Spoken Word Mastering

[Brent Evans]

In my never-ending quest for perfection, I'm now working on tweaking my mix for the sermons from church. The raw files contain massive shifts in peak level (sometimes as much as 20db), as our pastor has a very dynamic style. i find this not to be listenable, and I have it processed to about 6db between the loudest and softest peaks, normalized to -1db, with a noise floor averaging about -65 (to my eyes watching the meters). I find it very pleasing to the ear, but I'm curious if this is "normal" or if I need to relax it a bit.

For this sample, (http://www.filefreak.com/files/80918_vgkra/Preaching%20Sample.mp3) I have selected a brief portion that includes a shift from the softest level to the loudest level, and then back to the softest. If any of you would be so kind as to advise me, I would greatly appreciate it. This is a short clip, ~30s or so, and I could post a much longer segment if desired, but this is fairly indicative of the entire file.

Thanks!

[studio-c]

It won't be pretty, and I'd seldom use it for music, but SoundForge's "WaveHammer" plug (comes with it) does a good job of ramming everything to the rails. I usually batch process 32 hours worth of stuff as I go to lunch.

When I do multiple day recordings where they want it edited and mastered quickly, I use this. People listening in their iPods or cars comment to me that they like how they can hear everybody so clearly.

So you do lose subtleties in dynamic range, but it evens things out quickly and well.

I'm sure some extreme compression and the Levelizer in SS would work as well.

Cheers,
Scott

[Ian Alexander]

Brett, I listened to the clip twice on my cheap little computer speakers. Once with the volume at typical youtube levels. Worked fine. Didn't want to turn it up or down at any point. Then with the volume at the point where I could barely hear it. I could hear all of it, intelligibly. Again, no need to turn it up or down.

But through the clip, I could still get the intensity clues from his delivery and energy. Even with the reduction in dynamic range, you haven't taken that away.

Works for me.

BTW, if you're thinking of web delivery of these services, the dynamic reduction you've done will be necessary. It will help you keep things listenable with the data compression needed for streaming or downloading of such long programs.

[bcorkery]

Brent,

It sounds pretty clean from what I heard. I like the fact that there's still some dynamics in the sample, amen!

Every situation is different and I tend to do kind of what you described and then maybe mark a few of the really softer passages and bump them up a little and see how that sounds.

[UpTilDawn]

In my never-ending quest for perfection, I'm now working on tweaking my mix for the sermons from church. The raw files contain massive shifts in peak level (sometimes as much as 20db), as our pastor has a very dynamic style. i find this not to be listenable, and I have it processed to about 6db between the loudest and softest peaks, normalized to -1db, with a noise floor averaging about -65 (to my eyes watching the meters). I find it very pleasing to the ear, but I'm curious if this is "normal" or if I need to relax it a bit.

For this sample, (http://www.filefreak.com/files/80918_vgkra/Preaching%20Sample.mp3) I have selected a brief portion that includes a shift from the softest level to the loudest level, and then back to the softest. If any of you would be so kind as to advise me, I would greatly appreciate it. This is a short clip, ~30s or so, and I could post a much longer segment if desired, but this is fairly indicative of the entire file.

Thanks!

What I heard sounded pretty darn good over my desktop speakers.

I don't hear any problems.... as you said, it sounds pleasing to the ear.

[Cary B. Cornett]

Did you do any noise gating of this? I hear spots that sound as though the decay of the room gets "chopped" during a pause. As for the overall dynamics on the preacher, I pretty much agree with the earlier comments. Intelligibility is preserved without losing the sense of dynamics in the delivery. Were it not for the occasional odd thing with the background noise, I could have been fooled into thinking it was "all natural", and I imagine that is what most listeners WILL think.

So the peak levels before processing vary only about 20 dB? I might have expected more. When I was mixing some live theater stuff (using SSLite), I found that I could expect about a 30 dB variation in level on a performer's microphone (usually a wireless body pack setup). For the live mixes, I would routinely set up the console channel compressors with a -16 dB threshold to catch the occasional "runaway" peaks (singer "punching" a high note, etc.). I tried to set preamp gains so the highest peaks would hit the converters at about -2 or so.

Anyway, sounds to me like you are headed in the right direction.

[Brent Evans]

Did you do any noise gating of this? I hear spots that sound as though the decay of the room gets "chopped" during a pause. As for the overall dynamics on the preacher, I pretty much agree with the earlier comments. Intelligibility is preserved without losing the sense of dynamics in the delivery. Were it not for the occasional odd thing with the background noise, I could have been fooled into thinking it was "all natural", and I imagine that is what most listeners WILL think.

So the peak levels before processing vary only about 20 dB? I might have expected more. When I was mixing some live theater stuff (using SSLite), I found that I could expect about a 30 dB variation in level on a performer's microphone (usually a wireless body pack setup). For the live mixes, I would routinely set up the console channel compressors with a -16 dB threshold to catch the occasional "runaway" peaks (singer "punching" a high note, etc.). I tried to set preamp gains so the highest peaks would hit the converters at about -2 or so.

Anyway, sounds to me like you are headed in the right direction.


The raw file Runs through a 2:1 expander plug with the threshold set just above the noise floor, then through EQ to give a slight mid boost (the Countryman B3 we use is suprisingly flat) then through a compressor plug at about 4:1 above 20db (loudest raw peaks are about -6 usually), with auto makeup gain, then through the Waves L2 processor to bring levels up and limit to -1.2db or so.

My pastor usually has about a 20db difference in peak levels, but I have used the same processing with other speakers having up to a 40-50db peak level difference, with similar results (the after-processing swing was about 10db on that speaker, so I lowered the threshold in L2 to make up the difference).

I hear the choppiness from time to time but I've played with the expander and I can't seem to get it completely out. Any suggestions on that?

Thanks everyone who listened!

[Ian Alexander]

Having listened again at a higher volume, I'd say try it without the expander. Just consider the background noise as part of the live ambience and leave it there. The compression will push it down here and there, but that will happen as the level of the signal is high, so it won't be as easily noticed.

I think others have written here in a discussion about room tone, and I agree, that background noise can be acceptable at surprisingly high levels if it's consistent. Noise at a lower, but changing level, is more noticeable.

As is, the expander is even cutting into the preacher's breaths here and there. IMO, that's more harm than good.:)

[Cary B. Cornett]

The raw file Runs through a 2:1 expander plug with the threshold set just above the noise floor, ...
I hear the choppiness from time to time but I've played with the expander and I can't seem to get it completely out. Any suggestions on that? First you need to ask yourself what you are using the expander for. Usually it is for getting rid of some objectionable continuous background noise during pauses. Thing is, any time you get above the threshold the noise is there again, and often the "pumping" of the noise is more objectionable than just leaving it alone. If your preacher had trouble competing in level with the noise, gating would do no good. I would get rid of the expander, run the process again without it, and see what I got.

The compression, of course, ends up making all of the soft stuff, including noises, louder once the gain makeup has been applied. The real trick is to keep the gain from coming up during pauses. There are some compressors with the ability to "freeze" gain changes during pauses. This feature may go by some odd name like "intelligent release" or "auto release", meaning that the release time is program-dependent instead of being fixed. You might want to look for something like that and try it.

HTH

addendum:
I just saw Ian's comment, and I fully agree.

[Trackzilla]

+1 on a decent noise reduction plug rather than simple expansion, I've had decent results with Voxengo Redunoise for this sort of thing...it's a little trickier than some to tweak, but once tweaked it is pretty impressive.