General advice on monitoring sound level

[A. J. deLange]

Originally Posted by Ty Ford

As I have mentioned many times in the past. I have yet to hear a camera AGC circuit that does more good than harm. An AGC is not a replacement for a good limiter.
Ty Ford

I'm not myself a big fan of AGC. In fact I make my living running around yelling "Turn off the @#*&^ AGC and set the gain so the noise floor is at the LSB" and while the signals are not audio the principals still apply. OTOH I would never ever use a limiter because distortion (and this may be colored by my professional experience which is far more extensive than my audio experienc) is totally unacceptable to me. The need for either AGC or limiting is indicative of poor gain management. All that said if I'm out in the field by myself I can't manage zoom, focus, shutter speed, color balance and audio gain so I let the AGC handle the latter. Even if I wanted to manually ride the gain I can't reach the controls while the camera is on my shoulder. The results (XL series cameras) have always been excellent but then I've never tried to shoot a rock concert.

This could turn into a AGC vs limiter *&%%ing match which might as well be a MAC/PC discussion for the value it would add. It's important to recognize that taste and experience lead us to the opinions we hold, to remember that "de gustibus non disputandem est", and to encourage Alex to try any or all of the suggestions being made here to see what works for him.

[Michael Fossenkemper]

hahaha, well the mac/pc debate I think has ended. Mac is the winner being that they can run both operating systems now.

[David Ennis]

Correct me if I'm wrong, but I don't believe that limiters clip audio. Their transfer functions are not applied to instantaneous voltage, but rather to a sample of signal.

You can confirm this with any good audio editing program. Produce a sinewave tone at about -3 dB then put a curved envelope on it so that it tapers to zero at the beginning and end. Now apply limiting to clamp the output at - 10 dB. You'll find no distortion.

As with any active audio device, the signal can be clipped at the input of a limiter if its amplitude is out of range.

[Michael Fossenkemper]

depends on the design. Analog limiters can and will clip. digital look ahead limiters can react faster and while not technically clipping the result is similar. They try to round off of the tops to a certain degree depending on how hard you hit them. If it's gentle, then they will be more rounded, if it's hitting hard, they will be more flat topped.

[David Ennis]

I'm afraid I'm not convinced. Why would anyone design an audio limiter to clip? That's pretty nasty, since we buy them to prevent clipping. What would a clipping limiter be for?

[Dave Largent]

Originally Posted by Fred Retread

I'm afraid I'm not convinced. Why would anyone design an audio limiter to clip? That's pretty nasty, since we buy them to prevent clipping. What would a clipping limiter be for?

Not sure who you are responding to, here, but I'm
in agreement with you. Limiters are designed
to prevent clipping. Even soft limiters most times
prevent it. If your limiter is clipping you should adjust
it, if possible, so that it doesn't clip.

[Ty Ford]

right.

Ty Ford

[Bob Grant]

Just a few simple thoughts.
Firstly try to get the mic closer to the sound source you want and further from what you don't want.
If the levels of the general program is set to peak to say -6dB FS then the limiter if used should be mostly hitting the applause and even a bit of harm done to that isn't going to matter.
To record another track at lower level get a XLR splitter cable. Feed one output directly into one channel and the other to the other channel via a 10dB pad, these are available from most pro audio houses. That's assuming you don't have individual gain controlls on each channel, if you do just leave the pad out and set levels so one channel is good for most of all but the loudest program and set the other 10dB lower. Select best channel in post as needed.

[Michael Fossenkemper]

Nope, limiters aren't designed to prevent clipping. They are designed to restrain the level of a signal. They were designed for media that also was analog and therefore a tad more forgiving. This of coarse can be used to try to prevent A/D clipping. But with any analog design, there is a reaction time. This reaction time works well for radio, vinyl, tape, but not so well for Digital. I won't get into what happens on the D/A, but you can get as much as 6db of clipping if you pin your meters and have no overs converting it back from digital to analog. If you convert it to any compressed format like mp3, mp4 etc... then there is even more clipping occuring. Digital look ahead limiters can prevent clipping in the digital domain, but that doesn't really help with capturing analog and recording it to digital. And only then, there are only a few digital limiters that can guestimate overshooting and compensate, But we're talking in the $10,000 range and they of coarse loose sync do to the fact that they have to delay the signal in order to react quicker.

This is why compressors are used before the limiter. The compressor controlls the signal before it reaches the limiter. not only does it sound more natural, but does a much better job on rms levels (sign wave or vocals) where limiters do a better job on peak levels (spike like a snare or gunshot).
The very nature of using these non linear process's is distortion. The waveform is not the same, the definition of distortion. It's a tradeoff of one distortion for another. in the case of some camcorders built in limiters, the tradeoff doesn't seem worth it.

So in order to record a very dynamic signal that involves varying rms levels as well as dynamic peaks to a medium that is finite like a digital camcorder or any digital format, you have to make sure you don't push it to the limit. just because something says limiter, doesn't mean it's a free lunch and it doesn't mean you aren't clipping the signal or that the signal won't clip down the line.

[David Ennis]

Originally Posted by Michael Fossenkemper

Nope, limiters aren't designed to prevent clipping. They are designed to restrain the level of a signal...

Strictly speaking you could say that's true, and there are other applications for limiting output level. For videographers though, the reason is almost invariably to prevent clipping, and the limiters marketed to us for that purpose do not themselves clip, flatten, fold, bend, staple or mutilate audio signals to any significant degree, including pure sine waves, when operated within their design range. That would be absurd. That's the door I came in.

[Douglas Spotted Eagle]

Originally Posted by Fred Retread

Strictly speaking you could say that's true, and there are other applications for limiting output level. For videographers though, the reason is almost invariably to prevent clipping, and the limiters marketed to us for that purpose do not themselves clip, flatten, fold, bend, staple or mutilate audio signals to any significant degree, including pure sine waves, when operated within their design range. That would be absurd. That's the door I came in.

Limiters marketed to any market do indeed fold, spindle, mutilate, bend, whatever, because that's exactly what they're designed to do. They're not designed to prevent clipping, they're designed to hold a specific level at all times. There is no "videographer's limiter" and a "studio engineer's limiter" etc. They're all the same, just used differently in various contexts. I'd never use a really hard limit when recording a bit of dialog, I'd prefer to use softer compression starting earlier, because I don't like recording hard limits unless that's really, really, called for. If you record this way, you're married to it for ever. If you've got a dialogic that calls for extreme measures, then you'll want a compressor prior to the limiter so it's not quite so harsh. BTW, a limiter is just a compressor with more strict behavior. It's like a parent to the compressor's child.
Maybe you want a hard limit at 0dB, and you're feeding it +9 in the analog world. Perhaps you want it to limit at -6dBfs, and you're feeding it -2dBfs. You're going to get two extremely different results and audio qualities from both of those scenarios. Limiters have nothing to do with distortion overall, and they'll distort the signal by their very existence. You're associating "distortion"="bad" when in truth, distortion is anything that changes the paramaters of the audio from its orginal dynamic form. Color correction "distorts" the original image the eye sees, but it likely (hopefully) is a pleasing distortion.
If you're feeding the limiter a very dynamic signal, it may prevent truncation of the signal, but the waveform is still flattened out much like a truncated waveform due to the ceiling placed by the limiter. That will present itself audibly, as "distortion" because of the way your ear will respond. But it's not truncated. It can actually be somewhat repaired in post to make it more ear-palatable, where truncated audio really can't be.
Either way, it seems the thread is pretty far off track at this point.

[David Ennis]

Originally Posted by Douglas Spotted Eagle

...it seems the thread is pretty far off track at this point.

You're right, Douglas. This was not the initial poster's subject. I see that it turned this way from the third post and pretty much stayed there. Sorry I didn't notice. Your reply to my last post was kind of strong, so please just permit me to say that I know that compression, limiting, equalization, etc., constitute amplitude distortion. And I trust that if I had left out the firgurative excesses and simply said that units like the DXA-8 and MixPre, which tout their limiting functions in their advertising only in terms of preventing clipping, do not themselves clip audio when operated within range, you wouldn't disagree.

[A. J. deLange]

Watching the posts fly by I thought back to my generic advice on controlling dynamic range which I posted a day or 2 ago: "Turn off the @#*&^ AGC and set the gain so the noise floor is at the LSB" and wondered how it might apply to this industry. So I looked at the specs for a Shure SM81 which I suppose is a typical mic. It's noise floor is 16 or 19 dBSPL (depending on the weighting) and the maximum it can handle is +136 dBSPL. Using the lower noise floor that's 120 dB difference which is 20 bits worth. Given a complex (other than single tone) load an A/D needs 2 1/6 bits head room and 1 bit for the sign bringing us up to a total of 23 1/6 bits. Thus a 24 bit A/D (i.e. one that really delivers 24 bit performance) can handle anything this mic can do to it with out clipping and with almost a whole bit to spare. Now I would never go into a sound field at +136 dBSPL (that's 6 - 16 dB above the threshold of pain) but I understand it does get that bad at rock concerts. A 16 bit A/D will have 48 dB less dynamic range and would, set up this way, start to clip at 27 - 37 (don't forget that last 5/6 bit) dB below the threshold of pain.

For 16 bits I guess I'd set the LSB at the background room noise level and guessing that this is a good 10 -20 dB above the threshold of hearing pick up 10 to 20 dB out of the potentially problematic lost 27 - 37 that way. Clearly a second mic with 20 dB less gain would cover a good part of what's left over.