How to set levels?

[Adrian Tan]

What I've done in the past is record as loud as I can without the sound peaking, the idea being that if you lower the levels in post, you get rid of noise, whereas if you raise the volume you bring up the noise floor.

But it seems, especially with internal omnidirectional mics on recorders, sometimes you have a choice: record at medium gain, but pretty soft; or record at high gain and acceptable levels, at the expense of added noise.

Apologies for the audio ignorance, but, of these two choices, which is better?

I previously opted for the high gain and normal levels, but now am leaning towards medium gain and boost in post...

[Oren Arieli]

Can you give some more details about your recorder, mic and what you are recording? As a sweeping generality, you want to record with the signal with as little post-boosting as possible, but you need to be concerned with distortion and clipping. If it's a lav on a person, can you have them speak in their normal speaking voice (or chat with them while keeping an eye on the level).

If you're using a recorder with very clean preamps, you'll have more leeway to boost the level in post.

[Al Bergstein]

Adrian, There are a lot of discussions about where to set levels in various groups I've seen over the years, but given most situations, I've been told, and follow the advice, of keeping the general level peaking around -10 to -12 dB under 0 (max). It does matter whether you are doing this on a camera, audio recorder or field mixer. But let's review something about when you say, "levels" . Since you didn't specify what you meant by levels.

From a decent B&H quick lesson on the subject, but to the point.

"It's important to understand the distinction between gain and level controls. The gain knob controls the preamp. The preamp is a small amplifier that boosts the audio signal coming into the mixer's input. The level control determines how much of that boosted preamp signal gets to the main outputs of the mixer. You use the gain control to set your signal in the sweet spot, so it's loud enough to be heard in full, but not so loud that it clips and distorts. You use the level control to determine how much of that channel needs to be sent to the output and into the video camera or portable recorder."
The B&H Introductory Guide to Portable Field Mixers | B&H Photo Video Audio

and this from a 2009 discussion here on the audio board.

"The rules I've kind of followed are on prosumer/consumer, gear, try to stay between -20 and -6. MOST of the time, pegging around -12 is good, for an average speaking level, and when they get a little quieter, down at 20, and when the get more excited/loud, around -6. You don't want stuff too low (down in the 40s), but if you hit 0 you're screwed, so I don't like to be anywhere near -2 unless it's an accident. But the idea is if your TARGET is 12, and they spike to 2, it won't last long, and you can turn down your pots a little so they don't get that close again.

For more professional gear, you keep the levels even lower. -20 would be your average, maybe -30 for the lower stuff, try never to go above -10.

The logic behind the lower levels for better gear is, of course, the electronics are better and you get less inherent noise (i.e. generated by the recording device/camera itself) so you can have your levels lower without them being "contaminated".

These are the guidelines I follow, but I'm sure someone here will disagree. It's hard to get straight answer out of pro sound guys when I ask 'em (I am NOT one).

The idea is to leave room for error, especially in live/unrepeatable situations. So if your average/normal speaking level is set for something way lower than 0 (-12, -20), and you get hit with unexpected shouting/laughing/etc., then your volume controls are low enough that those spikes won't hit 0 when they occasionally occur. "

Many newer audio recorders and even some cameras, will allow you to record dual audio mono with one -10 dB below the other for safety. I always use that if I can. And remember, most decent audio editing programs will allow you to boost audio without noise in post. So it's always better being lower than higher. I do that all the time. Too loud is always bad. Clipping is death.

Lastly, Jay Rose, in his superb book, "Producing Great Sound for Film and Video" debunks the notion of working to 0 dB as a standard. It depends on whether the gear is pro, consumer or something in between, and is subjective to your equipment. You should always test your gear and understand how you are using it. He points out that zero levels could arbitrary. Worth reading his book.

Hope this helps.

[Roger Van Duyn]

Hi Adrian,

This thread has useful info:
http://www.dvinfo.net/forum/all-thin...2-setting.html

While I don't have a 302, I do generate tone from the mixer I have. In addition, I have the color bars setting set to generate bars and tone in both my cameras.

I use the tone to set audio levels whether feeding into my Zoom H4N or my A1's.

Generating tone takes a lot of the guesswork out, at least for me. My ears can play tricks on me. I like to see the audio levels when I'm setting them, so need a tone as a calibration standard.

[Adrian Tan]

Gentlemen, thanks for taking the time to reply -- very useful information!

[Paul R Johnson]

With digital dynamic range so wide now, not distorting at the top is the end game - so many people are under recording, which is usually quite recoverable. People start to 'see' waveforms and don't realise that a waveform image that hardly tickles is actually quite loud and usable, so I stopped the quest for maximum before peaking, settling on a nice safe average.

[Michael Silverman]

I agree with Paul in that I've had very good luck raising the volume of a clean audio signal, but I've never had good luck recovering audio that is distorted due to clipping. I'm always very interested in getting the best possible audio, but I've found that as long as the levels don't clip and are not EXTREMELY low then I can get very good results working with it in post.

[Tony Neal]

As Paul and Michael have said above, in the digital domain all you have to do is trim your inputs so that they peak only on the very loudest transient sounds. There is such a huge dynamic range available, especially at 24 bit resolution, that you can boost the level by as much as 18 dB or more in post with no ill effects (assuming your mike are good quality).
So there is no need to 'ride the faders' with digital, you just set and forget. Of course, if your mikes are too far from the source then the room ambience will muddy the sound and no amount of boosting in post is going to fix that.

[Michael Silverman]

I'm also a HUGE fan of the recorders that have dual level recording where one channel is a safety track at a lower volume. I use the Tascam DR-40 to plug into the PA system at the ceremony and the DJ's sound board at the reception. This works really well because during the ceremony the officiant may speak at a fairly soft level but then the music may come in at a much higher volume. At the reception the levels are all over the place too, so this has saved me several times when one track gets totally distorted but I have a safety track that records at a very nice level.