Peak level @ 0, -0.1, -0.2 or -0.3dBFS?

Posted: Tue Feb 22, 2005 4:11 am

Out of curiosity what do you guys use as standard?

I know (from listening experience and from a TC Electronic paper) that a lot of consumer cd players will additionally clip the signal beyond 0dBFS when D/A'ing and things are so hot.

In which case I suspect -0.3 instead of 0dBFS doesn't make any difference - probably should be at least -3dB or something..

But in other respects, there could be a difference in 0dBFS or -0.1dBFS peak level, I remember reading something about mp3 conversions generally not liking anything hotter than -0.1 or -0.2 (Bob Katz)?

Most commercial cd's tend to be either 0dBFS or -0.1dBFS (pop/rock/dance) so obviously these facts are mostly ignored in the loudness race (in which I "happily" participate myself).

Posted: Tue Feb 22, 2005 7:54 am

I use -0.1db

Also, anything out there will clip the signal beyond 0dBFS??

Moderator Joined: Sep 12, 2002 Posts: 1884 Location: NYC

I'm using -.3db on hotter CD's and -.1db on not so hotter CD's. I notice a tad difference by going -.3db but not much, mostly I just like the number 3.

Posted: Tue Feb 22, 2005 9:15 am

I've been using -0.3db just because I've heard that its safer. Michael brings up a good point though, -0.1db is probably safe if RMS volume isnt too high.

Posted: Tue Feb 22, 2005 9:17 am

On my side, I use -0.2db or -0.3db for hot wanted cd and go for -0.1db for smoother stuff.

Regards

Posted: Tue Feb 22, 2005 10:05 am

I process & limit at either 96 or 88.2k depending on the original Fs and set my limiter for -0.3dBFs so that after SRC I don't get any peaks above 0.0...

Posted: Tue Feb 22, 2005 10:09 am

...and dithering Laughing

Posted: Tue Feb 22, 2005 10:43 am

lucidwaves wrote:

-0.1db is probably safe if RMS volume isnt too high.

Anything greater than 0.3 and there is potential for intersample overs if you are not using an oversampled 100% brickwall limiter in my experience.

We've measured intersample overs using the L2 as the limiter and this is the number that we observed as safe.

The TC MD4 is a complete, brickwall oversampled limiter, so you can go higher if you like. If you're using the L2, I think it's better to be safe and stay around 0.2 or 0.3.

You can read the TC Electronic paper on this subject here, good read: http://www.tcelectronic.com/media/nielsen_lund_2000_0dbfs_le.pdf

Posted: Tue Feb 22, 2005 11:41 am

This is a great article ! Thanks!

Richard

Posted: Tue Feb 22, 2005 1:20 pm

leonardkravitz wrote:

We've measured intersample overs using the L2 as the limiter and this is the number that we observed as safe.

The TC MD4 is a complete, brickwall oversampled limiter, so you can go higher if you like. If you're using the L2, I think it's better to be safe and stay around 0.2 or 0.3.

And have you listened...?
Don't get me wrong, but it often seems like you guys like to match numbers instead of listening?
I good test IMO is to make some test CD's with different levels and play them on a range of diiferent CD players.

The Brickwall meter in the system 6000 is an odd thing IMO. It shows that newer albums has up to +3db output, however this isn't posible and it seems like the meter is "calculating" artificial overs that aren't really there.

Posted: Tue Feb 22, 2005 4:37 pm

I get a lot of 0dBFS masters too, Björn Engelmann in Cutting Room has done some great dance track masters, he doesn't seem to worry about making it hot as hell, so 0dBFS seems to be his standard for dance singles.

I guess I'll stick with -0.1dB for now on singles.

Moderator Joined: Sep 12, 2002 Posts: 1884 Location: NYC

No, I think the TC meters are accurate. there is a breakdown of different players and different CD's and how much they were clipping the material at the DAC. i think it was a phillips paper, I can't remember. But it was not uncommon for them to clip as much as +5db on the output. maybe I can find the paper and post it.

Posted: Sat Feb 26, 2005 11:40 am

that would be nice.....

i would really like to know a bit more....

after trying the m6000 with ammitsbøl i am wondering about the whole "intersample peaks" thing..... is it just another case of "the emperor's new clothes" ?

by the way..... how would one write a digital word of +3dBFS?????

1111 1111 1111 1121???? Laughing

Posted: Mon Feb 28, 2005 12:45 pm

Ammitsboel wrote:

I good test IMO is to make some test CD's with different levels and play them on a range of diiferent CD players.

Did you read the TC paper? This is almost exactly what they did.

Ammitsboel wrote:

The Brickwall meter in the system 6000 is an odd thing IMO. It shows that newer albums has up to +3db output, however this isn't posible and it seems like the meter is "calculating" artificial overs that aren't really there.

iznogood wrote:

by the way..... how would one write a digital word of +3dBFS?????

They ARE really there--not just possible, but CERTAIN. It's quite intuitive when you think about it. Here is an excellent paper that will clarify intersample overs:

http://www.cadenzarecording.com/papers/Digitaldistortion.pdf

Posted: Mon Feb 28, 2005 1:09 pm

thanks for the link....

i think this paper implies that all system not having oversampled peak meters do not accurately display the peaks.....

that means 98% of all DAW out there!!!

and it also means that if you mix in the analog domain and all of your signals are full code then you essentially have maybe 64 d/a's distorting at the the same time......

my conclusion is..... if this is true almost all digital audio up until now is useless!! Confused

by the way .... ammitsbøl tried putting some full code raw waveforms like sine and sawtooth into the TC intersample meter..... they all came out over odBFS!!! Confused

This topic sponsored by: