A difference of .1dB is only audible under ideal conditions, so I don't think it make sense to quibble at that small of a level difference.

OTOH, a 1dB difference is quite easily audible.

So if I really really wanted to absolutely maximize peak level for some reason I would probably start -.1dB and reduce it until the difference got big enough to both hear and actually care about.

For me personally, when talking about the overall level, anything less than ~.5dB difference just isn't enough of a difference to really care about. YMMV.

I agree. Think about how small a half a decibel is when you're mixing. I normally don't even bother with half-db tweaks when mixing. If it doesn't need a full db, it doesn't need anything.

The problem with pushing right up against the ceiling is that you leave no room for intersample peaks, which can be as much as 3db over the highest peak your meters register. This is exacerbated by MP3 encoders' high-Q filters. So if it's a choice between compromising fidelity and making the listener turn the volume knob up half a decibel...

And anyway, who are you competing with, trying to squeeze the last db possible? All your competitors on the radio?

Yea, I habitually master stuff at -.3 and when converted to mp3, I still sometimes get into the red.

@

Jeff Evans


Well Ben that is interesting. Personally I think even that is very high. There is still something to be said for staying well clear of a lot of converters that might still have trouble sounding good at -0.2 dB max level.  A good upper limit might be -1dB but then you are loosing some volume when you do this. How silly all this is when we can should be well clear of any of these levels.
(And we have got a whole whopping 96 dB of available range below it even at 16 Bit!)

I am thinking of enrolling in the Masters of Music technology at Newcastle Uni ! What do you think. Would you recommend it? 
 
We've had this disscussion before Jeff, the one about enrolling in the masters course at The University of Newcaste that is and I am glad we can put our differences behind us too.  My advice is this, I am still unemployed and it doesn't look like changing in the near future whether this is due to my attitude or being over qualified, (which I am now, for the grunt jobs in the audio/music industry) it's probably a little of both but the course hasn't enhanced my resume.  On saying that if you want a useless title (yes I know it is a part of my monkier) and the possibilty of a better paid job, then do it.  It won't teach you much that you don't already know, unless you want to learn Flash or to get a bit of research under your belt.  The one thing it does give you though is good resources, such as books and a copy of Flash if you did the Flash course and these are part of the course fees.  But I'm ambivalent about the merits of the course.
 
Peace Ben 

I don't know if the "your competitors on the radio" part was aimed at me but, I'll answer anyway: I have the misfortune of not liking anything whatsoever that gets onto mainstream radio, so, short answer-- no. However, I do want my tracks to sounds as professional as possible, and often, mainstream standards are what their name entails-- ubiquitous. So, though I may think my track sounds good the way it is and all the listener has to do is crank up the volume level, I also picture the scenario wherein the listener has put my track in a playlist, say, and the track before it peaks at -0.1dB, then mine comes along peaking at -1dB or less (which I may have kept that way in order to maintain audio fidelity and what have you) and the listener just skips ahead because it lacks the energy (volume) of the previous track, given that we all suffer more and more from attention deficiency, as well as one not possibly expecting the listener to know as much about music technology as we do (statement to be read in the most humble way possible). It needs to be on par with everyone else's work in my field, and you can be sure that everyone therein is 'fighting the loudness war', which is a very real thing and definitely needs to be adhered to if I would ever want my tracks to stand out or, at the very least, alongside. Nothing to do with radio, however. More so clubs... Interesting stuff about the different peak levels when exporting audio in different formats! I'll definitely keep an eye out for that and make the best compromises possible. Thank you all!

I also picture the scenario wherein the listener has put my track in a playlist, say, and the track before it peaks at -0.1dB, then mine comes along peaking at -1dB or less (which I may have kept that way in order to maintain audio fidelity and what have you) and the listener just skips ahead because it lacks the energy (volume) of the previous track

I understand completely. But this is exactly the trap so many fall into: the belief that 1 decibel in peak values will make the difference between being competitive or not. 

I usually set my peak limit at -1dBFS and let it do it's thing.



I just did an amplitude analysis on a AC/DC Back in Black .mp3... it has 1799 possible clipped samples.

Then I checked Adele's "Rolling in the Deep" .mp3... it has 25,000 possible clipped samples.  The signal just hit's 0dBfs and turns into a square wave over and over again.

Both peak at 0dBFS.



Doesn't bother me one bit... I'm sticking with -1dBFS peaks for my personal guideline... I think it keeps your digital from exciting the ugliest part of someones' analog and people don't seem to mind that at all as long as the average levels are perceived as not too quiet.


best regards,
mike

I also just did an analysis of Tom Petty's Refugee... I think it is a masterpiece of sorts so it seemed like a good example to mention.

It has less than a dozen peaks over -3dBFS with one hitting -1.3dBFS max.

Most of the peaks don't go above -4dBFS.

In other words, the peaks are, well... the peaks are actually peaks. :-)



OH BTW, unlike the other 2 examples... the Tom Petty .mp3 also has zero DC offset... you'd think someone could fix that on the other 2 tracks.

I think it might have been mastered by an actual mastering engineer.





Refugee sounds plenty loud when I play it.

Knowing that there are great songs like that out there bolster my opinion that I can get away with peaking at -1dBFS and happily disregard any opinions that the mix isn't loud enough.





Heck, that is one reason I have monitors with adequate power amplification ... so I can easily demonstrate to guests how loud a mix can seem.




best regards,
mike

 

Understood, mike and bitflipper! Thanks for your feedback! Very interesting comparison/analysis too, mike!

radio uses broadcast limiters, all set the same.

it will take the input from any source (cd) and make it just as loud as anything else.

the problem is trying to match that kind of limiting as a final product for a cd.
this is the wrong way to go about things.


a well-mastered pro release, will sound good on the radio.

a home made master, usually will sound pretty crappy on the radio.

even if the levels on both cd's are the SAME...... it's the WAY the pro release is mastered, that makes it sound superior when put thru ANOTHER level of limiting.

home-mastered tracks that have the low end dialed in pretty good, but do not have the high end properly adjusted, will sound bright, splatty, a bit harsh, when put thru the radio limiter.

that radio limiter, is the thing that brings all poor mastering to light.