Normalising question ...

I just read in Computer Music magazine, in a song they put together using just soft synths, they suggested bouncing each soft synth to an audio track as a WAV, then normalising each WAV before mixing ............ is this a good technique, to normalise each track?   

It would be easier to answer if the statement was put in context. Computer Music usually has decent advice, knowing the full context might put it in perspective. Nonetheless I'm sure many will be happy to answer without knowing the reason it was suggested.
 
Personally, as with cake2010, I'm not sure I see any benefit in normalizing, but knowing the magazine there is likely a reason it was suggested.

The trouble is, if you do normalize all of your tracks, when they summed at your main outs, there's a very good chance you'll get digital clipping.

"Normalize BEFORE mixing"? - so what happens if, as is quite normal, your mixing raises the overall level of the track(s)?

I'm with Cake2010, absolutely no point in normalizing if you're recording @ 24 bit.

Negating the effect of normalizing.

I am a little confused about what CM is suggesting. If one is working with soft synths wouldn't we balance levels before we freeze? I do. To than normalize seems counter productive. 

I also agree with using 24 bits. I can see normalizing if after the freeze a track or two is not fitting right. But that should not happen if you are watching your levels before a freeze. 

Also I don't bounce anymore I freeze.

Bristol is right.

Also CM is not the most accurate publication. If this were in SOS I would want to read the whole thing to find why. Being in CM I am not going to listen to it.

BTW what else would one create but a wav file through a bounce or freeze?

"they suggested bouncing each soft synth to an audio track as a WAV, then normalising each WAV before mixing"

swim at your own risk imho...  if the OP is relaying the article correctly, than the advice is pretty bad

Bristol_Jonesey


Another thing, as soon as you insert a soft synth you should know immediately what sort of level you're dealing with and you should address that right there and then, by raising or lowering the synth output, way before you even consider mixing/freezing.

I am of the opinion that people who need to normalize are not gain-staging properly throughout the project.

Or know what the heck they are doing. Much like CM.

I agree with Bristol and Cake2010. If you are concerned with S/N-ratio, you don't solve this by normalizing every single track since the noise is augmented accordingly. Use a decent audio signal to the system instead by correcting the distance to the mikes and adjust the volume levels on the amps. 24 bit gives a lot and sufficient headroom. Also never normalize to 0 dB since some systems (MP3-players, CD-players...) render digital clipping even though it sounds correct in Sonar. I usually go to minus 0.5 dB.

Its an old article - from this issue  http://www.musicradar.com/computermusic/computer-music-166-july-issue-on-sale-now-455790

The magazine was making a "chill out" track from scratch, using soft synths (and one vocal sample).  They got each sound (piano, drums, bass, pads, guitar etc) ready, with effects etc, and tracked out the song.  They also added some effects to the master bus.  

Then, to the mixing, and here I use their words :

"Let's get mixing!   With our master bus effects bypassed, we bounce each of our seperate parts down as WAV files and normalise them all."

They then set up a new DAW project, and re-loaded the master bus effects.  Then, in their words :

"We set the mixer faders for the stems to -5.0dB".    And way they went in the mixing process.

So, the way I work is the balance my relative volumes before bouncing, or at least I would balance the volumes of the bounced tracks without normalising them.   I'd get my Master bus around the -3dB level for all the (un-normalised) tracks together, then export the final stereo wave for mastering.

As far as I see it, if I normalised all the tracks, I'd just end up lowering their relative levels, and come up with the same -3dB result for the whole song together anyway ...

When I work with synths, I rarely have any sort of level problems. 99% of the time a synth track renders out at decent levels.

Occasionally, an audio track from guitar may not record with a strong level even though the input is showing it to be robust..... with a bit of red showing in the input meters.  So if the wave is a bit anemic, I will normalize it. 

I almost always run a 96% normalization on all my tracks AFTER I have exported the final mix into my wave editor. The tracks are lightly compressed at this point and have many peaks that are not exceeding 0db. 

Normalize simply raises the overall level a tiny bit in most cases with out changing the ratio like a compressor would do to achieve a few db more out of the track. 

this mp3 wave pic: which has been normalized




is this song:  http://www.soundclick.com...34&songID=10369122

Using Normalization is just like any other tool in the DAW.... it is there to be used, and it is possible to abuse it. 

Yes Mike, I usually save 2 "final" copies - one with synths frozen, one without.