PDA

View Full Version : Help with Normalizing in Sound Forge


econoblackcoffee
2007-05-29, 06:47 PM
I desperately need some help with normalization in Sound Forge.

1) What is the difference between peak normalization and RMS normalization? Does peak normalization raise the highest peak to "0" and all other peaks to an appropriate number relative to the "0" peak? RMS seems to work so much better for live recordings, but I just don't understand it.

2) What is the difference between normalizing and just using the "Volume" selection under "Process" to add gain to the recording?

3) Why under RMS normalization does the "Music" preset recommend normalizing to -16db? Don't I want the music to peak at "0"? When I use the preset, the wave looks like it does peak at "0" after the normalization is complete. Why is that?

4) Sometimes when I use the normalization feature it sounds like the sound is compressed afterward in certain loud spots. How can I normalize to the loudest possible degree without clipping or compressing the sound?

I know these are a lot of questions, and I greatly appreciate any help in any/all of them. I have been taping for about 9 years now, but I used to use the old-school analog EQ into a standalone burner to "master" recordings (my mastering usually doesn't extend beyond trying to get the levels hitting appropriately). I have just jumped into the school of SF, and I think I'm doing OK. I've done about 10 shows so far (off MD), and some of them have turned out perfect. The above questions stem from some issues I've had with those that haven't. Thanks.

Audioarchivist
2007-05-29, 08:15 PM
I desperately need some help with normalization in Sound Forge.

1) What is the difference between peak normalization and RMS normalization? Does peak normalization raise the highest peak to "0" and all other peaks to an appropriate number relative to the "0" peak? RMS seems to work so much better for live recordings, but I just don't understand it.

2) What is the difference between normalizing and just using the "Volume" selection under "Process" to add gain to the recording?

3) Why under RMS normalization does the "Music" preset recommend normalizing to -16db? Don't I want the music to peak at "0"? When I use the preset, the wave looks like it does peak at "0" after the normalization is complete. Why is that?

4) Sometimes when I use the normalization feature it sounds like the sound is compressed afterward in certain loud spots. How can I normalize to the loudest possible degree without clipping or compressing the sound?

I know these are a lot of questions, and I greatly appreciate any help in any/all of them. I have been taping for about 9 years now, but I used to use the old-school analog EQ into a standalone burner to "master" recordings (my mastering usually doesn't extend beyond trying to get the levels hitting appropriately). I have just jumped into the school of SF, and I think I'm doing OK. I've done about 10 shows so far (off MD), and some of them have turned out perfect. The above questions stem from some issues I've had with those that haven't. Thanks.
I don't use Sound Forge so much, I use Wavelab, but I can give my versions of answers.
1]
Peak levels are fast bursts of sound, and they are the ones that go up to zero. In oldschool analog tapedecks, you could go a bit above zero ( chrome and metal tape saturation absorbed the excess nicely) but in digital, above 0 means bad things happen. Trust me!
RMS is more like average volume level. I used to see amplifier/receivers with ratings like 1000w peak/300w rms, meaning it could give short bursts of 1000w, and steady 300w. Something like that anyway.
I don't use the automatic settings to normalize (Wavelab has them, too) and I'm not sure how it would translate exactly to a normalizing setting, but I think I set up things to let peaks reach zero (in some cases peaks above zero could happen, but I put a limiter plugin on, too) and rms levels to be in -14 to -15 range average, louder passages up to -12db. So normalizing to -16db will probably give music enough headroom to still have those peaks almost and up to zero.
If you look at the history generally of compact discs and the increase over the years of mastering levels, lots of todays music goes up to -6 or 5db rms, and it's compressed to shit with peak levels riding the 0db mark. Older stuff is like -20db, and peaks only got maybe -2 or 3db or maybe there could be one peak that got up to the zero mark on the whole record...plenty of natural uncompressed headroom - dynamoc range intact - not compressed.

Uh, ok. I'm answering more than one thing at a time.

2]
Not sure on SF, but Wavelab it's not much. Process is like an automatic feature, and using volume control is more manual control. Eventually still does the same thing. You might have to watch out for those above zero peaks, though.

3]
RMS -16db will (I think) still allow most music's dynamics to peak occasionally to zero. Very dynamic stuff (loud vs. quiet) might need an even lower setting like -18 or -20 to avoid too much clipping, like the modern stuff I mentioned.

4]
Stuff can sound compressed because it is after too much normalization. Too many peaks that would be above zero, too high rms level, probably was ok before processing...

5]
I, too use old school analog eq for years before doing the digital thing. I still will use it to ReMaster analog sources (my master cassettes really benefit from it still).
Like I say, headroom and not boosting those levels too much is a good thing. That's what our volume knobs are for. The modern mastering level race to -0db / competition to see who's record is louder was kind of started for the radio. Even though radio is compressed to hell, back when cd's were still mastered for the home audiophile market in mind, some brilliant record executive thought they could make their band's music stand out on the radio (like the commercials on tv stand out - louder), so they started pushing the mastering levels up a little bit at a time from one guys record to the next in a show of one up-manship...
Mastering properly is great, but over-remastering isn't so good. Some people would prefer an un-processed recording over anyone's ReMaster reworked sound version, even if it is an improvement. Of course I'm on your side, and I would want the copy of my recording to go out into the world sounding as nice as I could make it. Keep practicing and you'll do fine. My last bit of advice is that less is more. Don't overdo anything. If the processing you did distracts from the musical quality (ie:you notice the harshness of the processing more than the music) then you've done too much. Have you ever heard de-noised recordings that sounded de-noised? That's what I mean...
Good luck!