Tips and Tricks for the Normalize function?

[grannis]

 

(Process->Apply Effect->Normalize)

Looks like it should be useful for smoothing levels on a clip or track, but I can't quite figure it out and can't find the doc on it.

 

[Sock Monkey]

Will you don’t seem to understand what normalization does? It is not a compression technique. It is an automatic process that is identical to simply turning up the channel gain control. It just will automatically stops turning it up if the loudest peak reaches the set level. 

It’s downside to audio purity is that it alters the data and yes this could result in distortion especially if the software sucks. Or you push it to hard. All digital processors suffer this same issue.
But for a lot of material and if you don’t try and push it to -1 but keep it reasonable at like -6 the audio will be unscathed. 

Like most digital audio processors in the hands of knowledgeable users it is a useful tool. 

[msmcleod]

I think there's a general misunderstanding here regarding headroom. 

Regardless of how loud or quiet the waveform is, ALL audio within Sonar/CbB is converted to 32 bit float (or 64 bit float if using the 64 bit engine) as soon as its read from the file on disk.  This effectively gives you unlimited headroom during the mixing process (it can literally go to 1000's of db or 1/1000ths of a db).  It's only when it reaches your mains out that its then converted back to 24 bit integer, where at this point it could clip, or be too quiet.

That being said, the volume fader on a track/bus only has a finite range.  This is where normalize can be useful, as it can make a quiet sound louder or a loud sound quieter, meaning that the fader can be used on a more useful range. 

Yes, there's offset mode - but again, it's still limited in its range.

Normalize will not change the existing dynamic range of a clip.  All it does is make it louder or quieter.  

The most common use for normalize is to make quiet audio louder. If the audio is too quiet, then once you've pushed the track fader up to +6db,  it can't go any louder.  Normalizing the audio will make the whole clip louder, allowing the fader level to be reduced, and allowing you to move it up or down.

As for the OP's original question...  normalize will NOT smooth out the levels of audio.  All it does is turn it up or down.  To smooth out the dynamic range, use a compressor plugin.

[Glenn Stanton]

19 hours ago, Will. said:

No yoy dont create headroom. You'rw just lowering the volume. This means every plugin you run through still runs the risk of distorting. 

https://www.google.com/search?q=what+is+audio+headroom

[mettelus]

Audio Normalization is nothing more than raising/lowering the entire clip or track by a constant value to hit the target you specify. It does nothing to "smooth" or change the relative dynamics of the audio. Where this can be useful is to prepare tracks to a given level prior to mixing and/or so that they hit more common threshold targets for follow-on FX (specifically compressors) more consistently. Visually, normalization does essentially the same thing as applying clip gain (CTRL-click-drag) on the waveform in the Track View (just that with normalization you are specifying the target for the peak in the entire clip and a single gain is inserted on that entire clip to make it land there).

[msmcleod]

1 hour ago, mettelus said:

Audio Normalization is nothing more than raising/lowering the entire clip or track by a constant value to hit the target you specify. It does nothing to "smooth" or change the relative dynamics of the audio. Where this can be useful is to prepare tracks to a given level prior to mixing and/or so that they hit more common threshold targets for follow-on FX (specifically compressors) more consistently. Visually, normalization does essentially the same thing as applying clip gain (CTRL-click-drag) on the waveform in the Track View (just that with normalization you are specifying the target for the peak in the entire clip and a single gain is inserted on that entire clip to make it land there).

It's worth noting that it will raise the volume so the peak level hits the target you specify... not the average level.

It is useful for raising a very quiet recording to a sensible level though.

[Amberwolf]

The only thing I use the function for is if I have a bunch of little snips of something (usually percussive, or sounds I recorded to use as effects, like my dogs' breathing, panting or other wierd sounds they make that I can reprocess into something musically useful). 

 

I slice up the recorded or imported wave into separate clips (either manually or with the audiosnap feature that does a fair job at it), then select them and press my shortcut key for Process-Audio-Normalize and just hit Enter which does it to 100%.

Then I can use the sounds wherever, and use track volume or clip gain envelopes (or the destructive Process-audio-gain function) or compressors, etc., to manage the individual volumes of the sounds wherever they're being used. 
There's a few examples of this usage in a recent project here, with some panting and other sounds in certain sections as percussive timing effects,

https://amberwolf.bandcamp.com/track/ookami-no-kari-no-yume-wolfs-dream-of-the-hunt

and in this one replacing some shakers with pantings. 

https://amberwolf.bandcamp.com/track/a-peek-over-the-wall

[grannis]

great answers thank you all.

The fact that it normalises to the peak could in itself be quite useful (thank you @msmcleod)... so if you normalise to -6db, and see that a couple of peaky peaks are stopping the whole from getting close, you can then clip-gain those down and repeat. 

It's not something you could do with a limiter, clipper or compressor on its own. Has anyone experience artifacts from this approach? feels like you shouldn't but I'm not an expert on the comparative cleanliness of clip gain editing vs limiting or clipping.

As for the destructive editing (thanks @Sock Monkey) I guess a null test would prove or allay any concerns regarding quality degradation... not sure I can be bothered though - I wonder if anyone has already done it...

 

as an aside, I've always felt that batch processing of clips should be a useful tool to augment the real-time algorithms of limiting etc. I presume that plugins can't do this, it has to be a built-in DAW feature 

[Chaps]

6 minutes ago, Will. said:

No yoy dont create headroom. You'rw just lowering the volume. This means every plugin you run through still runs the risk of distorting. 

As a mastering engineer I will send the track right back to you cause I will run some issue in the mastring process to it. You cant normalize everything whether increase of decrease you are not creating headroom rather you're just lowering the overall volume. 

Sure, okay. If you say so.

[sjoens]

Why not ask google? It knows everything. 

https://www.google.com/search?q=does+normalizing+audio+raise+head+room&rlz=1C1RXQR_enUS1094US1094&oq=does+normalizing+audio+raise+head+room&gs_lcrp=EgZjaHJvbWUyBggAEEUYOTIHCAEQIRigATIHCAIQIRigATIHCAMQIRigATIHCAQQIRigATIHCAUQIRigAdIBCjE1NTEyajBqMTWoAgiwAgE&sourceid=chrome&ie=UTF-8

For "smoothing levels" it might be more useful for the OP to use a limiter instead.  But I really don't know.

Edited Wednesday at 08:36 AM by sjoens

[sjoens]

I love it when voting throws all the posts out of order.... 

[mettelus]

3 hours ago, sjoens said:

I love it when voting throws all the posts out of order.... 

Yeah, I was thinking the same thing... I pulled this up on my phone and the conversation is so disjointed now you cannot track the conversation. Moving a solution to the top makes sense, but re-ordering every post doesn't... is like ripping all the pages out of a book and randomly handing them to someone one at a time without page numbers.

[User 905133]

6 hours ago, sjoens said:

I love it when voting throws all the posts out of order.... 

Is the SORT BY DATE option not working any more?

UPDATE: It looks like our sorting preference gets reset.  😞 

Edited 14 hours ago by User 905133
answered my own question

[sjoens]

fixed

[Sock Monkey]

The difference is that applying normalize or gain processes the original audio permanently and this is known as Destructive Editing. 
Simply increasing your level using the tracks gain control or using clip gain is Non-Destructive and can be easily changed at any time. 
Some people don’t like the idea of destructive editing as it does render the audio file and has a potential to alter the audio files quality.  And it can’t be undone after the project has been saved and closed. 
I have never heard a difference in quality myself so I often use it on live recordings when a few tracks are recorded way too low. 
 

To use it this way I will set it at -6 db globally for all tracks. 

Be aware that if the audio clip has just one loud spike it will get normalized to that one spike. The track will still be too quiet overall. 
You sometimes need to manually remove any peaks first before proceeding. 
 

If you’re lazy you can just use a limiter like Boost11 set at -3 db and use the track  or clip gain to push it up . 

Edited January 16 by Sock Monkey

[Will.]

On 1/18/2025 at 12:32 PM, grannis said:

great answers thank you all.

The fact that it normalises to the peak could in itself be quite useful.

Not really, cause you're robbing yourself from the original headroom the file had. 

[Chaps]

5 minutes ago, Will. said:

Not really, cause you're robbing yourself from the original headroom the file had. 

I usually normalize a recorded audio track (vocals and direct guitar) to -1 dB  in Adobe Audition 1.5. and then hard limit it -.1 dB at a time until any outrageous spikes are tamed. Usually once or twice is enough. That leaves me with a good dynamics range with no spikes. Then I'll normalize the vocal audio between -6 dB or -12 dB for use in Cakewalk, depending on how many audio tracks I'm using (usually more than 30). Another situation where I normalize audio close to peak is with a recorded direct guitar where I want a lot of crunch. When the guitar track is where I want it after any processing, I freeze it and normalize to between -6 dB and -12 dB. I mostly use the track/bus gain controls to balance the volumes since, pretty much, every audio track I use has volume and pan automation. It is perhaps a clumsy way to do things but it works for me and it has become intuitive.

[Will.]

1 hour ago, Chaps said:

I usually normalize a recorded audio track (vocals and direct guitar) to -1 dB  in Adobe Audition 1.5. and then hard limit it -.1 dB at a time until any outrageous spikes are tamed. Usually once or twice is enough. That leaves me with a good dynamics range with no spikes. Then I'll normalize the vocal audio between -6 dB or -12 dB for use in Cakewalk, depending on how many audio tracks I'm using (usually more than 30). Another situation where I normalize audio close to peak is with a recorded direct guitar where I want a lot of crunch. When the guitar track is where I want it after any processing, I freeze it and normalize to between -6 dB and -12 dB. I mostly use the track/bus gain controls to balance the volumes since, pretty much, every audio track I use has volume and pan automation. It is perhaps a clumsy way to do things but it works for me and it has become intuitive.

If it works for you it works for you, i guess- who am I to judge right? It is just: I would never want to do what you do. OUCH! There are way better techniques out there and they all preserve the original files headroom and performance. 

Just remember whatever you normalize you're doing destructive editing and its  squash the headroom away. PLUS, Theres compression and limiting, OUCH! Oxygen is too important to the body. 

Think of normalization as a picture image that has been cropped. Once you print it out theres no way to restore back what you have cropped out. 

[Chaps]

18 minutes ago, Will. said:

If it works for you it works for you, i guess- who am I to judge right? It is just: I would never want to do what you do. OUCH! There are way better techniques out there and they all preserve the original files headroom and performance. 

Just remember whatever you normalize you're doing destructive editing and its  squash the headroom away. PLUS, Theres compression and limiting, OUCH! Oxygen is too important to the body. 

Think of normalization as a picture image that has been cropped. Once you print it out theres no way to restore back what you have cropped out. 

Normalization does nothing to affect the dynamics of an audio track, just increases or decreases the 'volume' in relation to the loudest peak of the audio. If I normalize an audio track that peaks at -6 dB to -12 dB I am actually increasing the headroom without changing the dynamics in the audio. Personally, I like all my audio tracks to be about the same volume before I start mixing them.

[Will.]

37 minutes ago, Chaps said:

If I normalize an audio track that peaks at -6 dB to -12 dB I am actually increasing the headroom 

Lol. You say what? No you dont. That's a misconception hobbiest and newbies have when using Normalization. 

When you Normalize a file you "Raise" its headroom along with the file volume of the file, running the risk of distorting your sound. You do not "increase" your headroom - you "RAISE" it up with your audio. 

Edited Tuesday at 03:38 PM by Will.

[Glenn Stanton]

actually -6 to -12 is "increasing" the headroom by 6db (lowering the level of the peaks is going to quiet all things). going from -12 to -6 is "reducing" the headroom (moving the peaks up and everything else is going up as well).