Thank you. I guess those are for OS X only. Any alternatives for Windows?

 

Yeah, I understand that it's still the part of the standard. But they do have these options also:

Apple Music Lossless (24-bit/48kHz)
Hi-Res Lossless (up to 24-bit/192kHz)

I understand that 99.9% of the people will not hear any difference. And I'm wondering if professionals can realy discern.
And what's the point of those? Marketing?
Knowing Apple, they may consider only 24 bit/48khz "True Loseless" in a few years. Sticking special badges or something. Like they're pushing Dolby Atmos now

 

That's clever
But on a serious note. Do you care about True Peak Limiting? Do you submit your tracks in 16 bit / 44.1 kHz or higher?

 

I get your point. But people won't go back to CDs and they use Spotify for convenience. Every corporation is funding wars in some way. Even when you put the fuel in your car or buy groceries. The only way to avoid it is to leave this planet.
Do you care about ture peaks, bit depth and sample rates in your music?

 

Yeah, that's what I thought. Nobody is following their guidelines. But I wanted to hear it explicitly from people who actually tested it on their own stuff

Seems like you did those tests on Spotify/Instagram and came up with the standard for yourself. Don't you mind sharing it?

 

as stated, I checked on all major streaming services

-true peak doesnt help translation/decodification if the track peaks at 0.0 dB

-at 0.0 dB there's nothing in this world that will save you from digital distortion

-I checked the maximum peak possible, yes: -1.0 dB translates/codifies perfectly on all platforms in 99% of tracks, and when I have very high LUFS like -4 LUFS I peak at -1.1 dB for more safety

-true peak also kills transients in my opinion, I tested it with FabFilter limiter, and I was very surprised how bad it sound compared to Newfangled ELEVATE. My final Limiter is ELEVATE at -1 / -1.1 dB with NO TRUE PEAK

I dont mind loosing even a couple of LUFS, for what I release -9 LUFS give the best sound and equilibrium, someone could like higher LUFS, thats fair, but who works with plugins ITB should export this way:

-no true peak
-maximum -1.0 dB

PS: I didnt discuss sample/bit rate, thats a VERY DIFFERENT SPORT ehehehe and its not connected with LUFS, loudness etc

PPS: all I have written is referred to ITB WORLD, so plugins, digital mastering and so on and so forth. People that work with hardware limiters and analog converters has a very different standard (but there arent many actually...)

 

Understood. Thanks for your recipe
But what sample/bit rate do you prefer yourself? If it's not a secret of course

Did you try oversampling in Fabfilter Pro-L 2? It goes up to 32x
And AOM Invisible Limiter G2 has like 512x (!!!) oversampling

 

After converting to mp3, there are not only intersample (true) peaks appear, but usual sample peaks. Not the case of true peak limiting only, it only gives some "natural" virtually (oversampled, reconstructed) "analog" peaks to be limited. After conversion all the peaks become changed.
True peak limiting kills a bit more dynamics and punch, but you achieve less aliasing, distortion, less digital. I don't know if is it a big problem solver for peaks and true peaks for mp3, but bit less distortion, same as dithering for mp3 (works in some cases, in others not). More ceiling is one of the answers for pcm peaks and true peaks

 

Yes, I've provided usual sample peaks for MP3s in my original post. So I had noticed those as well

Somebody consdidered my comment about oversampling in limiters funny, but as far as I understand, true peak limiting is actually achieved via oversampling. And we get less aliasing as you say

What settings/ceiling do you use most of the time?

 

-1dB output for peaks. True peaks limiting, oversampling yes, but only after any SRC.
Dithering with noise shaping (for mp3 there are lots of questions about dithering).

 

these days there are no secret weapons, infos goes so fast and deeply that it is impossible to hide something (unless you are Serban Ghenea and really never show-say a thing lol)

as stated, sample/bit rate has NO FIXED recipe, if you ever meet someone saying that he has "the fixed secret weapon for sample/bit rate" he doesnt know what he's talking about, and the same for OVERSAMPLING

Oversampling changes the sound of the production. Period.

There are no doubts on that, even for beginners of audio production. I will make an example.
If you tweak your plugins, mix, master then activate a very high oversampling only when exporting (for example with Voxengo Clipper, which is capable of the highest oversample on the market if I am not wrong, or with Acoustica Audio HARDER oversample/settings) the file resulting from that export will sound different than what you have been hearing during the tweakings, and the chances that "it sounds better" are not that high. I written something that implies long discussions, but if you think at my sentences you now know that "there are no right and wrong recipes". The best thing is to find equilibrium in your production stages, considering everything, from your monitor, room, computer power to pragmatism of production (deadlines and things like that)

Back to sample/bit rate: again, it DEPENDS.
ok, for bit rate lets say "the higher the better according to your DAW and computer power"
Lets say easily that way.
Now about sample rate.
if you are working at a production 100% ITB, with only samples and nothing recorded (vocals, guitars or something) then there's no reason to work at a sample rate higher than what you know will be the final export. So 44-1 for pop music, 48 for cinema/video. And this is something which is probably stated than 80% of professionals.

My 2 cents:
my favourite sample rates are 44-1 and 88-2
Everyone know that famous Jack Joseph Puig interview, so I wont quote. I can just add that I started working during the CD era, so the final product needed to be 44/16, and I also released the same tracks on vinyls printed from that identical master. One day asked the label if it was possible to put the 88/24 on vinyls. It was many years ago, but even my grand mother heard the difference.

These days I work at 88-2 only if I know I will use hardware gear like my analog synthesizers or hardware reverbs, or if I know I will record something

If I am sure that a production will be 100% ITB I start at 44-1 and finish it at 44-1

 

That was comprehensive. Thanks

Regarding oversampling. Can you explain what you mean by "sounding different"? I believe Ozone Maximizer has like 4x internal oversampling. And I used it on buses multiple times during production + now on master (and it doesn't sound bad to me at all). Maybe you're talking about slight distortion, but in heavy electronic music it's not always a bad thing, so I might not have noticed. Like if you clip 3 db on master - of course no oversampling will help you, there will be drastic audible diffrence. But if you put something like StandardClip with -0.1 soft clipping setting just to shave the peaks, do you really think it's gonna make it much worse with maximum oversampling?

Regarding bit/sample rates. I did record guitars and vocals in 24bit/48kHz. That's why I am so hesitant to use 16 bit/44.1kHz in the final master file. I would rather test 16bit/44.1kHz temporary master file myself and detect true clipping areas somehow, then make adjustments in the same spots for my original 24bit/48kHz project to make sure it's good in both cases. And feed 24bit/48kHz to distributor. Those platforms who need 44.1 will convert to it without peaks, and those who can take 48 will be able to take the original.
And a video is planned indeed, so that's another point to 48 kHz

I also wouldn't hesitate if I had to choose between 44 and 88 like you. But "16 vs 24" and "44 vs 48" is not that big of a difference. So why not use it if my computer handles it and Apple Loseless can stream it. Theoretically I'm losing nothing but might get something. Even if I myself can't hear it

 

AFAIK, Lukin said Ozone uses 9x oversampling.. Maybe exciter or so, don't know nor remember

 

I have written a lot already : ) I wont specify. I didnt write "sounds better", I wrote "sounds different IF YOU APPLY MAXIMUM OVERSAMPLING in OFFLINE after the mixing process...while you didnt listen to the realtime sound"

Yes. It will sound different, not necessarily better : )
but you will realise this very very simple concept with experience

lol : )
I sometimes use softclipping at 60% with 7 dB reduction in Standard Clip.
but IT DEPENDS, as I wrote, by the kind of music you make and the sound you wanna achive

Also, I sometimes oversample at 128x with Voxengo clipper : )

just to tell who am I: I am not a classical music producer (at least in this period of my life lol)
I am not scared of distorting hardly my stereo buss, you bet ahahah.

I wont write more about this : ) just walk on your way and find your solution, maybe read again and carefully what I wrote above. Not because "I wrote the RIGHT thing and THE TRUTH", but just because your answer is not related directely to my concept

Oh, one last thing and I am obliged to say: maybe 44 vs 48 is not that big difference (lets say easily...even if.....)

But "16 vs 24" THE WORLD CHANGE ehehehe : ) you will understand soon!

If you dont HAVE TO go on CD just avoid 16 bit, like true peak ehehehe

 

Yes, you did. Thank you very much. I appreciate that. Analyzing different opinions certainly gives some food for thought. But I feel like everyone except for lxfsn is giving slightly vague answers as if "sample rate+bit depth" is something sacred. He gave his exact settings: 16 bit / 44.1 kHz + dithering. And explained clearly that it "offers the absolute best compatibility with anything". I wish more people could explain it as simple in just two sentences and say why 24 bit / 48 kHz is overkill or not very compatible

 

So i've been studying mastering for 8-9 years, i'm only at "semi-pro" level career wise but i've been in direct email contact with a pretty high up guy that does top40 level stuff for labels, his peers are top level guys and he knows what the top pro's are doing due to talking with them. We've exchanged novels of information. I've also talked to various pros in mastering forums and listened to a bunch of mastering-specific podcasts over the years. After a while you do figure out who to trust and which information is coming from actual professionals or just coming from academics and salesmen. Here are my findings:

1) None of the top guys care about LUFS normalization. Most don't even look at LUFS meters. They optimize the loudness that sounds best for the song and let the normalization fall where it may. Normalization is just a playback thing and doesn't affect the sound. Normalization standards are likely to change over the years anyway, so that's why nobody bases decisions on it. Most guys ARE still trying to get it as loud as they can though, but there's also a point where a song can't go any louder without sacrificing too much. Every song has a different "breaking point", and it really depends on the arrangement & mixing (but especially the arrangement). For some songs the sweet spot is -5 LUFS, for others it's -10. Depends on genre also.

2) Spotify recommends -1 and -2db peak ceilings but no top guys are doing that. If you deliver a song to a label with -2db ceiling they will say what the fuck are you doing and send it back.

3) Top guys don't care about true peaks. Some use ISP, some don't. Some believe it dulls the sound too much, some use an extra limiter at the end of the chain dedicated solely to ISP but not making it work hard at all (maybe -0.1 or -0.2 of gain reduction or something). At the end of the day once it converts to mp3 you ARE going to get true peaks over 0. The encoding adds gain, it's unavoidable. You'd have to reduce it like 5db to truly avoid this and nobody is doing that.

4) Apple streaming has specific standards but nobody has really followed them for years. The whole "mastering for itunes" badge is a thing of the past. Not relevant anymore. Nobody is creating separate masters that are specific to the streaming platform.

5) A lot of top guys are annoyed by the confusion Ian Shepherd has brought to the industry. They're constantly getting clients who think their songs need to target -14 or they'll get *GASP* penalizzzzeeddddd. Ian has brought fear to the equation. Normalization is just a playback setting, and users can turn it off if they want. Always remember who the advice is coming from, Ian doesn't have any major credits. He has chosen the salesman path, and that's fine but it's not useful to people wanting to work on a commercial level. Nice guy though don't get me wrong.

(On a personal note, I bought Ian's course 3 years ago. That was a waste of $400. These days i wouldn't pay $10 for it. It wasn't terrible, but it was more of an indoctrination than an education. It was more preachy than teachy. I didn't learn anything advanced. I internalized his "be careful" philosophy for a year or 2 and i learned that following his guidelines was a mistake, so i guess i did learn something. It made me scared to push the limits and it made me release my EP at too quiet a level, as well as master songs for clients that would've sounded better if i pushed the loudness more. Live and learn.)

My personal setup: I have my output limiter at -0.2 and i don't use ISP. I don't like the sound, and i analyzed all my favourite music through my metering and every single song has true peaks over 0. I use 2 limiters (or 3 if the genre is a super loud one), one doing the heavy lifting (like 1-2 GR, sometimes less sometimes more depending on how compressed the song is) and then one at the end doing doing a tiny bit (like 0.5 GR max, mostly hovering around 0.2 GR) with oversampling at x32. My theory is that between them there's enough peak control happening to where true peaks aren't a big deal. With ISP off there's more "perceived loudness" imo, cos the transients are clearer. IMHO, any distortions that are introduced in the encoding process are gonna be masked by the loudness of the song anyway. And I'd rather have a song that perceives louder and has some encoding distortions that nobody can hear, rather than sacrifice some loudness perception in order to ensure there are no encoding distortions that nobody can hear anyway.

Personally, i don't look at integrated loudness at all. I do look at short term LUFS and momentary LUFS. For me, if the max peak mLUFS is quieter than like -8 or -9, then i probably need to consider pushing the loudness. If the mLUFS is louder than -4 or -5 then i need to double-check and make sure i haven't sacrificed too much to get it to that point. Most of my masters are hitting around -6 to -7 at their peak mLUFS before i deliver, but it varies. Depends on the song.
Imo the only time integrated LUFS matters is to check if my master is quieter than -14db. AFAIK, if it's below -14 then Spotify will turn it up with their shitty limiter algorithm. Nobody wants that.
(note: spotify might've changed the limiter algorithm or something since i last researched it a few years ago, but my masters are never below -14 iLUFS anyway so i've never bothered to research it again)

 

Last edited: Jul 28, 2023

Couldn't agree more with what you said!
All of those streaming platforms' requirements do not make any sense in the industry, the loudness war is still going on but now with more options to access critical aspects of a mix or a master. You are the only one making sense here.

Although, even oversampling changes the signal, it's a matter of taste and how precise you want your dynamics processors to react to it.

 

I deliver at 44.1k and 16bit, personally. The way i understand it (and i could be wrong, don't take my word for it) is that since Spotify is converting everything to mp3 (which is a 16bit format) then it will have an easier time converting from a 16bit file with 16bit dithering than from 24bit.
24bit does sound better, but to me the question becomes: do i want the song to be converted from 24 to 16 with Spotify's algorithm or do i want to do that conversion myself? I prefer doing it myself.
BUT, my knowledge on what Spotify's doing in the conversion process is a little spotty. Does Spotify use the same conversion algorithm regardless of if it's 24bit or 16bit? I don't know. And i don't think anybody else here really knows, so that's probably why you're getting vague answers.

At the end of the day, my advice to you is don't overthink it. Plenty of songs with all the emotional impact in the world are in 16bit. Nothing bad will happen if you export with TP overs and choose the wrong bit size.
16bit is the safe option.
24bit is the audiophile option that may not translate to streaming platforms.
So base the decision based on what the end platform will be.

 

Haha, i felt the same way when i saw your post on page 1

 

That's an interesting question, you have tons of software available on Windows to check the loudness values of a file.
In my opinion, with this "Apple Digital Masters" certification, they expect you to have an OS X workstation to use their "Master Droplets" plugins and an unix terminal for the "afclip" command.

My only solution (which is not handy at all) would be to virtualize an OSX environment on your machine or create an hackintosh and dual-boot it via Windows (Windows on one partition and OpenCore on the other). Take everything with a pinch of salt though because I've gone the lazy way and bought a refurbished Macbook to be on the safe side.

Edit: there is a software from PA ADPTR called "Streamliner" but it does not give the same insight as AUroundtrip nor does it gives you the ability to certify your masters under Apple' management.