Taming the Wild Voice Recording

I’m installing equipment that captures discussions in a small theater for the purpose of sharing the ideas with other people offset in time, distance, or both. It’s pretty simple. The microphones are Shure MX393 Stage microphones…


…feeding a Peavey PV6 analog mixer and then onto balanced audio tie trunks to another location. When they arrive, they are hum rejected and unbalanced with a Jensen transformer package…


…and then plugged into a Behringer UCA202. The computer is pretty much irrelevant, but UFN, it’s one of our linux machines.

When I did the initial sound tests everyone was shocked at how clear and crisp everyone’s voice was. That’s because nobody had ever heard a correct, uncompressed live recording with no processing or distortion before.

And that’s the problem. Because the system is completely unrestricted, it overloads at the drop of a hat.

No shock there. I’m expecting it.

Most of the time, the clipping isn’t objectionable and doesn’t affect the work. I suspect we are going to find a happy medium between overloading and the eventual customer’s ability to turn the loudness up far enough to hear the work. Low volume is actually much more of a problem.

But I wouldn’t mind gentle limiting somewhere in the chain so to make the process less user intense.

Oh, and you can’t use post processing. The turnaround is less than an hour and the show is an hour long. Once the computer gets the last of the show in real time, it essentially posts. The Operations people grab it in mid-air and produce the international copies.


As an experiment, I put one of the tests through Chris’s Compressor and what it did to the background room noise was pretty entertaining – and not unpleasant, but wholly unworkable for the actual performance.


A good case hardware compressor. I use a Behringer composer pro (relatively inexpensive and does the job nicely). Set the compressor threshold at the top of the “normal” voice level, then those voices will be essentially unprocessed. Set the ratio pretty high, but with a fairly soft knee so that you get increasing compression as the level becomes more dangerous. Add the limiter to the very top to prevent over 0 dB. The limiter on this device uses (fast) lookahead so that it can brick wall at the set level without clipping.

The compressor would typically be connected to the channel inserts of the mixer.

This is me writing that down.

There’s also the 4600 http://www.behringer.com/EN/Products/MDX4600.aspx if you want 4 channels in one rack space and don’t need the de-essing feature. But it looks like the PV6 doesn’t have channel inserts, so if you’re OK with just limiting the final mix then the 2600 would do the trick.

– Bill

There is plenty of headroom at the mixer. Of the four level lights -21, 0, +6, and CLIP, we never get past flickering the -21 light. The first place where level is a very serious problem is at the UCA-202.


The 4600 should be fine, though the Composer Pro offers a bit more flexibility (all the bells and whistles that could possibly want).
A good mixer should have plenty of overhead - some will happily cope with 24 dB above 0, but the UCA 202 certainly won’t. The UCA 202 is 16 bit, so 0 dB is an absolute limit.

Does the PV6 have “master inserts”? If it does then the compressor can be added into the signal chain at that point, which has the advantage that the main mix metering on the desk will indicate what is being sent out to the UCA 202.

Yes, it has effects send and effects return. However, the room where the mixer lives is short on operator space. There is a tiny table next to the operator’s knees big enough to place the mixer with the headphones underneath. No racks and no space for rack mount equipment.

Maybe I could wrap it in bubble wrap and put it under the floor.


That’s different from “master inserts”.

With an effect send/return loop:

the signal is “tapped” at some point in the signal chain (usually straight after the channel fader, though sometimes immediately before the channel fader). This signal is mixed with the signals sent to the fx out from other input channels. This fx “sub-mix” then goes to the effect processor(s) which will normally be set to produce a “100% wet” output (all effect and no dry signal). The output from the effect processor is the sent back to the mixer via the fx return input. The processed signal is then mixed into the unprocessed main mix.

The output is a mix of the “dry” signal that has passed through the normal route in the mixer, plus the processed signal from the Fx loop.

With master inserts:

The entire mix is sent out of the mixer to the effect processor(s) and then back to the mixer. The wet/dry mix is controlled by the effect processor(s).

Using an effect (send/return) loop will not help with taming the wild voice as the dry signal bypasses the effect processor.

Is the table 19" wide? You could put the mixer on top of the Behringer compressor on the table. Then you’d be able to see the meters on the Behringer. It has balanced outputs so you should be able to drive your balanced lines from it. Not ideal (master inserts would be better, individual input inserts would be ideal) but workable.

– Bill

This room was aggressively designed not to expand. The tiny table is slightly wider than the Peavey. That’s what, eight inches? It bumps right up against the operator’s knees.

Swell! :-/ How about the half-rack-width Alesis Nano Compressor? I’ve got one, and it’s pretty darned good, especially when size is a consideration.

I left with the impression that the whole process needs to be crammed into a cigarette-pack size room. That’s only at the microphone end of the system. At the other end of the process, after the stereo transmission line is a projection booth with two 6-foot tall racks and I can do almost anything I want in those.