PG Music Home Forums General Discussion Recording, Mixing, Performance and Production Audio Normalisation-Is It Normal?

Hi Experts,
Just to continue getting my money's worth in the PG forums this holiday period,and further develop some audio skills...

Is the use of audio Normalisation routine at any point in the mixing>mastering process?

I've been routinely normalising all my audio tracks relative to 0dB up to now,and while it doesn't appear to have degraded any work,I'm just wondering if the more experienced community use this technique?

If not,when and where is the best time to use this?

Cheers,

Neil C
Warrington
UK

Neil,

Yes. With caveats. To me it depends on the mix.

I use Audacity always as the next to last step in my process.

I do the best job of mixing that I can in the step before that, carefully watching my meters, and doing the best I can with EQ, also using compressors on certain tracks like bass and drums (or others if they need it.)

Then I will open the export in Audacity. I am looking for a .wav file that looks smooth and curvy (like a, well, wave) and not jagged or ragged with many spikes or transients. I am sure you know the picture I am talking about--it looks more like the image of a drum .wav than a song .wav---it looks "shredded." (The screenshot below gives you an idea of what a mix that needs compression and normalization will look like.) If I see that, I know I have no other choice but to compress and normalize. So I use a mild compression in Audacity (1.5:1), then normalize to -0.7.

Then I will use negative amplification to "trim the hair" of any remaining spikes that will reduce my headroom.

If it does not need compression at this phase, I will normalize it to -0.7 in Audacity before going further. My point is, if it is ragged you have to both compress AND normalize. If the mix looks elegant to the eye, all you to do is normalize, and yes, that is normal.

From here, I proceed to mastering in something like Ozone, or another tool, depending on the genre.

I NEVER boost to 0 dB with the next to last step .wav though. I always want to go into Ozone with a -0.7 mix so the effects used in the mastering tool (you have to have one) have room for transient shaping, maximizing, Eq-ing, etc. That is where you set your threshold.

I sometimes set the threshold to -.1 but most of the time -.2. I just don't think you have to max it out.

For the type of stuff I do, I find you begin to lose nuances when you max it to the limit and I would rather people hear dynamics.

I do not want to compress at the next to last phase unless I have to, because mastering tool presets add compression and I do not want compression times compression. But, if you start with a ragged .wav as your export, you have little recourse because it will be next to impossible to control your dynamics or loudness in the mastering phase before you clean things up a bit.

In the best of all worlds though, I try and mix so that I do not need to do this Audacity compression step before mastering, but it does not always work out that way.

Hope that makes sense.

Hi Neil,

I used to always normalise all tracks to 0.00 dB. For BIAB users who read this and don't know what 'normalise' means, the term refers to scaling a waveform (i.e. sound) so that its highest peak reaches 0.00 dB. Most peaks in the wave will be noticeably less than 0.00 dB. Normalising does not change sound integrity if it stays under 0.00 dB; in most cases, it simply raises the volume of the sound.

Then I read somewhere that it's always best to leave a little bit of audio headroom so that if a track has effects added to it, the effect processor has audio 'space' to work in so that it can maintain the integrity of the sound while adding, and after adding, the effect. In a number of articles, I came across the minimum recommended normalising value of -0.3dB and so I started using that.

These days, I normalise to -3.0 dB. I chose this value randomly after watching a friend mixing a song he had composed -- he is an audio engineer at a nationwide radio station and I noticed that he never normalised to 0.00 dB and always had substantial headroom above (and below) the highest peaks in the audio waveform. Since he is a professional in this field, I took his lesson on-board.

Audio that comes from BIAB mostly needs to be normalised. Wave volumes (that is, 'peak heights') are set by PG Music in a style such that the mixer will always show '90'. This means that some audio, when transferred to a DAW, will be noticeably quieter than other audio. To make it easier to work with these varying-volume audio stems, I always like to increase the height of the wave as it makes both editing and mixing that wave much easier. I'm happy working with -3.0 dB normalisation these days. It gives me a good clarity of sound.

Just my 2 cents worth.

Regards,
Noel

Good question. So.... yeah I tend to use it on everything as my LAST step. What it does is pretty simply. You select the level. As most have said they choose 0dB as the reference level. So, normalization looks at the file you are wanting to normalize and it finds the highest peak in the song. It moves that peak to 0dB. It amplifies everything else by the same amount. Nothing is amplified more than anything else, like would happen in a compression level to 0dB for example.

As I understand it, this is how it works. It's never given me any issues. They recommend -3dB for MP3. and Waves can go to 0dB..... BUT.... if you plan to have a mastering house master your song, they normally want around -6dB so they have room to work without squashing anything.

Learn something new everyday.

Thanks all!

...Deb

I use normalisation totally differently. When I pull the tracks into my daw (usually Reaper) I normalise all the tracks to -10db to -15db depending largely on the number of tracks. With all tracks playing I then like to be about -6db or less. I have a quick listen . I might renormalise (usually drums and bass) to put them up a couple of dB to have them better in the mix. I then do the mixing. At the end of the mixing process I like to have about -4db to -6db or less for headroom for the “mastering”. The final stage in the mastering chain has a limiter which I use to lift the track to -14 LUFS with a true peak of -1db. This way the songs all come out about the same level, reasonable balanced and ready for use.

In nearly all cases if I pull them into Audacity the songs are at -1db a different way to arrive at the same result.

I often use MP3 files to use when performing I put these through MP3Gain which in a sense renormalises again.

All seems like extra work but IMHO it in fact saves time.

Tony

I never use the Normalize function anymore. In Studio One there is a way to raise the track level gain by just selecting the event and dragging it up and down. This simply adjusts the volume level of the track without changing the dynamics of the track. It also does not change the Volume fader either. I like to get it around -12db per track and no more than -6db. I always leave headroom for when it is ready to master. Studio One also has a built in plugin called Mixtool that does this to the track as well as phase.

Most DAWs have a gain level knob of some sort for doing this.

Brian

Superb range of suggestions,many thanks everyone...and anyone else who wishes to throw in their Two Bob's worth..

I'm quite suspicious now that a problem piece of work I'm dealing with at the moment,also mentioned in this forum(which Brian also offered up some suggestions towards)may be due to the fact that all the guitar stems have been normalised relative to 0dB and that once mixed down,alongside the use of AudioSnap in Cakewalk,may be the cause of the issue.

I've never even considered normalising to a value other than 0dB,so this will definitely on the "To Try" list next time

Not a negative commentary on Cakewalk by the way,which must qualify as freebie of the year,merely my possible erroneous usage of AudioSnap.

Here's to Build 611 taking us to VST nirvana...

Thanks Guys!

Neil Cummins
Warrington
UK