averaging window

[tmcconnell]

I recently mastered a record for an irish band. Normally I have it done by professional mastering houses but these people could not afford it. I begged them to spring for it - but they insisted I do it.

To cut a long story story short, I read the books (in a hurry), I installed a vast quantity of absorbers, I bought xenon, and I did several spectrums of the room with several spectrum analyzers. I checked the master on many systems, over several iterations. I think I did ok.

Yesterday, one of the band members, a phd biophysicist with a good ear called and suggested that one of the songs was too quiet relative to the others. The cd had not been sent to reproduction yet. I looked at it and it looked like the others, but sure enough, it seemed quieter to the ear. Then I invoked my trusty PAZ meter, and checked the levels with several different parameters. The big difference showed up when I set the averaging window to about 250 ms. I used A weighting and rms settings. With "peak" the levels were identical of course since I had already mastered the material.

The difference in the levels between that song and the others was entirely dependent on the time over which the average was computed. This would seem obvious - but my question is: is there a standard window (length for averaging)?

I discovered that the difference for that tune and the one before it was greatest with a 250 ms averaging window - and that difference was about 4db. Then I simply pulled xenon up 4 db and set the release to a setting at which it sounded right (too long and bass was mushy, too short and bass went away). I was done.

The tunes then meshed nicely together. It sounded right - so alls well that ends well, but I still wonder if the length of the averaging window is something that's specific to the material or someting I should use as a standard setting.

[MASSIVE Master]

Your ears are going to tell you more than a loudness meter.

[scraggs]

yeah. you have to use the ears. if you can`t hear that one song is 4db quieter than the one before it, well....

also i think it`s a good idea to get the levels between songs pretty consistent BEFORE the limiter, not WITH.

[john hartley]

As John said above, your ears are better than a loudness meter for relative levels between tracks - that's how the client (and any other listener) is likely to do it.
A suggestion for practising this is to take the loudest track of your album and adjust the others to it.
Meters are great tools but ultimately, with everything in mastering, you must use your ears.

All the best,

John

[tmcconnell]

Thanks. I heard a difference, but I had asumed it was intended by the artists to be quieter material. In measuring that difference I got a range of answers depending on the averaging window - but then remembered that crest ratio is also an "average" to peak. So... why do people rely on crest factor if the time over which the average is computed can completely change the results??

[MAzevedo]

I don't think anyone 'relies' on crest factor. Most ME's are managing level my using a consistent, repeatable monitoring loudness and relying on their ears. I've got a Dorrough meter right in front of me, but I'm using is as an idiot check. Spectral content has too much to do with perceived loudness for overall level to really capture how 'loud' something is.

[tmcconnell]

Good perspective - thank you. t.