Official Mastering and Production thread

[SnappleMan]

Todays mastering standard is -0.1dB to -0.2dB, not 0dB.

These days, if anything hits 0dB it's pretty much considered clipping. But that doesn't stop idiotic rock and metal producers from overdoing it.

[zircon]

Yeah. It's kind of dumb. I usually don't quite go to 0db either, actually, but SLIGHTLY less just in case.

[GrayLightning]

Todays mastering standard is -0.1dB to -0.2dB, not 0dB.

These days, if anything hits 0dB it's pretty much considered clipping. But that doesn't stop idiotic rock and metal producers from overdoing it.

More than just them. I've seen so much professional releases lately hitting 0db! It's becoming an epidemic across the board.

[SirRus]

this is most certainly a n00b question, but what is a reliable program you all use to monitor the peak levels, and see if your songs are hitting 0 dB or not? Can you rest assured that if your song is not going above this level during the entireity of your song that it is free of clipping?

[OverCoat]

most software has dB meters in it, but any waveform editor will show you the FULL song [Audition, SoundForge, Goldwave, Audacity] and whether it goes above 0 dB or not, of course.

[GrayLightning]

this is most certainly a n00b question, but what is a reliable program you all use to monitor the peak levels, and see if your songs are hitting 0 dB or not? Can you rest assured that if your song is not going above this level during the entireity of your song that it is free of clipping?

http://www.elementalaudio.com/products/inspectorxl/index.html

That's one of the best.

I personally just use the one in Cubase with the free version of Inspector - http://www.elementalaudio.com/products/inspector/index.html.

As for your other question - Not always. SGX had this problem a few weeks ago where he was pushing his limiter too hard and the problems of it only manifested itself when he converted it to mp3.

[SnappleMan]

Well, someone should restate that there is good clipping and bad clipping. And I guess it'll be me. So yeah, there's good clipping, and the inverse of which would be bad clipping.

[zircon]

Well... digital clipping is ALWAYS bad. Analog-type saturation however is usually a cool effect.

[bLiNd]

You cant really judge clipping visually based on an mp3 since an mp3 compresses the audio even more. I master at -0.1db to its maximum punch without audible clipping but almost always the mp3 will clip on a VU meter, you cant trust it. And yes, inspector is a fantastic tool for this kind of thing.

[Splunkle]

Thanks for the quick answer guys, I think I get it now. I was using the Fruity compressor, which might have a limiter built in as well, as no matter how hard I push the gain, it never clips. Curious.

Also, I was wondering just what analog style clipping was. I know that guitar amps (well, most guitar amps) work by pushing the signal too hard through a tube amplifier, which gives that lovely distortion we all know and love, because overdriven tubes to cool things. But then I was also told that in the old days of recording, when everyone was using tapes, many engineeers delibrately cranked the volume up so that the tape would clip, because this produced some sort of spiffy effect, but a different effect from the tube amp.

So I was wondering, when people say analog style clipping, are they refering to one of those forms of clipping, or something else alltogether?

[zircon]

Well, analog clipping is simply when you push the volume above 0db on an analog device, like a tape machine. It can produce a variety of effects depending on the type of device, or even the exact model. Therefore, there's no one way to describe what happens.

[Yoozer]

The difference is that purely digital clipping flattens and analog clipping (because of the nature of the medium) saturates and even compresses before it goes all out of whack. It's the difference between harsh and "hey, that's actually a nice-sounding effect".

[Compyfox]

Sorry but I can't leave it this way.

You cant really judge clipping visually based on an mp3 since an mp3 compresses the audio even more. I master at -0.1db to its maximum punch without audible clipping but almost always the mp3 will clip on a VU meter, you cant trust it. And yes, inspector is a fantastic tool for this kind of thing.

This is why you limit it at -0.2dB!

I mastered so many tracks in -0.2dB with different loudness. In the K-System, -8dB RMS, -6dB RMS, hell even -4dB RMS and higher where you definitly hear the pumping. MP3 has a limit - and that limit is -5dB RMS, everything higher and it has digital clipping even though it is limited at -0.2dB. I engineered so many tracks, did 1:1 tests... no loudness change, no limiting change if you don't push the mp3 format to it's limits. So it's not true that mp3 compresses properly mastered material even further in terms of loudness.

Also if you use massive compression, exciting tools and other enhancers, and in worse case your limiter doesn't really "brickwall limit", your 0.1dB headroom is useless.

Let me tell you why:

CDs in the early days (even still) needed a certain headroom. That headroom is this 0.2dB (sometimes even 0.3dB). The reason was simple. The first CD player had problems with highly compressed material and it had a certain limit. Spikes (so called full scale samples) who caused clipping - most of the time certain synth patches/sweeps, bass, hell even hats - also caused the CD to literally barf (hiccup, skip), because the internal engine couldn't take it anymore.

Even though limiters were accourate, you still needed a certain headroom to prevent clipping. So every mastering engineer who knew this (I'm not talking about project studios who spreaded like mushrooms) used -0.2dB as savety for spikes. And all of their highend metering tools didn't show any clipping at all.

Nowadays there're tools out that catch every single issue (like Penguin Audio Meter, Inspector, soon the new limiter from PSP Audio Ware too if we talk about native plugins) - bitwise. Limiters and compressors are a tad more accourate than 10 years before, but that doesn't mean that they're flawless either. Like Gray mentioned... you wouldn't believe how many wannabe mastering engineers push the tracks to their limits, ignore every law of physics and sell their CDs full of bugs.

You see, this is not a wild and crazy rumor, it is a fact. No Limiter - and by that I really mean noone - works flawless! I tried so many in the last couple of years. Every limiter (hell even Waves!) with a brickwall limiting of 0dB clips. No matter which material I used (except material that was at K-20 and lower of course, which barely touches 0dB peak).

A lot has to do with the attack and release time of the limiter, too. But most of them only have a treshold and gain level. And this is where the -0.2dB (or even -0.3dB) come into play. Chances are that your limiter, as good as it might me, still let spikes through that're not registered by the peak meter (but tools like Inspector) who're still louder than 0.1dB. With the 0.3dB headroom however, you prevent every possible clipping that might occour.

You loose this race however if you:

- make the track louder again after it was mastered (for example CDex internal "normalise" function)

- your mastering limiter uses an EQ after the limiting

- encode into mp3 and the RMS level is stronger than -5dB RMS

- encode into AC3 and your track is louder than K-12 (-12dB RMS, K-System meter setup)

So... as you can see. Loud is definitly not better!

And I guess it is time to post a link to the loudness race homepage: www.loudnessrace.com

(server is unfortunately down a lot, and half of the facts are nonsense, but he has a point)

I could write more about the other stuff mentioned, cause I definitly do not agree with everything. However they're just another aspects of what you could "do" - a different opinion. I just wanted to "fix" this little issue.

But do not forget: Mastering is an own world for itself! There's way more behind it than just "highend" plugins (or even DSP cards) in a certain chain or all-in-one solutions. Else audio engineers would die out.

[sephfire]

Glad to see you in here, Compy.

Keep it coming guys!

[zircon]

Another thing to consider - if you are making EQ changes in the mastering phase of more than about 5db then you should really go back and work on the individual parts more. The EQ changes in the mastering phase should be minimal - you're just doing subtle, transparent edits to polish the finished track. All of the major EQ work should be done in the production phase.

This brings me to something that I learned from tefnek, though it's a technique that many major producers use as well. When working with sampled drums, it is a wise idea to put the individual drum parts on their own channels; eg. kick on one track, snare on another, hats/rides on another, toms on another. But after you do the necessary processing and mixing, what you can do is then route all of them into a 5th channel (in FL, you can do this with sends). You can then apply some saturation, perhaps a vinyl effect, a little EQ, and some compression. This can make the drums sound like they were recorded from the same kit, even if your drum track is composed of 30 samples from all over the 'net. Most notably, The Crystal Method uses this technique.

Edit: Some more stuff that I do..

* Always make sure that when you're changing one particular part of a song, you listen to it by itself AND in context of the mix. For example, let's you're doing a basic rock song and you have drums, bass, guitar, and voice. If the bass sounds "off", what you should do is first solo the bass part and treat that by itself. Once you think it sounds good, then play it back with the rest of the mix and see if you were right. This is a sound method of working because you don't want things to be tweaked and sculpted to hell, making them sound lifeless or unrealistic (even though they might sound OK together in the mix), nor do you want them to all be mixed as if they were playing solo, meaning they'll muddy eachother up when played together.

* If you're working with samples and you're spending an inordinant amount of time on one specific sample, trying to make it sound good, you might consider simply trying a different sample. Some people that I have talked to believe that EQ and processing can fix anything. I disagree. Sometimes, making a crappy kick drum sample sound good really is impossible, and you're better off finding a kick sample that sounds closer to what you want, and then tweaking that slightly. You'll save yourself a lot of time working this way. Of course, layering samples is another way to mask weaknesses and build on strengths.. not only for drum samples, but for other stuff too (for example, mixing different string sounds to get a more realistic and expressive blend).

* Vary up the frequency range that the listener is subjected to over the course of the song. No matter what genre you're working in, prolonged exposure to a certain frequency or sound WILL wear down the listener's ears. Try this for yourself. Find a sample of a 10khz sine wave and play it back at a normal listening volume. Annoying? Yep. While you're not actually working with pure tones in a mix, the concept is still the same.

Some common pitfalls include making hihats or other top-end percussion sound really bright and clear, and playing them throughout the entire song. Alternatively, having a low-end synth playing constantly at the same frequency range. Even something as simple as acoustic guitar strums can eventually become grating. Using "bridge" sections in your arrangement is one way to avoid this problem, but also, simply being more aware of what parts are playing for how long can help a lot too.

[bLiNd]

Well compy Im not going to argue with your "expertise" but in the world of club music, louder is better. My white skies club mix is a perfect master for a huge PA. Any mastered wave (properly mastered) wont clip visually or audibly, but any kind of after compression for videos or mp3s or whatever, can cause audible clipping as well. It depends on what your mastering for. If you are mastering for vinyl that will be played on a PA then you can hit the -.1db mark. Creamwares optimaster plugin doesn even have the option of an overhead of -.2db unless you do it manually with a fader. And im sure the German engineers at creamware know a thing or two about their own mastering plugin. Not one plugin ive heard in a software environment comes close to the transparency of the optimaster....

[bLiNd]

So it's not true that mp3 compresses properly mastered material even further in terms of loudness.

I didnt mean in terms of loudness. I meant that the compression of the actual audio to mp3 can cause slight peaks and jumps causing the "Red light" on a vu meter to turn "on". You can usually see graphically the limiting on a wav but if you open the same wav as an mp3 you can see that the limited wav has been compressed. Usually an ultra high bitrate on the mp3 will clip less (And when i mean clip i mean 0db) then a normal bitrate. And let me clarify again, that i mean VISUALLY seeing the clipping on a VU, not actual audible distortion.

[Compyfox]

I recently mastered Shael Riley's "Toybox" album. That album is in -8dB RMS, -0.2dB limiting. No matter if I open the WAV files or the mp3 files in Wavelab for a 1:1 check - both files are the same, none of them clip, the mp3 doesn't get "compressed" or normalised. A demo track that I mastered before for SGX with his mp3 issue was in -6dB RMS, also -0.2dB limiting. Even here, no clipping in the WAV, and not in the mp3 either.

I wouldn't have written this if I'm not sure about it, blind.

Also... You should know better.

Vinyls work on a carving base. The smaller the amplitude (loudness) for carving, the better the record, the less the chance that the needle hovers around but stays on track. I never really gave tracks into press, but I know from befriended engineers that a -8dB and higher RMS is definitly a NO-NO. Some even say that -10dB is way too hot.

The carved amplitude simply would get too big and turn into a squarewave, which also affects the other windings. This for example is also a reason why not to put a flanger or the bass/basedrum for club productions, or "spread" it out. Not only does it sound crappy on 08/15 standard club systems, neither does it work on the vinyl record itself. CDs work on a totally different basis, this is why loudness-driven engineers could easily step over it's limits.

Speaking of Creamware.

Maybe Creamware knows what they're doing, but that you can't even set the limiter to something different than only 0dB is a joke. No wonder that so many CDs are simply mastered the wrong way. Like I wrote in my last post... there are "always" spikes. Not hearable, but the metering tool catches it. And -0.1dB is just too less of a headroom.

Also if I hear the name "optimaster", I get the chills. Every mastering limiter/compressor that I know which uses that name or even part of it, squashed the life out of the music. Especially in radiostations. So it's like a red flag for me which tells me "don't ever touch it!". Not to mention, if you say that this is the only transparent thing you know, then you definitly didn't try the TbT plugins. DSP cards good and fine, but to me they're overrated.

No pun intended, but this is what I learned over the years from several pros (one of them Bob Katz, inventor and defender of the K-System) or read in books/AES papers.

[bLiNd]

All optimaster is, is a finalizer like ultramaximizer by waves. The sound quality of their compression algorithm is insanely good and it really puts waves to shame. You would just have to try it to see what im saying. The limiter can be set to whatever you want. But on the pre-normalizer there is a headroom knob that only goes down to -.1db. Anyway thats irrelevant. When i say that optimasters compression is TRANSPARENT, i mean that there is NO squashing whatsoever. NO PUMPING at all...

TbT?!! FREE plugins, now thats a big warning sign to me for a quality compression. Now if you choose those over creamwares plugins, you are just ignorant.

I have a ton of trance vinyls that i have ripped myself mastered at -.2db to -1.db. You dont need to drag this conversation on anymore. Listen to my white skies club mix and you will hear the quality of the compression ive been talking about. Show me some examples of your own work (your own music) with your plugin choices and maybe ill respect your opinion more.

[Compyfox]

Okay, now this was a clear threat towards my side but okay.

I won't lecture you, I just want you to please accept the facts. Nothing more. If you think your masterings are okay in your book, then they are. If you think "free plugins" aren't worth shit, then they aren't worth shit to you. If you say "-10dB RMS was in the 50ies", then so be it.

You know my works on JoeCam's "Hits & Misses" (which was mastered with a "free" limiter, btw), you heard stuff from Haroon Piracha that I mastered, Trenthian, Zeratul ("Just a Friend", just released on his page), soon even Shael Riley's "Toybox". If you say I talk crap - fine with me. It's your opinion as I have mine in terms of mastering. You say "make it loud", I say "K-12 is more than enough" - whatever, doesn't matter.

It's more ignorant to not consider or even accept other opinions/skills rather than standing ground. I mean... I didn't devote my sparetime in the last 3 years for nothing. And if one of the most well known audio engineers (Bob Katz again) agrees with with what I just wrote, is an AES member, over 15 years in the business and even told me a thing or two, and you say this is not right... then I don't know what is.

No threat, just my 2c.

[bLiNd]

well we just have different goals and different tools. I respect your knowledge about the technical sides of audio, they surpass me greatly. I know how to use my tools to get the sound I want, and same with you. But dont get me wrong, i do respect the time youve put into your know how and expertise. Anyway, I didnt really mean to attack you.

[suzumebachi]

what the hell is TbT? also, i don't quite understand how root mean square works when applied to decibels either... do you basically just mean the overall mean volume of the track?

[zircon]

Right. Peak volume represents.. the peak volume, eg. the loudest the track goes. RMS = Root Mean Squared, which is a representation of the average loudness. It is always lower than the peak.

[Argle]

Well... digital clipping is ALWAYS bad.

Nah, I wouldn't say always. Some of the more extreme electronic genres like industrial are built around really harsh sounds. digital clipping or distortion seems appropriate if you want something nice and abrasive.

[zircon]

No, it's always bad. As in, if you have clipping, it won't even play back properly on some devices.

There's a difference between that and distortion effects which simulate the effect, or recording the effect of the clipping and then compressing it to use in your song. But there is never any case where your actual waveform on the CD should exceed 0.0db.