Mpeg audio layer II peak limiting. Options

[Aciaci]

Audio processing and limiting is very basic thing to me. I have to limit peaks to -12dB..no problem in wav but when I encode it with video file to mpeg2 and audio layer II there cames these extra peaks from this encode process. I have studied why that happens and understood samething about it but haven't found any way to awoid it. The question is: is there one and is it only the thousands of dollars hardware or is there a software? Now I´m losing aboout -3dB rms. And yes..it must be as loud as possible..

The final mpeg2 multiplexing is done with adobe premiere and I have also sony soundforge where is that same problem. Win7.

[saratoga]

Lossy encoding will increase peak amplitude a little. It cannot be avoided, so you probably need to lower the volume more before encoding or else decrease the volume after decoding.

[DVDdoug]

I have to limit peaks to -12dB

Why? Where is that requirement coming from? I'm just wondering if it might be acceptable to keep the peaks below -12 before compression, and just live with the quirks of lossy compression.

Now I´m losing aboout -3dB rms. And yes..it must be as loud as possible..

A compressor/limiter (which I'm sure Sound Forge has) can boost the average/RMS level (and loudness) without boosting the peaks. You'll have to play with the settings and learn how to use it, but the usual procedure is to reduce the peaks and then boost the average with "make-up" gain. In your case, you may want to normalize to -12dB (or -15dB) instead of tweaking the make-up gain.

P.S
Of course dynamic compression reduces the dynamics of your program. But, a few dB of compression is usually acceptable (maybe not even noticeable), sometimes desireable, and if you want it "as loud as possible", you have to compromise...

This post has been edited by DVDdoug: Dec 21 2011, 00:52

[Aciaci]

It's a TV standard to limit peaks to -12db. Any peaks over that and the file is not approved. If I process the encoded layer II file I can't just save it in sound forge. I have to save as mpeg and then these peaks come again. Like I said normal audio processing is very simple but this audio layer II encoding is the problem. Is there a better encoder? Is there a encoder that can calculate exact input gain before encoding so the peaks are "normalized" to -12db? Now I have to test it many times so I can found the optimal level that doesn't clip over -12db.

This post has been edited by db1989: Dec 22 2011, 12:38

Reason for edit: deleting pointless full quote of last post

[Sunhillow]

If you have to do it professionally, you might ask at Institut für Rundfunktechnik in Munich for their Layer II encoder
http://www.irt.de/de/produkte-beratung/dig...rfunk/mpeg.html

there is a PDF available, but I don't know if it is free of overshoots. At least it is widely used in german broadcast

This post has been edited by Sunhillow: Dec 26 2011, 01:01