Where do you "master" to? (what dB)

[Compyfox]

So you are saying that not needing to limit at all is a good reason to limit to less than full scale? That doesn't make any sense.

It actually makes sense, if for example K-20 or the new EBU R128 "standard" is used. Then the limiter is only there for short and very high peaks just in case. But due to the nature of this material, chances are you don't reach -1dBFS digital at all.


Not being an uber-expert here, but this is my observation on this topic.


Else, I've made the observation/learned from experience that up to -0,5dBFS digital is sometimes needed due to the DA of CD and MD drives. I had a couple of test CD players and diskmen where the ceiling was actually higher as well (up to -1dBFS digital). Else the signal started to hickup (resultet in pulsing bursts or dropouts).


So it's not program material dependent, you need to take the DA into consideration as well.

Again, personal opinion/experience.

[IIRs]

So it's not program material dependent, you need to take the DA into consideration as well.

You need to consider the DA only in the sense that ISPs which are not a problem in the studio might cause audible distortion on cheaper consumer grade DACs.

I don't think mastering engineers typically check what kind of CD player you use before setting their limiter: the whole idea is that the mastered audio should sound fine on any player.

So I stand by my statement: the only reason to limit to any level lower than 0dBFS is too allow headroom for ISPs. ISP levels will vary enormously depending on the program material, therefore the only way to be sure you have left enough (or not left too much) is to use an ISP meter.

I'm surprised its proving so controversial to be honest!

[trimph1]

So it's not program material dependent, you need to take the DA into consideration as well.

You need to consider the DA only in the sense that ISPs which are not a problem in the studio might cause audible distortion on cheaper consumer grade DACs.

I don't think mastering engineers typically check what kind of CD player you use before setting their limiter: the whole idea is that the mastered audio should sound fine on any player.

So I stand by my statement: the only reason to limit to any level lower than 0dBFS is too allow headroom for ISPs. ISP levels will vary enormously depending on the program material, therefore the only way to be sure you have left enough (or not left too much) is to use an ISP meter.

I'm surprised its proving so controversial to be honest!

I see it as just another way to do things..if it works in your workflow then fine.

[dark_virus]

-0.3 always. Limiter celling set to -0.3db.

[Compyfox]

I'm surprised its proving so controversial to be honest!

Not at all, Dan. We all come from different areas in that section and I say limiting the ceiling IS important. MP3 usually ignores that (even with a limited ceiling prior to rendering, it will be gone after the encoding), but mediums like CD are not as forgiving, At least to my experience.


Then again, I used Inspector Free the time I dived into mastering and then moved on to iXL while it was still Elemental Audio (read - YEARS AGO!). And that statistics mode was one of the most brutal I've ever used/seen. I wish Inspector would still live on, or another firm would incorporate such a statistics view.

Then again, ton of great tools with show "true peak" or "intesample peak". Pro-L being one of it, recently released bx_meter as well, so is the DigiCheck from RME, and there are a couple of more tools like that.


If the (pre)master is friggin loud, then yes - ceiling all the way. If it's residing around the 0LU/EBU R128 standard, or K-20, then only for savety precausions.

One way or another, I usually have it setup at -0,3dB. Depending on the limiter I use, it either sticks to that, or my iXL/meter of choice tells me "dude, you have a serious problem".

Sticked to that for years, and I still kept my picky Phillips Diskman for very important material where I like to check more than twice just to be sure. If that one doesn't barf, then it works everywhere.


Again, just my 2c.

[IIRs]

I see it as just another way to do things..if it works in your workflow then fine.

The only alternatve so far proposed in this thread is to just pull a number out of your arse and hope for the best.

You can go for "-0.3 always" if you want, but you are ignoring the fact that ISP levels might be as much as a decibel higher than that in some circumstances. The only way to know if 0.3dB is enough is... an ISP meter.

[Fritze]

So it's not program material dependent, you need to take the DA into consideration as well.

You need to consider the DA only in the sense that ISPs which are not a problem in the studio might cause audible distortion on cheaper consumer grade DACs.

You just paraphrazed what Compyfox wrote, too. Cool.

I'm surprised its proving so controversial to be honest!

Get some fun, man. You appear a little stressful to me in this thread. That's most of the controversion I see. Peace.

[IIRs]

You appear a little stressful to me in this thread.

I did have a fairly traumatic encounter with a dentist this morning...

But I'm not in the least annoyed about this thread: just want to make sure the OP's question is answered properly!

[IIRs]

You just paraphrazed what Compyfox wrote, too. Cool.

No, I don't think I did. You might take the DAC into account to decide what level of ISPs to permit. But you then need an ISP meter to determine whether your specific material meets those criteria or not.

[Integratron]

Yes I know. The thread is about how much headroom to leave when mastering, ie: intersample peaks.

No it's not. YOU do it that way. I don't. Kim doesn't. Others don't. Some do.
So this thread is not about intersample peaks but about how much do you master to. When the OP said "how much do YOU master to" he didn't mean IIRs but anybody who wants to join the discussion.

Hey - what's 0.3 of a dB amongst friends?

[Integratron]

I see it as just another way to do things..if it works in your workflow then fine.

The only alternatve so far proposed in this thread is to just pull a number out of your arse and hope for the best.

You can go for "-0.3 always" if you want, but you are ignoring the fact that ISP levels might be as much as a decibel higher than that in some circumstances. The only way to know if 0.3dB is enough is... an ISP meter.

[Compyfox]

Dan didn't paraphraze me at all. Kim, Dan and I are right on several topics.

The thing with ISP or "Inter Sample Peak", also known as "true peak", you have two problems:

1) the DAC "interpolates" your signal (especially overcompressed ones) to reconstruct the data sausage back into analog form. Endresult, intersample overs, or better said: clipping

2) your limiter "might" be good and fine, but chances are that very fast peaks or "intersample peaks" result regardless. A lot of limiters back in the day (especially free ones) weren't suitable for brickwall limiting purposes. Thankfully that's a bit different nowadays.


What to do to prevent that (ISP):
Oversampling(!), properly limiting (including a ceiling for the DAC, hence the "old" 0,1dB/0,3dB/0,5dB), checking with a suitable meter.

There are no rules but one: check the statistics.


I can really, really recommend everyone to check out Inspector(!) from RogerNicholsDigital, even if it's that old. Or grab the free Solid State Logic X-ISM. Put your stuff to the test, see what the heck is going on.


If you need loud "master" files, this tool is essential. And this is what Dan is bragging about. What good is a limiter ceiling, if you can have intersample overs at 0,3dB as well?

If you're mainly working in EBU R128 or up to K-14, and you use a fairly modern limiter, chances are you never enountered this problem. Because nowadays a lot of (brickwall) limiters really are like a darn brickwall, or your digital peak doesn't even reach -2dBfS digital (as it is the case with EBU R128 and some K-20 material).



Try it for yourself. Overdrive your limiter, use meters with ISP/True Peak and see how good your brickwall limiter really is.



BTW:
Some back-info. Couldn't find another source (I had a lexica for audio terms but I can't find the link off hand):
Solid State Logic (Product Page of X-ISM)
Peasurize Music Foundation (I'd take some of their philosophies with a grain of salt now)

[Integratron]

[Dean Aka Nekro]

Just so I don't get painted as a metalhead, sometimes I actually do master lower than 0dBfs and I do pay attention to intersample peaks. I do it that way for projects that can't tolerate any distortion, and require a smooth clean sound. Often they're the more sparse and dynamic projects.

-Kim.

Nothing wrong with us metalheads Kim at all, It is not like we all think and indeed practice audio smashing, I hear more of that side-chained 'Electronic' dance muck smashed than i do majority of metal albums...I ain't no fan of 'Death Magentic' or anything else Metallica have released since maybe their self-titled (aka The Black Album), Although there are a few decent tracks on load, reload and St.Anger but mostly filler/treading water. Which is where the 'Metalheads just smash material and the audience does not care' type stigma has come from i think. I ain't going to recommend any albums that are not smashed as there are too many good ones, Metallica are well not really metal as i know and love it in its many forms

Whats wrong with a little paint either? You know you really love ya corpse paint really:

You love it dude \m/ ^"^ \m/



Shining Black Leather

[Dean Aka Nekro]

I'm surprised its proving so controversial to be honest!

Same here Dan mate, I mean i always check via RND InspectorXL. Thought using an ISP meter was the 'norm' dude It always has been for me and the people whom have mastered mixdowns i have done, No matter how nice thier D/A of choice is they have always been specific about that, Therefore so have i been as when having to do home-brew masters (yeah i would rather NOT but well money does not always factor into the job these days now i am part-time and mostly working for/with no-budget work) it has IMHO payed off to apply the same principles minus the very nice specific/custom gear obviously and a room as nice (If were are talking more than one track as if its a lone wolf then its not really mastering and i also i do not do the other essentials which make up alot of a real ME's job such as getting the information side all done...etc PQ codes but rather simply making say 10 tracks sound like they belong together, sequenced and thats it aka home-brew job as they are usaully friends). Get more money doing teching for bands than i do studio stuff

All the best your way anyways, Peace

Dean