Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Normalization Revisited (Read 4873 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Normalization Revisited

I brought my post over from another topic because it was getting buried and it didn't get the response I thought it warranted.

>>
Isn't the "real" problem with normalization is that it un-dithers a dithered signal (and may raise the noise floor)? Re-dithering after normalizing a lossless *.wav file (a final step) seems like a 9.9/10 way of doing things, especially with sources that are way weak in signal strength. For certain types of music normalized & re-dithered sure would beat Replay Gained files (because all players support it). I have a few tracks on CD's which EAC reports as in the 30 to 40% range (signal strength). The opposite is true too....tracks in the 98 to 99% range. It's been said that lossy encoding can push this level into clipping. How would you get around this? Only way is to normalize DOWNWARD to prevent this (if your doing a digital rip, then you've got no control over the recording level.....with analog rips, you do). The only thing u can do is post process the wav file....and that necessary evil is called "normalization".
>>

Am I off my rocker? Let me know...  ...  ...don't be shy.

Xenno
No one can be told what Ogg Vorbis is...you have to hear it for yourself
- Morpheus

Normalization Revisited

Reply #1 –
Quote
Isn't the "real" problem with normalization is that it un-dithers a dithered signal (and may raise the noise floor)?


I know this isn't what you're asking for, but I think the "real" problem with normalization is that most users use it because they think it will make their files equally loud.


Quote
It's been said that lossy encoding can push this level into clipping.


ReplayGain is the one and only _good_ solution to that. Peak level normalisation doesn't really make sense in any situation IMHO.

By the way - the process WaveGain uses to change the volume on WAV files is identical to the normalisation process other software uses, but, as John33 pointed out in that other thread, it can dither. So - wouldn't that be the solution to your problem?

(Perhaps I misunderstood you... )

Normalization Revisited

Reply #2 –
With the average normalization routines that don't use dithering, you are indeed increasing quantization errors, and the resulting noise is quite unfavorable from a psychoacoustic viewpoint. I once read that correlated quantization noise sounds about 12 dB worse than analog noise with equal energy. This is mostly audible at low volume levels. So, when you can't use lossless methods (ReplayGain) for whatever reason, and want to use normalization, dithering is strongly advised. If you 1) can't use dithering, or 2) could use ReplayGain/MP3Gain/etc. instead, DON'T normalize! (this is for all you EAC users...)

Normalization Revisited

Reply #3 –
Thanks guys....

Volcano >> but, as John33 pointed out in that other thread, it can dither. So - wouldn't that be the solution to your problem?

It sure would  if it does indeed do that.

I've heard alot about normalization being a bad thing, and per my first sentence "Isn't the "real" problem with normalization is that it un-dithers a dithered signal (and may raise the noise floor)?", was trying to figure out what the reasoning was behind it. I should have mentioned distortion as a by-product of the un-dithering, as u guys did.

Xenno
No one can be told what Ogg Vorbis is...you have to hear it for yourself
- Morpheus

Normalization Revisited

Reply #4 –
... and how about converting the original wav to 24 bit or 32 bit, then normalize this one and make your mp3, ogg, mpc etc from it?

This way the additional quantization errors will be much less audible

Normalization Revisited

Reply #5 –
Quote
This way the additional quantization errors will be much less audible

still, the "audible" factor here is the additional "layer" of dither noise, which can be quite noticeable when compared with the original soundfile

this is the reason why i dont really like normalizing wave files
Be healthy, be kind, grow rich and prosper

Normalization Revisited

Reply #6 –
Quote
It's been said that lossy encoding can push this level into clipping. How would you get around this? Only way is to normalize DOWNWARD to prevent this (if your doing a digital rip, then you've got no control over the recording level.....with analog rips, you do).

The big problem and a big part what makes normalization completely useless is that how much you have to lower the volume depends on the music you are encoding. A level that prevents clipping in one clip may still cause clipping in another one. To make things worse, most people that use normalization normalize to 98 or 99% and think it will stop clipping. Not.

As already noted, ReplayGain doesn't suffer from this problem, but note that WavGain may if you encode the wavgained files.

The chance is not very big because the reference level is quite low, but it's still there.

Normalization Revisited

Reply #7 –
Just wondering, what is the status of of CLI wavegain? Is it on par with all the windows binaries floating around and is it still being updated? Also, is there a mirror for the replaygain website, because www.replaygain.org doesn't work for me.
 

Normalization Revisited

Reply #8 –
Quote
Just wondering, what is the status of of CLI wavegain? Is it on par with all the windows binaries floating around and is it still being updated? Also, is there a mirror for the replaygain website, because www.replaygain.org doesn't work for me.

Dunno about wavegain, but the Replay Gain website is now hosted at http://replaygain.hydrogenaudio.org/