Normalization?

A

AumStudioBrian

New member
I'm confussed on how to get songs onto a CD and have them all sound the same volume? I'm working on a 13 song Album for a friend and all the songs were recorded at different times, over a 3 month period. I'm having trouble getting all the songs to sound similar. Any advice would be helpfull.

*(I'm using Pro Tools and Waves Plugins if that helps with any of your replies. Thanks)*
 
Disposable said:
My guess would be normalization and/or compression...
Click to expand...
Guess again........ normalization does very little to increase overall, unless the track has no transients........

Some form of Limiter processing is useful in levelling out transients and increasing overall level........
 
But if Brian is used to waves plug ins presumably he's familiar with compression and limiting? Is the problem more to do with different rooms, mics, instruments, tracking, mixing equipment? Is there an EQ approach to making this all sound more homogenous? I don't know - just asking.
 
I recently pulled Har-Bal out of mothballs and began using it again since it just got an update to v1.5. It's a parametric EQ with a visual spectrum type interface (as opposed a picture of hardware with knobs). The spectrum shows both average and peak curves of your song along with any reference curve you may wish to match the balance of.

I'm using that guy right now to do the exact same thing. It sounds better to since I'm not trying to push my song into a compressor or limiter to adjust loudness. If I want to adjust dynamic range & headroom I can always grap a comp/lim to adjust the dynamic balances and match the collection closer at some other point in the process. Right tool for the right job.

Voxengo CurveEQ has a spectrum matching function to adjust the EQ balance and add color using the current song and a reference curve of your choice. I've tried CurveEQ to balance a set of songs but usually I end up just using it on a song to add cool color - not to match another songs loudness level.

I'm getting some good results using Har-Bal visual interface to do this so far. It really helped me with the resonances & details in the bass as well as broader strokes for the mids and highs to get the overall balance correct. But I'm a visual guy anyway...

I just wouldn't use a compressor/limiter to match loudness levels since you'll be changing your songs' dynamic headroom that way.
 
kylen - I'm struggling a bit with HarBal, which I see as an ear training device rather than a serious mastering solution. Any tips? (Without hijacking Brian's thread.)
 
Normalization is a gain multiplier that operates on peak signal levels - nothing more.

Once the peak value of the signal is determined, a gain multiplier (or divider) is applied to cause the peak (and consequently, the entire signal along with it) to reach the desired threshold (typically 0dBFS, but it doesn't have to be...)

There are 2 problems with normalization -
1) it raises the noise floor along with the rest of the signal

2) It tends NOT to raise overall level significantly because transient peaks often spike very close to 0dBFS, and so the actual amount of gain applied is usually minimal.

Example- you have a signal who overall level is around -10dBFS, except for a number of transients who peak at -1dBFS.

The rookie figures normalization will raise the level since just by looking at the waveform it's obvious that the bulk of the signal is nowhere near 0dBFS. They start the process expecting to gain a good 8-9dB at least -- when the process concludes, they find they've only had a gain of about 1dB........

I'm sure you can figure out why that is -- Normalization is a gain multiplier based on peak value.

Well what if you limit the peaks first, then normalize -- sure - that will work, but a) that's limiting not normalizing; and b) you'd be degrading the signal a digital generation down using 2 processes (limiting then normalizing) when limiting can both tame peaks AND raise overall gain as part of the same process.
 
The Waves L2 plugin is the best for increasing the volume without mutilating the track.
 
Waves L2 is a pretty expensive plug in for most home reccers (and has been the cause of much angst among some very knowledgeable people here because it's squashed the life out of many commercial tracks). What's wrong with using some automated fader riding to bring down the peaks and hence allow an increase in the overall level?

Also I'm not sure that the original question has been answered yet.
 
Garry Sharp said:
What's wrong with using some automated fader riding to bring down the peaks and hence allow an increase in the overall level?
Click to expand...
That's exactly what a limiter does.........

Garry Sharp said:
Also I'm not sure that the original question has been answered yet.
Click to expand...
It has -- a combination of limiting and judicious use of EQ are tools used to tonally balance multiple tracks..........
 
I got ya Blue Bear...
Would you guess I've never used a limiter? Haha...
never even really knew what it was, to be honest.

I'll have to try it out...
 
Garry Sharp said:
Waves L2 is a pretty expensive plug in for most home reccers (and has been the cause of much angst among some very knowledgeable people here because it's squashed the life out of many commercial tracks). What's wrong with using some automated fader riding to bring down the peaks and hence allow an increase in the overall level?

Also I'm not sure that the original question has been answered yet.
Click to expand...

Having done a good number of compilations I can tell you that trying to match recordings from sessions that are months apart is no easy feat (let alone a bunch of different bands recorded at different studios with different styles).

A couple of suggestions (in no particular order):

1. Work from the loudest song first and get an overall level (e.g. -10 RMS Dbfs). Then work the other songs toward the loudest using a combination of compression and limiting (keeping in mind that all songs should not necessarily be at the same volume). Automation is good too, but it usually works best over a wider range, while limiting is better for short transient material.

2. Use a spectrum analyzer (unless you have experienced ears) and match the EQs between "like" songs.

3. Listen for things that stand out from track to track for example sibilance, uneven frequency response, and try using your Waves multi-band compressor to even things out, or automating an EQ.

4. Burn a CD of the tracks and listen on as many systems as you have patience for, see how well the CD translates overall and from track to track. It can be surprising to hear how differnent tracks sound when played on multiple systems and can bring out the worst and best.

Hope this helps ...
 
Garry Sharp said:
kylen - I'm struggling a bit with HarBal, which I see as an ear training device rather than a serious mastering solution. Any tips? (Without hijacking Brian's thread.)
Click to expand...
It's serious enough for me - I balanced a dozen songs for a project in an evening - I've never worked that fast before and I am very bad with a parametric EQ but not this one. Now I'm adding a touch of upward expansion (the project was kinda flat coming from an old tape) and will proceed to push all the stuff into a mastering limiter (Voxengo Elephant2 - it's very transpanent) like masteringhouse and others mentioned.

masteringhouses's step #2 is what harbal does - I just do it first because the stuff I deal with I know the EQ is more out of balance than the dynamics. I'd rather adjust the 'density' last anyway but that's just me - and I'm a garage guy so who cares what I do :D

I have noticed something with Harbal - it's taken a couple of months to get it understood, I had to get decent monitoring and acoustics so I could hear what I was doing and understand which peaks to push on, I don't allow any bumps in the Harbal spectrum below 200Hz or so. If you see a bump your bass is too uneven - this does not remove the dynamics or punch in fact it does just the opposite. For the mids or highs I use wider strokes to balance the EQ.

As far as the normalization thing goes - even if you have something like Cool Edit Pro that does rms normalization it will only get you close - you still have to manipulate the levels by hand at some point to get them to fit together. And if you sequence an acoustic duet with a crunched metal tune you'll laugh at what auto-normalization will do...hehe or cry.

Everything changes and goes faster when the monitoring improves - is that obvious or what - this is audio !
 
Kylen - that's very interesting. My room and monitoring gear will bever be anything like good enough for more than basic mixing, but I can still learn from HarBal and will remember your comments.

Blue Bear - presumably limiting gives different results than using the faders to bring down peaks yourself??
 
Gary, as I understand it a limiter puts a ceiling on a track volume above which it cannot go, but allows for dynamic variation below that limited level? This is obviously not quite the same as manualy moving a slider to try to control the peaks. Though having never used one, I'm just going off what I've read!!!
 
Glyn, the cause of so much frustration with much recent commercial stuff is that they use the limiter to squash everything right down, which means they can then bring everything right up again, making it all very loud with no dynamic variation!

At least using the fader riding thing, you and not a piece of software are making the loudness decisions, and can then use much more gentle limiting to finish it off, as it were.

Bob Katz describes a technique called upward compression. Blue Bear or somebody will correct me if I've got this wrong, but in essence it seems that you take two versions of your final stereo mix, and apply compression to one of them. This brings down the peak volumes relative to the quiet parts, but preserves much more dynamic range.
 
Garry Sharp said:
Bob Katz describes a technique called upward compression. Blue Bear or somebody will correct me if I've got this wrong, but in essence it seems that you take two versions of your final stereo mix, and apply compression to one of them. This brings down the peak volumes relative to the quiet parts, but preserves much more dynamic range.
Click to expand...

I think that the better term for this is parallel compression, to me upward compression is akin to expansion.

see:
http://www.nthelp.com/mastering/faq/specialtechniques.htm#anchor463964

BTW, this isn't a BK invention, see:

http://homepages.paradise.net.nz/~rhulse/Side%20Chain/sidechain.htm
 
You must log in or register to reply here.
Back
Top