Dramatic dialog remastering

[gjmail]

I am re-doing a dramatic reading of a novel that was recorded on reel-to-reel tape, and which I transferred to CD using a home stereo burner (not the one in my computer - the kind you use as a home stereo component) and then ripped the CD into Sound Forge on my computer. It's going well, but I could maybe save a lot of hours of experimentation if someone who has done this before gave me some tips on Normalizing, or the Wave Hammer Volume Maximizer if you have seen this process on Sound Forge or something similar. I burned the CDs just barely below the clip level, so normalizing to peak does virtually nothing. The signal mostly stays below minus 4-6dB, but often hits a loud word that maxes out near 0dB. Normalizing to RMS, using dynamic compression to avoid clipping, is about the same as the Wave Hammer (though WH is much quicker & easier).

What I'm worried about is losing dramatic effect of the loudly/forcefully spoken words by compressing them in order to bring up the volume of the rest of the signal. I've experimented a little, and it doesn't seem to diminish the dramatic effect appreciably, but I want to be sure to get the best finished product possible, so I will have to experiment a lot more before I'm sure it's OK. Have you had experience with this problem? Is there any intrinsic need to maximize the volume, or will I do just as well to leave the volume as-is and just let the listener turn up the volume on his stereo to accomplish the same thing? Or does that (turning up the volume of the finished product) worsen the signal-to-noise ratio? (The recording is already somewhat noisy, even after applying noise reduction software to it). Scanning a region of my recording with the scan set to ignore any areas below -40 dB yields an RMS of around minus 20-25 dB. I also don't have any idea whether that is loud enough, or if there is an "industry standard" recommended level for that figure.

I would appreciate any pointers you could give me. Thanks!

Richard Thayer

[juppu]

Cool subject... There's quite a bit of work like that around.

I personnally use the Waves L1 Limiter and the Steinberg Denoiser on things like that. I work at a radio station and have to edit some interviews from time to time, some of which are of a poor quality. I don't know if these plugins are available for Forge, but you can use them within anything that accepts VST plugins.

The L1 is basically a hard limiter with three controls: threshold, output level and release. It sounds pretty damn natural unless you go over the top (-40dB threshold etc...). Make sure you don't reduce more than 3dB at the maximum peak, it's very unnoticeable like that. What you can do is put 3 or 4 of them one after the other and make all of them reduce 3dB - very subtle, very loud.

The denoiser reduces noise. http://www.audioforums.com/forums/smile.gif You also have a control for ambience reduction, so it's great for my purposes - to make things intelligible. It works pretty well at moderate settings. I think it's some kind of a frequency-concious expander + maybe does some phase-twiddling and dither-sort-of noise shaping. Anyway, it takes away noise.

Beware! The Denoiser has a bug, it gives a short signal burst just before the actual signal starts, so you have to find a way around it. I bounce the thing down, edit the unwanted bit out and export the remaining region.

This is how I do it and it works.

I hope this helped...

Juppu

[Plec81]

First of all.. NEVER normalize audio. What you're doing is applying massive amounts of DSP to your entire file. This brings your audio down one generation and it WILL sound worse. You never gain anything from it anyway. What you're doing is getting the loudest peak to hit -0.0dBFS. This does nothing for you. The ear always responds to RMS and not Peaks. So normalizing is totally useless.

Interesting work you got cut out for you there. Since it's supposed to be dramatic you should not compress it. Every single dB of compression takes away the dramatic effekt. I would've just used a peak limiter to make sure that the loudest peak hits -0.2dBFS. Maybe a dB or two of limiting. With that you have maximized the experience for the listener. As far as noise reduction goes. Be very careful when using bad quality plugs for this. Most plugins on the market aren't that good. The only one that's decent is Waves Restoration Bundle. Listen very carefully. If it sounds more alive and more dramatic and more natural without any NR, just leave the noise as it is. Of course, if it's a lot of noise there's nothing to do about it but to use the best tools you have for the job...