wytse! My only comment: album normalisation is still preferable even with a random playlist. If you use the loudest song on the album as the reference then you can't game the system by making the rest of the album quieter, but songs that were intentionally mastered quieter remain quieter for the playlist as well.
@@DanWorrall I have Tidal and what I have noticed it if you mix genres in a playlist you're kind of asking for trouble if you expect normalization. Your playlist may vary.
Re: 6:36 "I don't think you need dB LUFS because it's a loudness unit." Interesting point. I had previously interpreted "LUFS" (Loudness Units Full Scale) to refer to the full-scale loudness level, so "dB" was needed so that the user knows "-14" means decibels relative to full scale (and not, say, -14% of full scale). But the AES's Technical Document AESTD1008.1.21-9 on loudness for streaming backs you up: "A unit of loudness difference (LU) is equivalent to a decibel (dB)." And throughout the document they use phrases like "-14 LUFS" and "-20 LUFS" without "dB". Thanks for bringing this to my attention. -Tom PS: The document is freely available online; do a search for the doc number.
I’m glad you touched on dynamic range because that’s exactly why I target -6 to -5 LUFS is because it’s the only way I can know if I’m getting the kind of limiting and clipping on my track that the tracks I’m referencing are getting since I’m looking at those track’s numbers. It’s not because I want to be “louder” necessarily. It’s so I have something to bass my dynamic range off of.
For those clients who say "but I want it louder because of my Genre", we could also explain that after Spotify turns them down, they have just sacrificed all the punch and impact of their Kicks and Snares for no good reason, and could get that Punch back by mastering in the sweet spot of the dynamic range.
And if they don't believe you, show them an A/B test where you take there master and cram it up to the top (with no dynamic range left), then turn it back down afterwards (like Spotify would), and A/B those 2 versions.
@@JimhawthorneNet You lose nothing when it gets normalised. If the kicks and snares are mixed with punch in mind, normalisation will not ruin this. At -14 LUFS, it will still sound perceivably louder than a skimpy track that is mastered much lower.
@@Kiloeve good mix/master will still sound good, but the less it complies to standardization, the more it will be altered by the streaming service. So if it's our target, what's the point?
@@---pp7tq If it's being altered in a linear way (assuming you're above -14 LUFS), I would've thought people stopped caring as much and just focused on what is right for the music, but I am just naive to assume that.
It's just using the devices internal gain to level the album (or playlist), there's no dynamics being adjusted with normalisation enabled only overall gain. so just master to whatever you want and it'll be loud if they have normalisaiton on (in comparison to the previous and next track played) and it will be as loud as you mastered it for everyone with it off.
My Dad refused to buy any products with loud commercials, it really aggravated him when an ad was extra loud and he kept having to adjust the volume. We used to think it was hilarious how much it bothered him
This was sort of a game changer for me. I went back and realized I had turned normalization off on Spotify and forgot about it. I’ve been spending so much time trying to make my songs as loud and present as others. But now with normalizing they all play at the same volume and I don’t have to destroy my tracks anymore 😅
thanks dude, i easily got dragged into the loudness war mindf*ck over ten years ago with my own music when the ipod phenonemon took off and i made some horrendous sounding stuff for internet delivery , but whilst dragging some old mixes from DAT tapes i realised how much my mixes suffered during that time, I'm glad i went down that rabbit hole because i learned alot about my equipment but still, the masters i made for vinyl production still hold up better because i at least knew the amount of dynamics required for that form of delivery at the time. i am happy the streaming services are using this method to control overall volumes as it was very frustrating feeling the need to compete against the 'pop' world of loudness when most of the source material i was mixing from was very lo-fi and sample based to start with. i don't watch alot of youtubers in the mixing /mastering sphere but your channel is one i always check for, your voice is a mainstay in my kitchen and my family always recognise when i am listening to one of your videos. thank you for all of the great info and advice you have provided for years. x x x
Two mixes at - 9 doesn't automatically mean the perceived loudness is the same, one mix could be perceived way louder, it's not about LUFS only, but balance, crest factor, dynamic, frequencies etc the whole LUFS discussion is bs!
Thank you for such a straightforward explanation of the subject of how streaming services handle uploaded material. It would be nice if everyone could discuss things in life in a civil manner, without berating, or attacking each other, and/or attacking each other's point of view on a subject. Until then, please continue making these wonderfully informative videos. They are appreciated.
bandcamp doesnt do any loudness compensation to my knowledge. in general i'm a fan of the fact that bandcamp does its best to just be a hosting platform and not try to do any algorithm or social platform nonsense on top.
@@maegnificant The moment it does recommend stuff to you based on your and other people's likings, it becomes a recommending algorithm like on the major social medias, instagram, youtube, etc.
Yooo about humans being more sensitive from the back, I've totally experienced this, when I go to bed I usually like to play some youtube video about music production or something like that on my phone while I try to sleep, so I just leave the volume really low as to hear it while also being quiet enough as able to sleep, but when I turn my back on the phone the sound just feels way louder and then I have to adjust the volume again
Might be that you listen more for what you can't see. or that your acoustics make it feel like its louder. or a survival instinct to be ware of potential danger behind you. maybe all of those. :P
Hah! I do that as well. Without a nice podcast in my ears, I suffer terrible nightmares. I subscribe to several audio production podcasts of course (The UBK Happy Funtime Hour was the GOAT), and mostly my dreams are about walking around LA and around studios with mostly American dudes talking about music production, songwriting, gear, history of the recording industry etc. Funny thing is, sometimes I come across a very specific and difficult audio production problem to solve, and the answer will immediately come to mind and make it's way out of my mouth in words I did not know I was capable of speaking. So I guess subliminal learning works.
I'll start with saying that I'm 100% with you on the overall message, and that "turning off" the feature on Spotify is a bad idea. That said, it's not completly true that getting the signal at different levels is the same, because you'll just turn it up or down. Many systems are built in a way that sound correction will tend to zero as much as you pump up the volume. For example, with my technics and pioneer amplifiers, if you have loudness on and eq boost to move poor speakers, the more you pump up the volume the more the signal will be back being direct. The reason is simple, if an user would just crack it up because he's having a party, without that feature, he would blow up the speakers.
Love the vid man! This is a great topic. On top of being in music production for a long time im also involved in other audio related hobbies. Car audio beimg one of them, i highly recommended leaving loudness normalization on. Reason being when we tune car systems we we used various test tones at various frequencies to set the gains correctly so there is no clipping. We also spend alot of time balancing each media source so that when you go from say the radio to a cd or a Bluetooth device that the overall volume stays close in each. This helps ensure the system doesn't go into clipping when switching media sources. Messing with loudness settings on these streaming services could result in a lower clipping volume due to the signal becoming louder
Very good follow up video, I was one of the guys that thought you deliver at - 14lufs and wasn't understanding how can you do it everytime with every genre. I think this video really clears up everything you need to know about this subject!
I was discussing this with Ian Shepherd recently as I too was confused that the AES target isn't really a target, it's the minimum loudness it might be heard on the users device, so a loudness floor really. Ian said he's advised AES to not call it a target anymore because clearly if you put a -7lufs track into its own playlist and turn on normalisation, it's still going to play at -7lufs not to the loudness 'target' of -14lufs. It might theoretically be attenuated to -14 if there's enough quiet music alongside it I suppose, but I don't have any playlist showing -14 other than UA-cam playlists.
I'm sometimes wondering if LUFS is a good way to measure loudness of a track. If I listen to a metal record mastered at -14LUFS and a pop track mastered at -14LUFS, I often have to change the volume anyway, because they don't sound equally as loud. What are your thoughts about this?
Check out Sage Audios UA-cam channel , they have a recent mastering for loudness video. They are saying that a track mastered to say -9 LUFS will sound slightly louder that a track mastered to -14 LUFS even after being turned down by the streaming platform and that things like maximisation can make one track sound louder than another even when they have the same LUFS value. When a track is pushed too hard and normalised by the streaming platform, it will actually sound quieter though. This is contrary to what I learnt at audio engineering school where we were told that anything above -14 LUFS will be attenuated by the streaming platform and will always sound quieter. Realising alot of what I learnt is actually technically incorrect.
Contrary to what some may think, you CAN actually capture (Record) Spotify tracks. And I've played back some of these LOUD tracks and discovered they are in fact hitting -9 LUFS which is commercial CD level loudness. I've noticed this on commercial artist tracks and not on smaller or amateur artist releases. So, Spotify can say whatever they like, not all tracks are being streamed equally.
@@7thResonance I would also be very interested in that. My tests also came out below -13.9 LUFS. Only when turning off the normalisation I get results way above that level (which should not be a surprise, especially after this video).
Coming from mixing live TV, where we work to either -23 or -24 LUFS (depending on country usually), I am pleased to see that music is using the LUFS standard. -14 LUFS though is incredibly loud. While there are reasons that DAW produced music can achieve this and consumer playback devices with weak amps will require it, what opportunities within your mixing are you missing out on in the 10 LUFS difference between TV's standard and the music streaming service standard?
I’m absolutely loving that the music streaming services are adopting deliverables specifications like have been in place for some time in the tv & broadcast world. I’m loving more that there seems to now be a global standard between streaming music and television/film emerging.
Great video and subject matter. I was caught between what I want my album sent out as, but you made it extra clear with providing a "Streaming Ootimized" version. Thanks man!
Hi! I work in the classical music area (mostly). And a classical piece might suffer in quality even at a "low loudness" as -14. You want the dynamic and almost no active limiting. So, if the master is -17Lufs. Will Spotify really possibly damage the sound quality by using an own limiter to come to -14? I heard you suggest it a few times. As far as I know, they only raise it to -1tp. But they don't limit or compress. The Lufs remains below -14. Am I wrong?
For those who didn't know, Daft punk's RAM album was mastered to -14 LUFS and won a Grammy for Best engineered album. It's also considered one of the best albums for audiophiles. So there's that.
Loudness normalisation (if enabled) will make as loud as a track mastered to 0db, the difference comes when users have normalisaiton turned off, then you'll see tracks on Apple, Tidal, Amazon etc hitting -6 or even -5db.
@@vigilantestylez So, I just went to Spotify to run a little test with this album. When I turn normalization OFF the album definitely jumps up in volume. Theoretically, the Spotify algorithm shouldn’t be doing anything at all, right? What’s going on here?
Great video mate. I've been covering this for a while now and it's got to the stage where explaining Loudness Normalization now feels like a Musical version of a Haemorrhoid!!! I still feel it's challenging when all of the platforms have different Normalization Levels which makes it more beneficial to Master to 'at least' -14LUFS Int but there's some much room between -5LUFS and -14LUFS so why not use it and make the track sound better! Keep the great content coming mate.
Another thing I do for clients is include loudness-matched reference tracks. I already have reference tracks in the mastering session, set to -14 LUFS, so I'll just export those reference tracks along with the client's tracks and tell them "This is how it's going to sound on streaming playlists."
I found myself nodding and laughing at your rant. It also made me check my settings, because you're right; I had no idea if it was on or off (it was on and I left it on).
When you mentioned mastering to louder than -14, what do you do to deliver it to the client optimized for streaming? Are you simply turning down a -11 lufs master 3 db? Sorry if this is an obvious thing-- just making sure. Wanted to be clear on the difference between the loud master and the optimized master you presented to the clients in your example. Also, love your content-- thanks so much for making these!
Finally someone who understands and also explains it well! i master my tracks -9 lufs (cd standard) for listening to my customers on their smartphone or whatever so it is perceived louder and additionally a -14 lufs version for streaming. I then always explain it to my customers as well. No idea why there are so many people who do not understand, apparently it is despite all efforts for some people not understand what is loudness normalization. Most prefer to listen to youtuber or instagram influencers who adorn themselves with singer credits for which they have mastered. A good example is someone else from England who is just using half-knowledge and misinformation to give misinformation to young producers (e.g. turn off TP-limiting). At the end of the day I don't care what anyone else does, I have stopped trying to explain this to people, some people are like a brickwall limiter when you try to explain it to them. If people like to take the extra distortion they should master -2 lufs and 2 db peak or something just to revive the loudness-war of 00s. Sometimes you don't hear it. I, on the other hand, stick to the guidelines and see no reason why I shouldn't do it.
Thanks Wytse! Very interesting! But... what if I upload a track (just one, no album) which is mastered -1 dB true peak and it has let me say -20dB LUFS or even less? What does Spotify then to "compensate" the missing loudness? They would have to limit the track as normalization won't work here. So.. quiet tracks would be really changed in the sound of the track.
Ok I got the answer for myself when watching again and read the spotify paper in full screen :D... as Spotify only normalizes, they would do nothing in this special situation as thye cannot go above -1 dB TP. Means... When I or anybody else prefers very quiet masters he or she should watch out to limit way less -1 TP when this will go to a streaming service that the normalization can "work"? Otherwise you could (obviously) master the song louder, lol. But there will always be a slightly difference in the dynamics then. Seems to be a lose lose situation :D
I've had songs at -6LUFS that sounded thin and phasey, and I've had songs hitting at -11LUFS that were in my face. The numbers are really only good for target matching. Not something I overthink, just something to consider. "Do you have a similar tonality and perceived loudness to your competition at the target level?" That's really all that matters.
step 1, get the level of energy you want from the track. As loud as you need it to be to feel the way you want it to feel. step 2. make sure its not too quiet. roughly speaking. this is what i do. my tracks end up at different loudness this way. and they are loud enough but to the point where it sounds right to me.
Yep, the loudness of two tracks at - 14 Lufs could be perceived differently, one could be perceived louder than the other, so the focus on Lufs as to loudness is simply stupid
I think Mr. Bill's take on loudness and why genres like EDM and Metal mix so loud. It's the insecurity of not having your painstaking details heard. Of people not having their systems at a proper volume level to understand the intent. I'd argue that nobody will understand your intent as an artist wholly, regardless of how many noticable details there are.
Metal isn't "mixed loud"... it actually IS loud. It's arranged that way... same thing with a Hans Zimmer orchestral arrangement... or heavy dubstep... it IS loud... you can't just put a mastering limiter on it and slam it... that will never work... but if the waveform is full like pink noise (plus huge sub bass) for any duration, it's going to check in very loud... even up to -3 LUFs at the loudest parts, WITHOUT audible distortion. You can actually "write" loud.
Just because the genres bake mixing into their sound design and arrangement doesn't mean it's not mixing. And Mastering is different from mixing. Usually in those genres you use clippers anyways.
@@ethangregorymusic Yeah to me mixing and mastering are all part of sound design and vice versa... I guess it's a tough thing to describe and should probably be talked about on that level more often. If you clip and saturate your sources, mixing them together to make a fat wall of sound is pretty easy.
@@ethangregorymusic I think the point I was trying to make is that without complete control, as the producer, a mixing or mastering engineer probably views loudness a bit differently than the people that focus entirely on making loud music from the get go. Again, it's a different conversation but these videos almost never touch on it because the people making the video are focused on mixing other people's music.
Thank you so much for this discussion. I'm a noob, just finished my first master for a client that I recorded and mixed. But wondering about what Lufs to deliver. I had it at -14 on the money, and -0.3 peak, but not sure if that's a true peak. I checked with Loudness Penalty to get things perfect for streaming. But worried its not loud enough, as a problem I had as a noob artist, when I ordered my mastering and didn't understand compression except that bad thing UA-cam does, I asked for "No compression, preserve the dynamics." It sounded good until it was on Spotify in a playlist and I realized I could not promote it because it sounded quiet and weak compared to everything else in the playlist. So for my clients, I want to avoid having them go through that experience. After watching this video, I think I'll go for -11 Lufs which sounded good in my DAW.
I usually seem to end up somewhere between -9 and -11 LUFS, and call it good. However, I've to date left the TP at -1 dB. In your example you didn't actually change to -2 dB when it was boosted for dynamic range reasons. Is this something you actually do for "optimized for streaming" masters? When does that actually make a big difference... I assume -13 LUFS doesn't suddenly make -2 dB TP necessary, but at some point it likely does, right?
Hi Wytse! The theory you talking about is absolute correct. But if I stream from Apple Music Adele - "Can I get it" my T.C. Electronic analyser displays a program Loudness is -9 LUFS. Elton John & Dua Lipa-Cold Heart (PNAU Remix) is at 8.7 LUFS. Apple Music does not reduce anything. Even my own production Phonofinish - Dance with me has -10 LUFS (which is the original loudness) at Spotify. I can not see any Loudness reduction. Cheers André
Hmm, strange I got these numbers from my lufs meter while playing the songs on Apple Music 🤔 Adele - Can I Get It -15.9 Lufs,True Peak -6.12 Cold Heart PNAU Remix -16 lufs, true peak -7.14 Phonofinish - Dance with me - 16 lufs true peak -5.25 (get a weird spike just in the very beginning -0.10 true peak but after resetting true peak is around -5 -6)
The comment about high dynamic range sounding weird is a bit misleading, because it's highly dependent on the musical content. In classical and acoustic music it's quite common to have large dynamic range. I think the whole discussion should be focused around the music rather than numbers. When people talk about "loud" they are thinking of a full sound that blasts you continuously (like in metal or heavy genre), high LUFs is a result of that musical idea, and not the goal in itself.
@@analogkid4557 Yeah I don't like the low dynamic range metal style usually, although I think it can make sense in some cases, e.g. for something like Electric Wizard's Funeralopolis (but it would probably be better with a little bit more DR, less irritating).
Very insightful, loud commercials are sadly still a thing sometimes, especially the ads on youtube. On TV it has improved but mid-roll ads on youtube still have that extreme loudness in some cases.
I think this is because adverts - being short - just don't have quiet bits or silence like longer programmes, interviews, or podcasts. Adverts don't have the wide dynamic range of movies, for example. They might have the same peak volume, but the short term average is typically higher, or at least feels that way because there are no quiet bits in adverts; it's just totally in your face for 30 seconds, like someone is shouting at you.
Hi Wytse! A little off-topic (but not so off).: Are you going to review the new Slate Digital FG-X2? I have the fisrt version, and version 2 it is free for former users. I think would be cool for a video (and even better if you could run version 1 to compare to version 2). Thanks!
Do you target a similar LUFS level for YT as well? I see that this video has been normalised down by 1.1db - I'm wondering if being over the -14 LUFS matters as much for spoken word / gaming YT content. Thanks for the video, very interesting :)
I always master to -10LUFS and they still sounds great on the streaming platforms. I hear no difference between my masters and what shows up on Apple Music or Spotify. The integrity remains so I’m sticking with it. An engineer could go crazy trying to master for each form of media.
Thanks a lot for this video! I have always mastered my tracks in 2 versions, one at ~12 LUFS and one just below 14 LUFS because of the Spotify thing. Now I understand things better! Thanks a lot mate! No more 14 LUFS versions!
The key phrase here. The KEY phrase "Streaming Optimised Masters" Not -6LUFS with -0.1, Glad you are on board here. Im done bashing my head into the wall. One thing that's not clear here, You mention being at -11LUFS but that is not an issue, That totally right. As the guide is for the average measurement over the course of a track. Not just a short section, So we could go to a break down section and be say -16LUFS for a period of time. This bring the average of the track down with it. You may have a moment of loudness at -5LUFS but still have the track average at -16LUFS (Just as an example)
As a (casual) listener, I had loudness normalization turned off, because I did not really have faith that it was being implemented in a 'correct' way. Indeed I thought: "Surely the mastering engineer would know better than some corporate type in a tech company who wants to capitalize on the fear of the loudness war among their subscribers". Now that I know that there is a standard and that it is transparent I will turn it on. Thank you.
I will be happy with I clean the Audio and not that the Audio burns only to be loud.. I think that such Audio can last and as an artist I will make it loud when I want and I will also make such a mix but i cam do Master my self but i want Artist idea to be free not captured in Forum , to make my music stand in line with other similar ones that burn due to the neglect of -1 -2db true pick and crumpled dynamics. This video is finally something concrete that I can attach every time I send music for mastering.. great video and i was stunned with Master Engineers who have expensive equipment and do not know the basics. Thanks
VERY interesting video! Does targeting accurate intended LUFS require playing the track for the full track duration? That was my understanding with Ozone 9, then Ozone 10 took away the ability to toggle on/off LEARN Threshold, and Ozone 10 LEARN now auto-shuts-off after only 5 seconds!
I was thinking about that a lot the last time. Thank you for the inspiration! Last year I started mastering everything to -14LUFS -2db TP. And in a way I figured out for my self, that's most balanced, between dynamics and loudness. Maybe because 12dB is a double of volume or so. That is the maybe best depth for speaker use. Now, the Dancehall Artists I worked together with, gave me reference tunes mastered on ..!!!! -0.1 dB TP and -8LUFS or -9LUFS. So, they haven't been very happy with the -14 -2 mastering... In the last weeks I updated my studio-setup and started mastering all the tunes again to -10LUFS -0.2dB TP. I was really not in the mood to push it further cause everything louder starts to distort in a way that destroys all the real beauty of the production. Might be useful with big amplifiers but for home use its worth to discuss... Thank you also for this loudness video with all the penalty thing last year, it helped a lot! Blessings
Very good work and good video. I understand that streaming services need the -14 LUFS. But what I dont understand is the following: When Iam buying music on amazon for instance for dj-ing , then the songs are mastered at -5 LUFS . An they even clipping at 3 TP. And that are artists like Matrin Garrix and Jay Hardway. So, when I want to have a version for DJs to buy and use, then I have to do it the same way? Otherwise the DJs need to trim my tracks louder everytime? I dont get it. Please explain it to me. Thank you 🙂
Thanks for your patience and continued explanations on this topic, Wytse. It apparently is one that continues to vex people without formal training and expertise, and your expert understanding is very helpful. Greetings from Southern California. 🧠
I'm a professor and researcher in neurosciences, and this reminds me very much of several fundamental principles I teach graduate and medical students about human brain function where many seem to not be quite reaching that level of full understanding. Don't get me started on the topic of resting state functional "connectivity." It isn't connectivity!!!! 🤦🏻♂️😉
thank you for making awesome videos! I was looking for someone to explain bit depth when mastering.. I have seen all your videos regarding sample rate, but I would like you to explain bit depth please! I always work with 48k sample rate and export my mixes at 48k 24bit (for the extra headroom) and for mastering most of the time I render at 44.1kHz 16 bit because that's just what used to be the standard for CDs... but with streaming services being the mayority of where listeneres will consume our music.. what would be the advantages or disadvantages of mastering at a higher bit depth?
I have a question, not releated to this sublject but another that you may be able to throw some light on. Do you think a music studio should be kept at a certain temperature? I have a purpose buily insulated studio in my garden. However in winter when i first go in it can feel very cold and the equipment to touch feels cold. Do you think it would be a good idea to keep an ambient temperture all the time. What would the benefits be over it being cold etc etc. Thanks for an awsome channel. :-)
You mentioned that the master should always be above -14 lufs in order to be normalized to -14 lufs. However, given that the peak is set to -1 dbfs I would imagine you need to master above -13 lufs. Mastering to -14 lufs would result in the track being adjusted to -15 if I'm not mistaken.
100% agree. Just saying, what I've once heard some engineer saying about listeners turning off the normalization: "the kids don't get their earphones as loud as they want to listen" Which I think is plausible in some cases. Anyway that might just be a reasons why some people might turn it off. Your point is still correct and everyone should calm down with pumping up. But honestly I feel like we are on a good path. Take care tot ziens
Thanks for this valuable information! Quick question - When looking at say, SoundCloud, looking at the meter of the track in SoundCloud, some tracks are way under or all over the place and some tracks have it hammered all the way to the top. Is this SoundCloud applying a limiter to their tracks? Some of these sound good, while others have some weird dynamics going on. The tracks that are way down, don't sound as loud.
Wearing headphones and listening to a playlist - a song plays at -24 LUFS - annoyed, you turn it up - the next song plays at -2 LUFS - instant hearing loss.
Because ads aim to stick out and grab more attention. Listening to the sound the ads have for hours is something nobody would enjoy. probably. Regular content creators usually prioritize more dynamics for a natural sound as compared to the radio style commersial voice overs
Well that's because most creators rarely know much about audio. Certainly those talking into a mic, UA-cam doesn't apply lufs targets to typical videos unless they are official song releases (the ones labelled "topic"
9:44 ok thanks for the video but what about the - 2 dB TP level, if the master is louder than -14 LUFS, because of the possible „extra distortion in the transcoding process“ (quote end)?
Atmos loudness is measured using the format in 5.1 fold down loudness and cannot be delivered louder than -18lufs. There's no loudness target for the binaural headphone playback on Atmos devices like Android or Apple, but the headphone loudness will generally fold down and increase another 2-3db depending on where the loudness is in the Atmos space. There's no single speaker format for listening so loudness largely depends on how you're listening to it and on how many speakers etc as it folds down from the master..
I deliver my mixes at -10 LUFS. I know some people that go even louder. Like Michael Brauer who mixes are as loud as -8 LUFS. Just goes to show that it really doesnt matter if it sounds good.
If the listener’s delivery codec distorts or clips on overshoot above the TP spec per platform/codec, then that distortion or clipping is baked in to their output. The only way I can control that is by staying within published LUFS and TP range.
I'm wondering about this : why / how is bringing down our master versus providing the streaming plateform with this famous optimized file, different/ better ? Also: im aggregator for a digital distributor (believe digital) that puts out music that I provide on all streaming plateforms. They work only with one audio file. So what's the point of these optimized for streaming services masters ?
I love your videos and you absolutely nail the explanation regarding normalization. However I would point out one thing that I've come to learn regarding mastering minimalistic classical music and especially solo piano tracks. It is not always necessary to aim at -14 LUFS (integrated). Less will actually still do bc going 'all the way' up to -14 LUFS in most cases will require some sort of limiting which is not always ideal dealing with very delicate acoustic recordings. It is better to aim at preserving a high dynamic range and avoid limiting or do heavy compression. With such tracks it is a good idea to mainly look at the 'short term' loudness and secondly try to go as close as possible to -14 LUFS using a non destructive approach (which would include no limiting). What I want to point out, is that you're not necessarily 'in a dangerous zone' being below -14 LUFS when it is very intimate and dynamic acoustic track. You can easily be a little below hitting a loudness penalty of +0.1db (or a more) and still have it working wonderful on Spotify - Lastly, thanks for making these videos and your honest opinions - I find your insights very useful.
-7 lufs seems to be where all the main music streaming services are settling their delivery to your device with normlisation enabled (where available). I measured the same track played back and none of them observe -14 loudness normlsation, they're all just playing the track at the same level it was mastered, which for this track was -7.2 LUFS other than UA-cam. Tidal appears to be around the -14 ballpark though. Spotify Stereo: -7.0 LUFS Quobuz Stereo: -7.0 LUFS YT Music Stereo: -7.3 LUFS YT Music Video: -7.3 LUFS Tidal Stereo: -13.0 LUFS Tidal Video: -7.9 LUFS Tidal Atmos: -12.6 LUFS Amazon Music Stereo: -7.0 LUFS Amazon Music Atmos: -13.2 LUFS Apple Music Stereo: -7.1 LUFS Apple Music Atmos: -12.1 LUFS UA-cam Stereo: -14.1 LUFS UA-cam Video: -14.1 LUFS AAC File on phone: -7.2 LUFS
@@gregrodrigueziii8075 Yep. running the whole track through YLM threshold to -14 integrated. Loudness normalisation seems to be ad-hoc how it gets applied when not inside an album. Tidal and UA-cam seem to be normalising playlist too to -14, haven't seen any track anywhere neat -14 integrated through Apple Music, Spotify or Amazon though.
@@SamHocking Thats interesting. Specialy now that they have change for their usual replaygain. Its odd that before, they do appear to be in -14LUFS when I checked years ago, but thats them using replaygain and not actual LUFS algo. I wonder whats happening.
@@gregrodrigueziii8075 Could be that when not played as part of a wider album or playlist it was a bit random where they set the level? Left me none the wiser what's going on really.
@@SamHocking i dont know why they are making it complicated. Honestly for us in post prod since its been the standard for audipost for a long time, we just make sure our room is okay, speakers properly calibrated, and for some reason, if we mix with our ears, we are always near -23 or -24 LUFS int. target. Its a bit a differen topic, but, it really isnt that complicated, but for some reason measuring the streaming platforms will get anyone confused as to why there isnt seem to be a standard existing when there should be one existing.
It is enlightening to listen to very compressed masters back to back with more dynamic ones with the normalizers on, ie "sound check" at Apple....it really shows how poor squashed mixes sound....why not use all that glorious headroom for a snare to bounce etc....space is good! I like -16 LUFS even better. Music is so fun to crank up when there is a high dynamic range...thanks for the vid.
Thanks for sharing your knowledge and the great work you do. I am a bit old school, and I think there's a particular situation where I would take your opinion with a pinch of salt. I'm talking about mastering electronic music intended to be plaid by a human person in a club, techno, for example. In that situation, other tracks shouldn't play louder than yours. The biggest reason is that the DJ will have to increase the loudness; even worse, they will use the mixer gain, not the volume knob. I would like to know your opinion, but in the club scene, let's say, louder tracks are usually perceived as better or even easier to play. I'm not sure if there is "loudness normalization" yet if you are a DJ. Let's say that the normal is as loud as possible if you can keep a bit of dynamic range.
I appreciate you talking about this further, and seems you've read my comment, since you posted the same link about spotify (even if you don't quote me on your comments replies) ANYWAY... We agree Spotify normalize tracks for a better user experience, for the reasons you told at the beginning of the video. And this is GOOD. But this doesn't mean you must master your music with this level in mind. As i wrote in the other video, as engineer you MUST work for the best of the music and so for the listener. Certain songs, are good at -16 -14, while others sounds better at higher lufs, not because they are louder, but because the dynamic processing you apply change the sound pressure and the crest factor, and this change how the song is perceived. I have nothing wrong with services normalization, i'm against to master music because of EXTERNAL decisions, that has NOTHING TO DO WITH THE SONG YOU ARE WORKING WITH. Talking about the mid part of your vid (it's not -14) i'm getting confused. I mean, you made a video about that stuff, also a shirt, along with different quotes in other videos about that targets, and now... you do not follow your own suggestions... good for your clients, don't get me wrong, but you remind me politicians that say everything and the opposite... maybe you changed your mind? Why just don't make a video with that clear statement then? Would sound way more genuine, i mean, no one is perfect, everyone do mistakes, me, you and i think even god (that's why we are on earth probably). If your master is good and valuable you don't need another master for streaming, cause your master will translate well also over decoding. I'm doing that from a lot of time, and i never had any single issue, not even one, so why bother with it? I just work for the song, the artist and the listeners, when you know how to tackle everything there is no need for a separate version. @16:20 this is not called copying, but referencing @17:00 you forgot to read what i pointed out, on the same spotify link you showed. "The web player and 3rd-party devices (e.g. speakers and TVs) don’t use loudness normalization." Sorry but the conclusion is laughable, i don't wanna sound or be harsh, but you are saying that quieter music is a better production, this is wrong, again. You don't have to reach any number!!! Both loud or "quiet" you must LISTEN your music, and master it accordingly with how it sound the best, that's it, that's very simple!!!! But you seems to strugle with that idea, you keep to hold that "standard" as a god. First of all THIS IS NOT A STANDARD FOR MASTERING, you keep misunderstanding that, it's a standard for music normalization on a streaming service, this is completly another story. If an engineer is not able to use their own ears, but to follow numbers instead, to me he is not an engineer, i'm sorry to say that, but mastering a song, is not to open a GUI and try to bring your music into a green area, or reaching a number, this is called GAMING. Mastering is about LISTENING DISCLOSURE: Since we are in the f***** internet i want to point out that i have nothing against you, my comment is not toward into any kind of "war" or something, and i'm not an hater. It's just a pure and honest way to reply to your video, i respect you and your thoughts, i just have a different one
I too am a producer of the hip hop genre for 24 years now. I completely and fully understand what you're saying. I guess the answer to your question that you pose regarding telling people to turn off loudness normalization is this. And truly, this is the only reason I can think of. I just think that people turn off normalization so they can squeeze out every last drop of volume when most of us are sitting in a car more typically equipped with a stock stereo system which probably isn't very loud to begin with. I know my Subaru has a very poor stuck system and even maxing the volume, at least just until it starts to sound distorted still leaves me wanting more volume. Therefore, turning the normalization off would afford me a couple or few DB of a gain boost. With that said although, I completely agree that it doesn't make the track sound better at all. Just louder.
Preach it Wytse!
My only comment: album normalisation is still preferable even with a random playlist. If you use the loudest song on the album as the reference then you can't game the system by making the rest of the album quieter, but songs that were intentionally mastered quieter remain quieter for the playlist as well.
Good point! But the streaming services are not doing that right now... Right?
@@Whiteseastudio maybe Tidal, but I'm not certain.
@@DanWorrall I have Tidal and what I have noticed it if you mix genres in a playlist you're kind of asking for trouble if you expect normalization. Your playlist may vary.
@@DanWorrall Nobody is observing -14 or even -16 for Apple. UA-cam seem to because I assume it'sdone at the upload point rather than users delivery.
@@InsideOfMyOwnMind I've experienced this on Tidal as well, especially for some reason with black country, new road
"The listener is the one[s] that, basically, decide if our hobby becomes our job." Brilliant.
True 💯🧠
Re: 6:36 "I don't think you need dB LUFS because it's a loudness unit." Interesting point. I had previously interpreted "LUFS" (Loudness Units Full Scale) to refer to the full-scale loudness level, so "dB" was needed so that the user knows "-14" means decibels relative to full scale (and not, say, -14% of full scale). But the AES's Technical Document AESTD1008.1.21-9 on loudness for streaming backs you up: "A unit of loudness difference (LU) is equivalent to a decibel (dB)." And throughout the document they use phrases like "-14 LUFS" and "-20 LUFS" without "dB". Thanks for bringing this to my attention. -Tom PS: The document is freely available online; do a search for the doc number.
I’m glad you touched on dynamic range because that’s exactly why I target -6 to -5 LUFS is because it’s the only way I can know if I’m getting the kind of limiting and clipping on my track that the tracks I’m referencing are getting since I’m looking at those track’s numbers. It’s not because I want to be “louder” necessarily. It’s so I have something to bass my dynamic range off of.
@floating pants yes you are.
Thanks for elaborating. I had the misconception of having to stay under -14 LUFS at all time 😄
same here....
Lot of vids teaching that, too!
For those clients who say "but I want it louder because of my Genre", we could also explain that after Spotify turns them down, they have just sacrificed all the punch and impact of their Kicks and Snares for no good reason, and could get that Punch back by mastering in the sweet spot of the dynamic range.
And if they don't believe you, show them an A/B test where you take there master and cram it up to the top (with no dynamic range left), then turn it back down afterwards (like Spotify would), and A/B those 2 versions.
@@JimhawthorneNet You lose nothing when it gets normalised. If the kicks and snares are mixed with punch in mind, normalisation will not ruin this. At -14 LUFS, it will still sound perceivably louder than a skimpy track that is mastered much lower.
@@Kiloeve good mix/master will still sound good, but the less it complies to standardization, the more it will be altered by the streaming service. So if it's our target, what's the point?
@@---pp7tq If it's being altered in a linear way (assuming you're above -14 LUFS), I would've thought people stopped caring as much and just focused on what is right for the music, but I am just naive to assume that.
It's just using the devices internal gain to level the album (or playlist), there's no dynamics being adjusted with normalisation enabled only overall gain. so just master to whatever you want and it'll be loud if they have normalisaiton on (in comparison to the previous and next track played) and it will be as loud as you mastered it for everyone with it off.
My Dad refused to buy any products with loud commercials, it really aggravated him when an ad was extra loud and he kept having to adjust the volume. We used to think it was hilarious how much it bothered him
Your Dad and my Dad must know each other, LOL. :)
I'm the same lol
I always muted all ads - still do
He is destined to restore balance to loudness
All ads on TV in Europe must comply with European broadcasting uninon loudness standards. Not sure about the situation in the US.
Love your rant.... I been saying the same thing for years!! Fighting ignorance is hard.
You are totally right, man. I'm sick of explaining this to ignorant people that argue you that louder is better... They are stuck in 90's.
This was sort of a game changer for me. I went back and realized I had turned normalization off on Spotify and forgot about it. I’ve been spending so much time trying to make my songs as loud and present as others. But now with normalizing they all play at the same volume and I don’t have to destroy my tracks anymore 😅
thanks dude, i easily got dragged into the loudness war mindf*ck over ten years ago with my own music when the ipod phenonemon took off and i made some horrendous sounding stuff for internet delivery , but whilst dragging some old mixes from DAT tapes i realised how much my mixes suffered during that time, I'm glad i went down that rabbit hole because i learned alot about my equipment but still, the masters i made for vinyl production still hold up better because i at least knew the amount of dynamics required for that form of delivery at the time. i am happy the streaming services are using this method to control overall volumes as it was very frustrating feeling the need to compete against the 'pop' world of loudness when most of the source material i was mixing from was very lo-fi and sample based to start with. i don't watch alot of youtubers in the mixing /mastering sphere but your channel is one i always check for, your voice is a mainstay in my kitchen and my family always recognise when i am listening to one of your videos. thank you for all of the great info and advice you have provided for years. x x x
Well said, however, I want to know why measuring directly out of Spotify I still find some at -7LUFS etc?
-9 LUFS is the magic number for me!
Same 👍🏼👍🏼
Short term or integrated?
Two mixes at - 9 doesn't automatically mean the perceived loudness is the same, one mix could be perceived way louder, it's not about LUFS only, but balance, crest factor, dynamic, frequencies etc the whole LUFS discussion is bs!
Thank you for such a straightforward explanation of the subject of how streaming services handle uploaded material. It would be nice if everyone could discuss things in life in a civil manner, without berating, or attacking each other, and/or attacking each other's point of view on a subject. Until then, please continue making these wonderfully informative videos. They are appreciated.
bandcamp doesnt do any loudness compensation to my knowledge. in general i'm a fan of the fact that bandcamp does its best to just be a hosting platform and not try to do any algorithm or social platform nonsense on top.
Yeah. That's why I really can't use any of these streaming platforms. Loudness compensation is just horrible.
It is a social platform and one with an algorithm to discover music you might like, what do you mean?
@@TachyBunker it does not have an algorithm.
@@maegnificant The moment it does recommend stuff to you based on your and other people's likings, it becomes a recommending algorithm like on the major social medias, instagram, youtube, etc.
@@TachyBunker but it doesnt do that
Yooo about humans being more sensitive from the back, I've totally experienced this, when I go to bed I usually like to play some youtube video about music production or something like that on my phone while I try to sleep, so I just leave the volume really low as to hear it while also being quiet enough as able to sleep, but when I turn my back on the phone the sound just feels way louder and then I have to adjust the volume again
Might be that you listen more for what you can't see. or that your acoustics make it feel like its louder. or a survival instinct to be ware of potential danger behind you. maybe all of those. :P
Hah! I do that as well. Without a nice podcast in my ears, I suffer terrible nightmares. I subscribe to several audio production podcasts of course (The UBK Happy Funtime Hour was the GOAT), and mostly my dreams are about walking around LA and around studios with mostly American dudes talking about music production, songwriting, gear, history of the recording industry etc. Funny thing is, sometimes I come across a very specific and difficult audio production problem to solve, and the answer will immediately come to mind and make it's way out of my mouth in words I did not know I was capable of speaking. So I guess subliminal learning works.
I'll start with saying that I'm 100% with you on the overall message, and that "turning off" the feature on Spotify is a bad idea.
That said, it's not completly true that getting the signal at different levels is the same, because you'll just turn it up or down.
Many systems are built in a way that sound correction will tend to zero as much as you pump up the volume.
For example, with my technics and pioneer amplifiers, if you have loudness on and eq boost to move poor speakers, the more you pump up the volume the more the signal will be back being direct.
The reason is simple, if an user would just crack it up because he's having a party, without that feature, he would blow up the speakers.
Love the vid man! This is a great topic. On top of being in music production for a long time im also involved in other audio related hobbies. Car audio beimg one of them, i highly recommended leaving loudness normalization on. Reason being when we tune car systems we we used various test tones at various frequencies to set the gains correctly so there is no clipping. We also spend alot of time balancing each media source so that when you go from say the radio to a cd or a Bluetooth device that the overall volume stays close in each. This helps ensure the system doesn't go into clipping when switching media sources. Messing with loudness settings on these streaming services could result in a lower clipping volume due to the signal becoming louder
Very good follow up video, I was one of the guys that thought you deliver at - 14lufs and wasn't understanding how can you do it everytime with every genre. I think this video really clears up everything you need to know about this subject!
I was discussing this with Ian Shepherd recently as I too was confused that the AES target isn't really a target, it's the minimum loudness it might be heard on the users device, so a loudness floor really. Ian said he's advised AES to not call it a target anymore because clearly if you put a -7lufs track into its own playlist and turn on normalisation, it's still going to play at -7lufs not to the loudness 'target' of -14lufs. It might theoretically be attenuated to -14 if there's enough quiet music alongside it I suppose, but I don't have any playlist showing -14 other than UA-cam playlists.
-99 LUFS Balloons 🎈
YES! 😂😂😂
Hahahhaaaaa, theres a theme song for Loudness Wars the Movie 😄
You killed it!! genius!!
I'm sometimes wondering if LUFS is a good way to measure loudness of a track. If I listen to a metal record mastered at -14LUFS and a pop track mastered at -14LUFS, I often have to change the volume anyway, because they don't sound equally as loud. What are your thoughts about this?
Check out Sage Audios UA-cam channel , they have a recent mastering for loudness video. They are saying that a track mastered to say -9 LUFS will sound slightly louder that a track mastered to -14 LUFS even after being turned down by the streaming platform and that things like maximisation can make one track sound louder than another even when they have the same LUFS value. When a track is pushed too hard and normalised by the streaming platform, it will actually sound quieter though. This is contrary to what I learnt at audio engineering school where we were told that anything above -14 LUFS will be attenuated by the streaming platform and will always sound quieter. Realising alot of what I learnt is actually technically incorrect.
Contrary to what some may think, you CAN actually capture (Record) Spotify tracks. And I've played back some of these LOUD tracks and discovered they are in fact hitting -9 LUFS which is commercial CD level loudness. I've noticed this on commercial artist tracks and not on smaller or amateur artist releases. So, Spotify can say whatever they like, not all tracks are being streamed equally.
interesting, can you link some examples? my tests all came at -13.8 to -15 LUFS.
@@7thResonance I would also be very interested in that.
My tests also came out below -13.9 LUFS. Only when turning off the normalisation I get results way above that level (which should not be a surprise, especially after this video).
@@MysteryCheaterMan yeah ....
We need to know if you had loudness normalisation on or off
This is my favorite video of your channel! You are answering most of my master questions! Thx!
Coming from mixing live TV, where we work to either -23 or -24 LUFS (depending on country usually), I am pleased to see that music is using the LUFS standard. -14 LUFS though is incredibly loud. While there are reasons that DAW produced music can achieve this and consumer playback devices with weak amps will require it, what opportunities within your mixing are you missing out on in the 10 LUFS difference between TV's standard and the music streaming service standard?
Amazing video!
Best I've seen on loudness and thanks for all the articles in the description!👏👏
I’m absolutely loving that the music streaming services are adopting deliverables specifications like have been in place for some time in the tv & broadcast world. I’m loving more that there seems to now be a global standard between streaming music and television/film emerging.
Some good sound tech info AND pholosophies here Wytse.
Keep educating. 👍
Who says we can't educate our audience ?
Hey, can you give us a small hint? If my track has about -9 Lufs, how to get it to -14 Lufs?
Is it Impossible?
Good point. Please do!
Thank you for sharing the knowledge! Some people that think you do things out of malicious are ridiculous. I love your honesty and transparency ♥
Netflix is always 25% quieter and it makes me scramble for the volume when switching to something else.
Great video and subject matter. I was caught between what I want my album sent out as, but you made it extra clear with providing a "Streaming Ootimized" version. Thanks man!
Hi! I work in the classical music area (mostly). And a classical piece might suffer in quality even at a "low loudness" as -14. You want the dynamic and almost no active limiting. So, if the master is -17Lufs. Will Spotify really possibly damage the sound quality by using an own limiter to come to -14? I heard you suggest it a few times. As far as I know, they only raise it to -1tp. But they don't limit or compress. The Lufs remains below -14. Am I wrong?
Your advice on the "client-dilemma" was very helpful thank you very much for that !!! The two version idea is brilliant 🤯
I still experience tv advertorials louder than the program you try to watch.
That's because commercials and movie trailers/music have different ilufs requirements than normal TV material
For those who didn't know, Daft punk's RAM album was mastered to -14 LUFS and won a Grammy for Best engineered album. It's also considered one of the best albums for audiophiles. So there's that.
Loudness normalisation (if enabled) will make as loud as a track mastered to 0db, the difference comes when users have normalisaiton turned off, then you'll see tracks on Apple, Tidal, Amazon etc hitting -6 or even -5db.
@@SamHocking yes. Absolutely.
Fab album.
@@ZigUncut indeed.
@@vigilantestylez So, I just went to Spotify to run a little test with this album. When I turn normalization OFF the album definitely jumps up in volume. Theoretically, the Spotify algorithm shouldn’t be doing anything at all, right? What’s going on here?
Great video mate. I've been covering this for a while now and it's got to the stage where explaining Loudness Normalization now feels like a Musical version of a Haemorrhoid!!!
I still feel it's challenging when all of the platforms have different Normalization Levels which makes it more beneficial to Master to 'at least' -14LUFS Int but there's some much room between -5LUFS and -14LUFS so why not use it and make the track sound better!
Keep the great content coming mate.
Thanks for the excellent and informative content!
Another thing I do for clients is include loudness-matched reference tracks. I already have reference tracks in the mastering session, set to -14 LUFS, so I'll just export those reference tracks along with the client's tracks and tell them "This is how it's going to sound on streaming playlists."
you are just so amazing!!! love your attitude and I am learning so much from you ! pls keep it up !
I found myself nodding and laughing at your rant. It also made me check my settings, because you're right; I had no idea if it was on or off (it was on and I left it on).
When you mentioned mastering to louder than -14, what do you do to deliver it to the client optimized for streaming? Are you simply turning down a -11 lufs master 3 db? Sorry if this is an obvious thing-- just making sure. Wanted to be clear on the difference between the loud master and the optimized master you presented to the clients in your example. Also, love your content-- thanks so much for making these!
Finally someone who understands and also explains it well! i master my tracks -9 lufs (cd standard) for listening to my customers on their smartphone or whatever so it is perceived louder and additionally a -14 lufs version for streaming. I then always explain it to my customers as well. No idea why there are so many people who do not understand, apparently it is despite all efforts for some people not understand what is loudness normalization. Most prefer to listen to youtuber or instagram influencers who adorn themselves with singer credits for which they have mastered. A good example is someone else from England who is just using half-knowledge and misinformation to give misinformation to young producers (e.g. turn off TP-limiting).
At the end of the day I don't care what anyone else does, I have stopped trying to explain this to people, some people are like a brickwall limiter when you try to explain it to them. If people like to take the extra distortion they should master -2 lufs and 2 db peak or something just to revive the loudness-war of 00s. Sometimes you don't hear it. I, on the other hand, stick to the guidelines and see no reason why I shouldn't do it.
You taught me a few things here -- I always appreciate that ... and now I want that Sonible true:level meter!
Thanks Wytse! Very interesting! But... what if I upload a track (just one, no album) which is mastered -1 dB true peak and it has let me say -20dB LUFS or even less? What does Spotify then to "compensate" the missing loudness? They would have to limit the track as normalization won't work here. So.. quiet tracks would be really changed in the sound of the track.
Ok I got the answer for myself when watching again and read the spotify paper in full screen :D... as Spotify only normalizes, they would do nothing in this special situation as thye cannot go above -1 dB TP. Means... When I or anybody else prefers very quiet masters he or she should watch out to limit way less -1 TP when this will go to a streaming service that the normalization can "work"? Otherwise you could (obviously) master the song louder, lol. But there will always be a slightly difference in the dynamics then. Seems to be a lose lose situation :D
I've had songs at -6LUFS that sounded thin and phasey, and I've had songs hitting at -11LUFS that were in my face.
The numbers are really only good for target matching. Not something I overthink, just something to consider.
"Do you have a similar tonality and perceived loudness to your competition at the target level?"
That's really all that matters.
step 1, get the level of energy you want from the track. As loud as you need it to be to feel the way you want it to feel. step 2. make sure its not too quiet. roughly speaking. this is what i do. my tracks end up at different loudness this way. and they are loud enough but to the point where it sounds right to me.
Yep, the loudness of two tracks at - 14 Lufs could be perceived differently, one could be perceived louder than the other, so the focus on Lufs as to loudness is simply stupid
How lovely that the industry decided to give consumers the “courtesy” of preventing hearing damage all on their own!
I think Mr. Bill's take on loudness and why genres like EDM and Metal mix so loud.
It's the insecurity of not having your painstaking details heard. Of people not having their systems at a proper volume level to understand the intent.
I'd argue that nobody will understand your intent as an artist wholly, regardless of how many noticable details there are.
I agree. I mix metal and I try to keep DR 10 db or higher.
Metal isn't "mixed loud"... it actually IS loud. It's arranged that way... same thing with a Hans Zimmer orchestral arrangement... or heavy dubstep... it IS loud... you can't just put a mastering limiter on it and slam it... that will never work... but if the waveform is full like pink noise (plus huge sub bass) for any duration, it's going to check in very loud... even up to -3 LUFs at the loudest parts, WITHOUT audible distortion. You can actually "write" loud.
Just because the genres bake mixing into their sound design and arrangement doesn't mean it's not mixing.
And Mastering is different from mixing. Usually in those genres you use clippers anyways.
@@ethangregorymusic Yeah to me mixing and mastering are all part of sound design and vice versa... I guess it's a tough thing to describe and should probably be talked about on that level more often. If you clip and saturate your sources, mixing them together to make a fat wall of sound is pretty easy.
@@ethangregorymusic I think the point I was trying to make is that without complete control, as the producer, a mixing or mastering engineer probably views loudness a bit differently than the people that focus entirely on making loud music from the get go. Again, it's a different conversation but these videos almost never touch on it because the people making the video are focused on mixing other people's music.
Masterfully explained, I did not understand this topic until now (as an amateur beatmaker, not a sound engineer). Thanks 🙏
Thank you so much for this discussion. I'm a noob, just finished my first master for a client that I recorded and mixed.
But wondering about what Lufs to deliver. I had it at -14 on the money, and -0.3 peak, but not sure if that's a true peak. I checked with Loudness Penalty to get things perfect for streaming. But worried its not loud enough, as a problem I had as a noob artist, when I ordered my mastering and didn't understand compression except that bad thing UA-cam does, I asked for "No compression, preserve the dynamics." It sounded good until it was on Spotify in a playlist and I realized I could not promote it because it sounded quiet and weak compared to everything else in the playlist. So for my clients, I want to avoid having them go through that experience. After watching this video, I think I'll go for -11 Lufs which sounded good in my DAW.
Man you're my hero! Haha this video is awesome! And for real! Nobody is going to turn off the loudness normalization 😂😂 I cracked up so hard!
I usually seem to end up somewhere between -9 and -11 LUFS, and call it good. However, I've to date left the TP at -1 dB. In your example you didn't actually change to -2 dB when it was boosted for dynamic range reasons. Is this something you actually do for "optimized for streaming" masters? When does that actually make a big difference... I assume -13 LUFS doesn't suddenly make -2 dB TP necessary, but at some point it likely does, right?
Hi Wytse! The theory you talking about is absolute correct. But if I stream from Apple Music Adele - "Can I get it" my T.C. Electronic analyser displays a program Loudness is -9 LUFS. Elton John & Dua Lipa-Cold Heart (PNAU Remix) is at 8.7 LUFS. Apple Music does not reduce anything. Even my own production Phonofinish - Dance with me has -10 LUFS (which is the original loudness) at Spotify. I can not see any Loudness reduction. Cheers André
Yeah I've measured some Spotify songs directly from the output at -7
Hmm, strange I got these numbers from my lufs meter while playing the songs on Apple Music 🤔
Adele - Can I Get It -15.9 Lufs,True Peak -6.12
Cold Heart PNAU Remix -16 lufs, true peak -7.14
Phonofinish - Dance with me - 16 lufs true peak -5.25 (get a weird spike just in the very beginning -0.10 true peak but after resetting true peak is around -5 -6)
@02:17 Ah ! So this is why I like to listen to my mix "backwards" sometimes by turning my chair around ! 😄
this is a great analysis of the nuances of mastering. thanks Wytse
The comment about high dynamic range sounding weird is a bit misleading, because it's highly dependent on the musical content. In classical and acoustic music it's quite common to have large dynamic range. I think the whole discussion should be focused around the music rather than numbers. When people talk about "loud" they are thinking of a full sound that blasts you continuously (like in metal or heavy genre), high LUFs is a result of that musical idea, and not the goal in itself.
I mix metal and I hate the way it is done today. I usually shoot for -12 and 10db or more of DR. Sounds so much better.
@@analogkid4557 Yeah I don't like the low dynamic range metal style usually, although I think it can make sense in some cases, e.g. for something like Electric Wizard's Funeralopolis (but it would probably be better with a little bit more DR, less irritating).
Very insightful, loud commercials are sadly still a thing sometimes, especially the ads on youtube. On TV it has improved but mid-roll ads on youtube still have that extreme loudness in some cases.
I think this is because adverts - being short - just don't have quiet bits or silence like longer programmes, interviews, or podcasts. Adverts don't have the wide dynamic range of movies, for example. They might have the same peak volume, but the short term average is typically higher, or at least feels that way because there are no quiet bits in adverts; it's just totally in your face for 30 seconds, like someone is shouting at you.
Hi Wytse! A little off-topic (but not so off).: Are you going to review the new Slate Digital FG-X2? I have the fisrt version, and version 2 it is free for former users. I think would be cool for a video (and even better if you could run version 1 to compare to version 2).
Thanks!
Super informative video!! Happy you're educating everyone, keep it up Wytse! :)
Do you target a similar LUFS level for YT as well? I see that this video has been normalised down by 1.1db - I'm wondering if being over the -14 LUFS matters as much for spoken word / gaming YT content.
Thanks for the video, very interesting :)
As a professional post-production company, we deliver our AV masters in -16 LUFS.
This is exactly what I needed to hear, thanks.
I always master to -10LUFS and they still sounds great on the streaming platforms. I hear no difference between my masters and what shows up on Apple Music or Spotify. The integrity remains so I’m sticking with it. An engineer could go crazy trying to master for each form of media.
This might be your best video you have done! Thank you!
Supergood explanation. Good rant as well :) You make sense in mastering -14 dBFS. And I like how you make us care about the dynamic range.
Thanks a lot for this video! I have always mastered my tracks in 2 versions, one at ~12 LUFS and one just below 14 LUFS because of the Spotify thing. Now I understand things better! Thanks a lot mate! No more 14 LUFS versions!
Are the major labels delivering different masters for streaming than for their manufactured CDs?
Because CDs are at True Peak of -0.3 correct?
The key phrase here. The KEY phrase "Streaming Optimised Masters" Not -6LUFS with -0.1, Glad you are on board here. Im done bashing my head into the wall. One thing that's not clear here, You mention being at -11LUFS but that is not an issue, That totally right. As the guide is for the average measurement over the course of a track. Not just a short section, So we could go to a break down section and be say -16LUFS for a period of time. This bring the average of the track down with it. You may have a moment of loudness at -5LUFS but still have the track average at -16LUFS (Just as an example)
As a (casual) listener, I had loudness normalization turned off, because I did not really have faith that it was being implemented in a 'correct' way. Indeed I thought: "Surely the mastering engineer would know better than some corporate type in a tech company who wants to capitalize on the fear of the loudness war among their subscribers". Now that I know that there is a standard and that it is transparent I will turn it on. Thank you.
I will be happy with I clean the Audio and not that the Audio burns only to be loud.. I think that such Audio can last and as an artist I will make it loud when I want and I will also make such a mix but i cam do Master my self but i want Artist idea to be free not captured in Forum , to make my music stand in line with other similar ones that burn due to the neglect of -1 -2db true pick and crumpled dynamics. This video is finally something concrete that I can attach every time I send music for mastering.. great video and i was stunned with Master Engineers who have expensive equipment and do not know the basics. Thanks
I cannot begin to tell you how helpful this video is💥🧙♂️ big thanks
Thank's a lot.
This was a pretty important information for me, in terms of creating my own style and opinon about this topic.
All the best :-)
VERY interesting video!
Does targeting accurate intended LUFS require playing the track for the full track duration? That was my understanding with Ozone 9, then Ozone 10 took away the ability to toggle on/off LEARN Threshold, and Ozone 10 LEARN now auto-shuts-off after only 5 seconds!
I was thinking about that a lot the last time. Thank you for the inspiration! Last year I started mastering everything to -14LUFS -2db TP. And in a way I figured out for my self, that's most balanced, between dynamics and loudness.
Maybe because 12dB is a double of volume or so. That is the maybe best depth for speaker use.
Now, the Dancehall Artists I worked together with, gave me reference tunes mastered on ..!!!! -0.1 dB TP and -8LUFS or -9LUFS.
So, they haven't been very happy with the -14 -2 mastering...
In the last weeks I updated my studio-setup and started mastering all the tunes again to -10LUFS -0.2dB TP.
I was really not in the mood to push it further cause everything louder starts to distort in a way that destroys all the
real beauty of the production. Might be useful with big amplifiers but for home use its worth to discuss...
Thank you also for this loudness video with all the penalty thing last year, it helped a lot!
Blessings
Very good work and good video. I understand that streaming services need the -14 LUFS. But what I dont understand is the following: When Iam buying music on amazon for instance for dj-ing , then the songs are mastered at -5 LUFS . An they even clipping at 3 TP. And that are artists like Matrin Garrix and Jay Hardway. So, when I want to have a version for DJs to buy and use, then I have to do it the same way? Otherwise the DJs need to trim my tracks louder everytime? I dont get it. Please explain it to me. Thank you 🙂
Thanks for your patience and continued explanations on this topic, Wytse. It apparently is one that continues to vex people without formal training and expertise, and your expert understanding is very helpful. Greetings from Southern California. 🧠
I'm a professor and researcher in neurosciences, and this reminds me very much of several fundamental principles I teach graduate and medical students about human brain function where many seem to not be quite reaching that level of full understanding. Don't get me started on the topic of resting state functional "connectivity." It isn't connectivity!!!! 🤦🏻♂️😉
Excellent video, it helped me understand even more about this dark science... thank you. 😎
thank you for making awesome videos!
I was looking for someone to explain bit depth when mastering.. I have seen all your videos regarding sample rate, but I would like you to explain bit depth please!
I always work with 48k sample rate and export my mixes at 48k 24bit (for the extra headroom)
and for mastering most of the time I render at 44.1kHz 16 bit because that's just what used to be the standard for CDs... but with streaming services being the mayority of where listeneres will consume our music.. what would be the advantages or disadvantages of mastering at a higher bit depth?
I have a question, not releated to this sublject but another that you may be able to throw some light on. Do you think a music studio should be kept at a certain temperature? I have a purpose buily insulated studio in my garden. However in winter when i first go in it can feel very cold and the equipment to touch feels cold. Do you think it would be a good idea to keep an ambient temperture all the time. What would the benefits be over it being cold etc etc. Thanks for an awsome channel. :-)
You mentioned that the master should always be above -14 lufs in order to be normalized to -14 lufs. However, given that the peak is set to -1 dbfs I would imagine you need to master above -13 lufs. Mastering to -14 lufs would result in the track being adjusted to -15 if I'm not mistaken.
-14LUFS with the peaks at -1dBTP.
@@Whiteseastudio Right but if you go above -14 LUFS Spotify says to set your true peak to -2. We should do that?
@@Whiteseastudio Ah I see, understood!
100% agree. Just saying, what I've once heard some engineer saying about listeners turning off the normalization: "the kids don't get their earphones as loud as they want to listen" Which I think is plausible in some cases. Anyway that might just be a reasons why some people might turn it off. Your point is still correct and everyone should calm down with pumping up. But honestly I feel like we are on a good path. Take care tot ziens
Thanks for this valuable information! Quick question - When looking at say, SoundCloud, looking at the meter of the track in SoundCloud, some tracks are way under or all over the place and some tracks have it hammered all the way to the top. Is this SoundCloud applying a limiter to their tracks? Some of these sound good, while others have some weird dynamics going on. The tracks that are way down, don't sound as loud.
Wearing headphones and listening to a playlist - a song plays at -24 LUFS - annoyed, you turn it up - the next song plays at -2 LUFS - instant hearing loss.
UA-cam ads are always way louder than the creators videos.
Because ads aim to stick out and grab more attention. Listening to the sound the ads have for hours is something nobody would enjoy. probably. Regular content creators usually prioritize more dynamics for a natural sound as compared to the radio style commersial voice overs
Well that's because most creators rarely know much about audio. Certainly those talking into a mic, UA-cam doesn't apply lufs targets to typical videos unless they are official song releases (the ones labelled "topic"
one of the best explanations on youtube for „turning off normalization“
9:44 ok thanks for the video but what about the - 2 dB TP level, if the master is louder than -14 LUFS, because of the possible „extra distortion in the transcoding process“ (quote end)?
Thanks for clarifying!
Im curious how this adapt to Dolby Atmos Format (Apple Music)
Atmos loudness is measured using the format in 5.1 fold down loudness and cannot be delivered louder than -18lufs. There's no loudness target for the binaural headphone playback on Atmos devices like Android or Apple, but the headphone loudness will generally fold down and increase another 2-3db depending on where the loudness is in the Atmos space. There's no single speaker format for listening so loudness largely depends on how you're listening to it and on how many speakers etc as it folds down from the master..
Thanks for adding to my slowly congealing knowledge in this area Wytse 🙏
I deliver my mixes at -10 LUFS. I know some people that go even louder. Like Michael Brauer who mixes are as loud as -8 LUFS. Just goes to show that it really doesnt matter if it sounds good.
If the listener’s delivery codec distorts or clips on overshoot above the TP spec per platform/codec, then that distortion or clipping is baked in to their output. The only way I can control that is by staying within published LUFS and TP range.
If it's for the sake of showing something or proving a point without caring about the quality of the audio, it's prolly a bad decision
Thanks again the most important fact in mastering!
Fully agreed. I check my master with loudness penalty app. Give the client a -14 and around -6 version.
I'm wondering about this : why / how is bringing down our master versus providing the streaming plateform with this famous optimized file, different/ better ? Also: im aggregator for a digital distributor (believe digital) that puts out music that I provide on all streaming plateforms. They work only with one audio file. So what's the point of these optimized for streaming services masters ?
I love your videos and you absolutely nail the explanation regarding normalization. However I would point out one thing that I've come to learn regarding mastering minimalistic classical music and especially solo piano tracks. It is not always necessary to aim at -14 LUFS (integrated). Less will actually still do bc going 'all the way' up to -14 LUFS in most cases will require some sort of limiting which is not always ideal dealing with very delicate acoustic recordings. It is better to aim at preserving a high dynamic range and avoid limiting or do heavy compression. With such tracks it is a good idea to mainly look at the 'short term' loudness and secondly try to go as close as possible to -14 LUFS using a non destructive approach (which would include no limiting). What I want to point out, is that you're not necessarily 'in a dangerous zone' being below -14 LUFS when it is very intimate and dynamic acoustic track. You can easily be a little below hitting a loudness penalty of +0.1db (or a more) and still have it working wonderful on Spotify - Lastly, thanks for making these videos and your honest opinions - I find your insights very useful.
-7 lufs seems to be where all the main music streaming services are settling their delivery to your device with normlisation enabled (where available). I measured the same track played back and none of them observe -14 loudness normlsation, they're all just playing the track at the same level it was mastered, which for this track was -7.2 LUFS other than UA-cam. Tidal appears to be around the -14 ballpark though.
Spotify Stereo: -7.0 LUFS
Quobuz Stereo: -7.0 LUFS
YT Music Stereo: -7.3 LUFS
YT Music Video: -7.3 LUFS
Tidal Stereo: -13.0 LUFS
Tidal Video: -7.9 LUFS
Tidal Atmos: -12.6 LUFS
Amazon Music Stereo: -7.0 LUFS
Amazon Music Atmos: -13.2 LUFS
Apple Music Stereo: -7.1 LUFS
Apple Music Atmos: -12.1 LUFS
UA-cam Stereo: -14.1 LUFS
UA-cam Video: -14.1 LUFS
AAC File on phone: -7.2 LUFS
You're checking integrated right? not short momentary LUFS.
@@gregrodrigueziii8075 Yep. running the whole track through YLM threshold to -14 integrated. Loudness normalisation seems to be ad-hoc how it gets applied when not inside an album. Tidal and UA-cam seem to be normalising playlist too to -14, haven't seen any track anywhere neat -14 integrated through Apple Music, Spotify or Amazon though.
@@SamHocking Thats interesting. Specialy now that they have change for their usual replaygain. Its odd that before, they do appear to be in -14LUFS when I checked years ago, but thats them using replaygain and not actual LUFS algo. I wonder whats happening.
@@gregrodrigueziii8075 Could be that when not played as part of a wider album or playlist it was a bit random where they set the level? Left me none the wiser what's going on really.
@@SamHocking i dont know why they are making it complicated. Honestly for us in post prod since its been the standard for audipost for a long time, we just make sure our room is okay, speakers properly calibrated, and for some reason, if we mix with our ears, we are always near -23 or -24 LUFS int. target. Its a bit a differen topic, but, it really isnt that complicated, but for some reason measuring the streaming platforms will get anyone confused as to why there isnt seem to be a standard existing when there should be one existing.
It is enlightening to listen to very compressed masters back to back with more dynamic ones with the normalizers on, ie "sound check" at Apple....it really shows how poor squashed mixes sound....why not use all that glorious headroom for a snare to bounce etc....space is good! I like -16 LUFS even better. Music is so fun to crank up when there is a high dynamic range...thanks for the vid.
Spot on bruh....keeping rising levels to what actually sounds good.
Thanks for sharing your knowledge and the great work you do. I am a bit old school, and I think there's a particular situation where I would take your opinion with a pinch of salt. I'm talking about mastering electronic music intended to be plaid by a human person in a club, techno, for example.
In that situation, other tracks shouldn't play louder than yours. The biggest reason is that the DJ will have to increase the loudness; even worse, they will use the mixer gain, not the volume knob. I would like to know your opinion, but in the club scene, let's say, louder tracks are usually perceived as better or even easier to play. I'm not sure if there is "loudness normalization" yet if you are a DJ. Let's say that the normal is as loud as possible if you can keep a bit of dynamic range.
Thank you very much!!
Appreciate this video very much. Cheers.
I appreciate you talking about this further, and seems you've read my comment, since you posted the same link about spotify (even if you don't quote me on your comments replies) ANYWAY... We agree Spotify normalize tracks for a better user experience, for the reasons you told at the beginning of the video. And this is GOOD. But this doesn't mean you must master your music with this level in mind. As i wrote in the other video, as engineer you MUST work for the best of the music and so for the listener. Certain songs, are good at -16 -14, while others sounds better at higher lufs, not because they are louder, but because the dynamic processing you apply change the sound pressure and the crest factor, and this change how the song is perceived.
I have nothing wrong with services normalization, i'm against to master music because of EXTERNAL decisions, that has NOTHING TO DO WITH THE SONG YOU ARE WORKING WITH.
Talking about the mid part of your vid (it's not -14) i'm getting confused. I mean, you made a video about that stuff, also a shirt, along with different quotes in other videos about that targets, and now... you do not follow your own suggestions... good for your clients, don't get me wrong, but you remind me politicians that say everything and the opposite... maybe you changed your mind? Why just don't make a video with that clear statement then? Would sound way more genuine, i mean, no one is perfect, everyone do mistakes, me, you and i think even god (that's why we are on earth probably).
If your master is good and valuable you don't need another master for streaming, cause your master will translate well also over decoding. I'm doing that from a lot of time, and i never had any single issue, not even one, so why bother with it? I just work for the song, the artist and the listeners, when you know how to tackle everything there is no need for a separate version.
@16:20 this is not called copying, but referencing
@17:00 you forgot to read what i pointed out, on the same spotify link you showed. "The web player and 3rd-party devices (e.g. speakers and TVs) don’t use loudness normalization."
Sorry but the conclusion is laughable, i don't wanna sound or be harsh, but you are saying that quieter music is a better production, this is wrong, again. You don't have to reach any number!!! Both loud or "quiet" you must LISTEN your music, and master it accordingly with how it sound the best, that's it, that's very simple!!!! But you seems to strugle with that idea, you keep to hold that "standard" as a god. First of all THIS IS NOT A STANDARD FOR MASTERING, you keep misunderstanding that, it's a standard for music normalization on a streaming service, this is completly another story.
If an engineer is not able to use their own ears, but to follow numbers instead, to me he is not an engineer, i'm sorry to say that, but mastering a song, is not to open a GUI and try to bring your music into a green area, or reaching a number, this is called GAMING.
Mastering is about LISTENING
DISCLOSURE: Since we are in the f***** internet i want to point out that i have nothing against you, my comment is not toward into any kind of "war" or something, and i'm not an hater. It's just a pure and honest way to reply to your video, i respect you and your thoughts, i just have a different one
I too am a producer of the hip hop genre for 24 years now. I completely and fully understand what you're saying. I guess the answer to your question that you pose regarding telling people to turn off loudness normalization is this. And truly, this is the only reason I can think of. I just think that people turn off normalization so they can squeeze out every last drop of volume when most of us are sitting in a car more typically equipped with a stock stereo system which probably isn't very loud to begin with. I know my Subaru has a very poor stuck system and even maxing the volume, at least just until it starts to sound distorted still leaves me wanting more volume. Therefore, turning the normalization off would afford me a couple or few DB of a gain boost. With that said although, I completely agree that it doesn't make the track sound better at all. Just louder.