Rodabod wrote:I would have thought that using proper noise reduction might be more useful than just using eq alone. There are good modern noise reduction technologies available which can be very effective.
...Watch out for over-compression your material
...Remember that the order that you process in is vital - for instance, compression before noise reduction might not be a good idea.
i agree completely with rodabod. this sounds like the precise plan i would try first... noise reduction of some sort, and then a dynamics processing that would increase the level of the quieter parts to where they ought to be, and essentially leave the loud parts alone.
i have to get on my other computer to look it up, but i know there's either a Noise Reduction built into Cool Edit, SoundForge, or a plug-in that i have somewhere, that's really advanced. you select a segment of your program that represents the noise signature, and then it actively filters the entire program to drop out that noise when it's the only thing present, and decrease its presence when there are other things happening on top of it. i need to look up exactly what it is, i've only messed with it a couple times as noise is rarely the problem with my recordings, i'm dealing more with bad EQ, phase cancellations, and dynamics issues...
i would definitely deal with the noise first. and then i might try the Loudness Maximizer, but there may be a better approach even than that. i don't recall if SoundForge has this, but i know Adobe Audition does, where you can *draw* a compressor. it's an X-Y plot, with incoming level on the X axis and outgoing level on the Y. flat, it's just a diagonal line from lower left to upper right (upper right is 0,0 and everything else is negative relative to that point). so for example if you drew a horizontal line at the -10dB level on the Y-axis, that would take absolutely anything that comes in, and make it -10dB, your meters would be completely motionless sitting there at -10dB no matter what was playing, be it loud, soft, tape hiss etc.
so what you could do with that (dynamics processing -> graphic compressor, i believe it's called in CoolEdit/Audition) is to analyze your program material and get a feel for what the levels look like, and then use a compression scheme that will smooth it out a little.
like, if the good program material all pretty much falls in the -15 to -5 range let's say, and the quiet parts are down more like -30 to -15 and in there, what you could do is make a graphic compressor that will take anything below -40 and cut it out completely, take the range from -30 to -8 or -10 and make it end up basically being a much more horizontal line than it is, bringing up the left side (the -40 side) quite a bit (say, to -15 or so) and leaving the point -8,-8 or -10,-10 exactly where it is.
so what you'd end up with is, from left to right on the X-axis (incoming level) from -infinity to -40, it would be cut completely, to -infinity on the Y-axis. at -40 on the X-axis, the output would be maybe -15 on the Y-axis, and then there'd be a nearly horizontal line to the point -10,10, and from there on up it would be flat, normal, a 45 degree diagonal line.
the -40 low cutoff point and -10 point are both just wild guesses. without having analyzed your actual program material, i have no idea what the real points will be. but you get the idea...
there's a low threshold that's maybe 10 dB lower than what you think your quietest useful program material might be (you can push it close to what you think your quietest level is, but going too far can make things sound more weird than they need to).
there's a section that represents your "quiet material", the stuff that needs to be made louder, which will go from your cutoff threshold up to whatever point you think is the crossover between "too quiet" and "alright", and again here i'd maybe push the actual point a little higher into the "alright" range, but NOT TOO HIGH. if you go too far, you're gonna make it sound weird. i'm guessing.
experiment even with a cutoff of anything below 40, and then drop in two points, one at -40, -20 and another at 0, -15. that'll give you a pretty tight output range, basically only a 5dB range... that will probably sound pretty unnatural. but somewhere in between that and the more conservative approach i outlined above, there will probably be a happy medium that sounds okay.
i would NOT try to make it sound like everything is all consistent and at the same level. i would try and make it so the parts that are too quiet are still quiet but not *too* quiet. i would not try to make it so the parts that are *too quiet* end up sounding the same as the stuff that's loud and clear. or rather, i would try it, but think it highly unlikely that it will happen without sounding unnatural, and potentially very annoying. i don't know how much Listener Fatigue is an issue for what you're doing, but tweaking stuff too much can result in something that sounds *really good*, but after a minute or a few minutes, you have become agitated and don't want to continue listening. kinda like what can happen with rock music being overproduced. eventually its true unnatural nature comes through.
do you have SoundForge 4.5, or a newer version? i can check my plug-ins when i'm on my other computer, and check the software too, to make sure it has this same functionality that i know Audition has. it's been a few years since i stopped using SoundForge because i thought CoolEdit Pro had it beat for 1 and 2-track editing.
out of curiosity, what's the nature of the program material? what kinda 80's detroit show? is it bad ass?
