Howdy folks
 
I've been sent the stems of a friend's track for a remix project. After loading the stems into X3 Producer I noticed that there was very little wave profile and levels were peaking at maybe a max of -12db. I've done a bit of googling on normalizing and from what I can see I should be able to do that without compromising the audio. Being a bit cautious I'm thinking of normalizing to -5db rather than 0db. I quickly tried that on a couple of tracks within the project and the wave profile is certainly much more visible and I'm didn't blow my monitors :). I didn't save anything at that point and I'm wondering if anyone here has any thoughts on the matter.
 
Just a note on the stems. It's not going to be an easy job for my friend to redo them so I'm hoping that normalizing will be the answer.

Are you mixing or what? Are the wave files in 24 bit or greater? What sort of music is it? And what are the instruments? Has any FX been added? What is the RMS dB value?
 
I would do the processing that you think is needed and go from there. I would only normalize if that was the only way to get the final levels I wanted. 
 
 

The first question will be: What bit depth did he send you?
 
2nd: How many stems are you going to sum?
 
3rd: Have you tried summing all the sends to see how the levels stack up to see if you really need to normalize them before mixing?

I must be blocked. LOL

There are a couple of useful tricks for working with low levels
- Gain nob - seems obvious but sometimes people forget. It works great. Can't be automated. Pretty much the same as normalizeing in a lot of ways.
- when in track view and drop down is set to 'Clips' on the track header, you can click in the gray divider and drag up and down to increase the size of the wave form so you can see the wave form better for quiet parts of the track
- Normalize... Normalizing to a particular level for a particular tune is a bit like setting the gain in a way that sticks. In other words, say you have a recording of 20 songs in a single project and each song needs separate gain settings. Instead of using the gain nob, you can normalize the audio and thereby control the gain into the fx sends for each song in a way that is remembered over time. If you use the gain nob on one tune, then alter it for the next tune, then go back to the first, you will need to reset the gain nob on the first tune. Can you remember where it was set? If you use normalization, you will not need to reset when you return to an earlier tune.

If you do normalize, make a backup copy of the file before normalizing or any other destructive editing.

John
Are you mixing or what? Are the wave files in 24 bit or greater? What sort of music is it? And what are the instruments? Has any FX been added? What is the RMS dB value?
 
I would do the processing that you think is needed and go from there. I would only normalize if that was the only way to get the final levels I wanted. 
 
 

Remixing. 24bit. Sort of folky dub. Drums, percussion, bass, guitars, synths, the usual suspects. Stems are dry to allow further processing. As mentioned the highest peak was about -12db. 
 

mike_mccue
The first question will be: What bit depth did he send you?
 
2nd: How many stems are you going to sum?
 
3rd: Have you tried summing all the sends to see how the levels stack up to see if you really need to normalize them before mixing?

1. 24bit
2. I'm at work, but from memory would be about 18-20
3. With everything at unity gain everything does sound fine but the master level was definitely lower than I would normally mix to.
 

John
I must be blocked. LOL

Why?
 

gswitz
There are a couple of useful tricks for working with low levels
- Gain nob - seems obvious but sometimes people forget. It works great. Can't be automated. Pretty much the same as normalizeing in a lot of ways.
- when in track view and drop down is set to 'Clips' on the track header, you can click in the gray divider and drag up and down to increase the size of the wave form so you can see the wave form better for quiet parts of the track
- Normalize... Normalizing to a particular level for a particular tune is a bit like setting the gain in a way that sticks. In other words, say you have a recording of 20 songs in a single project and each song needs separate gain settings. Instead of using the gain nob, you can normalize the audio and thereby control the gain into the fx sends for each song in a way that is remembered over time. If you use the gain nob on one tune, then alter it for the next tune, then go back to the first, you will need to reset the gain nob on the first tune. Can you remember where it was set? If you use normalization, you will not need to reset when you return to an earlier tune.

Thanks gswitz, thats really helpful. I'm still getting my head around Sonar so if people oftem forget about the gain nob I'd not have had much chance of using it. And I tried to find a way to increase the size of the waveform but obviously missed the trick. I really didn't want to normalize at all so I'll follow your gain nob approach. Thanks again.
 
Cheers
 
Mark

Lynn
If you do normalize, make a backup copy of the file before normalizing or any other destructive editing.

Thanks Lynn, I was only going to normalize within X3 so the original files will still be as supplied. Looks like I have the solution now though.
 
 

John
I must be blocked. LOL

You guys are crazy (not you Soft Enerji).
 
Ok, for more info on audio wave form scaling see this help file...
http://www.cakewalk.com/Documentation/default.aspx?Doc=SONAR%20X2&Lang=EN&Req=EditingAudio.13.html
 
So... reasons for using gain or normalization can include 'gain staging' 
http://www.soundonsound.com/sos/sep13/articles/level-headed.htm
 
Gain staging is really the main reason to use the 'Gain' nob, thus the name. :-)
 
You can bring up the levels into your FX and etc so you can mix well.
 
Keep in mind that you also have Gain nobs on busses and they can be turned down as easily as up. If you are clipping slightly on the master bus, you can turn down the master bus gain a little to solve your problem, rather than re-setting the levels of all 100 tracks individually.
 
Normalization has less justification, imho. As I described in my post, if you are doing gain staging that you want to stick, you might try auditioning the tracks dialing in gain and then apply normalization to stick that gain permanently to that tune. (specifically, listen to the track noting the loudest moment during the tune when the gain is set where you want it. If the loudest moment is -2.5, then normalize to -2.5 and remove gain adjustment). Then you can set gain back to 0 and if you return to that section of the recording, you're good to go. This only applies if you try mixing multi-track recorded concerts in an efficient manner where your goal is just to bounce out that tracks and you don't plan to spend a ton of time on it. If you are going to really polish each track, you're better off breaking each track into separate projects. 
 
I have a video in my signature walking you through some of these ideas in a way that I think of it.
 
If you are only mixing one song per project... your friend's stems only apply to a single song... then you probably don't really need to normalize anything. Keep in mind that you don't have to normalize so that every available bit is used during the loudest moment of the selection. You can normalize to -3 dB or any other amount you choose.
 
Cakewalk FX (and others) often emulate real hardware... thus, they permit you to overdrive the hardware, just as you could in real life hardware. For that reason, you don't usually want to normalize to 0 (as loud as it will go). If you do, you will send the loudest possible signal into your compressors or EQs or whatever and they may distort for you on purpose... like that's what you want!! So back off a little. Use the gain nobs and listen to what you get back from the compressors and EQs and etc.