Venues actually tend more toward 95-105dB, not 85dB... though that is based on some random resources on the net (this is not new research, I built my gray noise a couple years ago)... 85dB was a compromise between club and (loudish) home stereo.
I can't really provide a link, as there was no resource that just said "here's how to make gray noise." I had to figure it out myself.
I did misspeak slightly, though... you actually need to shape
pink noise, not white noise. It's important when going from SPL to digital that you still follow that 1/
f line.
Grab the ISO:226-2003 curve values - this can mean needing to pay for the ISO if you can't find the values. I found them in a matlab thingy back when I made it, but can't find it online any more - if you can find something similar, I recommend that as a cheaper alternative. I would paste them here, but it's a lot of data (most of it useless, as we're not going to be mixing for very much below 60dB SPL).
Next, grab yourself some pink noise, slap on a 32 band EQ, and adjust the channels to match the values from the ISO standard. Depending on the range of your EQ, you may need to use two in parallel - my EQ has a range of -18 to 18, and the maximum range at 80dB is 37.16dB. (Looking back at my spreadsheets, it was 80dB, not 85, that I used). I highly recommend not using the values for 20Hz and 20KHz (and possibly 25Hz, your choice there, but I don't use it), as they are outlier values that will give your curve extremely harsh edges.
After that, you'll probably also want to add on a strong low pass around, say, 50Hz and a high pass around, say 15KHz, just to soften things a little bit further.
Finally, adjust the level so it's at 0dB, and there ya go, gray noise.
If you don't like 80dB, then you are of course quite free to choose a different value. I actually have three gray noises that I built, 60, 80 and 100.
You can extrapolate the curves to any dB SPL value you want, but I'm not going to go into that here.
I am using this technique in the DJ software I wrote to measure all the tracks in my setlist's frequency spectrums, then at performance time, measure the level of the room (I am working on a realtime meter that uses a radio transmitter to send this directly to my laptop so that this part happens automatically), and adjust the curve of each track in real time to fit the room so that it can adjust the sound to the levels in the room... but that part is still in prototype stage, I don't use the room leveling yet when I play out. I admit that I am not completely sold on using it all the time anyway, for the same reasons as what I detailed earlier about following the curve perfectly not being
interesting enough to me.
Here's a screencap of what my gray noise looks like in ProQ2. Note that the spectrum is rotated 4.5dB per octave to make it more readable, as is the standard with ProQ2 (and many other EQs).
