Understanding Sound level numbers -60db to 10db

rfrink

rfrink

Member
Hi Everyone. I'm in the earliest stages of knowledge and experience with sound.

I notice that my work often has inconsistent sound levels. For example...a scene that should be very loud might be just a little loud...and a scene that should be only just a whisper...seems to boom through the speakers.

Other things that I've noticed is that my voice overs also have inconsistent levels. When cpmpared side by side, I can really notice the clips that were recorded at different sessions. Not only is the cadence different..but the level as well.

So, just recently, I've really been trying to pay attention to the sound level meters...on the Camcorders and in my Post software.

How can I put the level numbers in perspective. They go from say -60 to +10.

After all what is a minus (-) sound level? Negative sound? anti sound...? I know it's just a notation...but I never paid attention to it untill recently.

At 0db..the meters all start turning red. Is this when the levels enter an increasing region of distortion and degrading quality due to excess power?

What are some typical sound level guidelines for:

A normal conversation from 6 ft away.

A jet engine or explosion up close.

A soft whisper in a lover's ear

ambiant noise from a theater

A voice over/narration "news reporter" context.

Hopefully, this thread will give me a few tools to help control and understand apropriate sound levels for a variety of situations.

And...techniques as well. Do I try to capture the sound with best quality at a mid level...then edit it in post...or try to capture it at a level apropriate for the footage/story/scene.

Whattaya think?

Sincerely, Rob
 
dB is a measure of a ratio, comparing two things. It's an exponential relationship so that a signal that is twice as loud as another is 3dB louder while one that is half as loud is 3dB softer. Because it is a comparison the numbers don't mean anything by themselves ... it's only when you reference them against a standard that they become meaningful. There are a number of different standards around - for example, you'll hear about 'line level' as being +4dBu (note the 'u'). 0dBu is standardized as an average signal level of 0.775volts. So a signal that is a positive dBu number is stronger that 0.775 volts and a signal that is a minus dBu number is weaker, some fraction of 0.775 volts.

See http://en.wikipedia.org/wiki/Decibel

As for the meter scales, there are a number in use and we don't have enough info about what specific gear you have to interpret what you're looking at. For example, a digital meter on a digital camera would have a meter that sets 0dB at full scale, the maximum level it can go without clipping. Normal sounds would be recorded with an average level somewhere between -12dBFS and -20dBFS to give you a good volume without danger of hitting the max on sudden peaks. Analog cameras might have an older VU style meter that sets 0VU somehere less than full scale and in their case 0 is the normal operating level and peaks can briefly stray into the red zone as long as they don't go too far or stay too long. Some devices give you a choice or a combination. That's where you've got to read the manual to know for sure what you have.

As for recording dialog on the set, you want a good recording level so the signal is well above the circuit noise but you don't want it so hot that it distorts or clips. Try for consistencey. It's a judgement call. You can adjust levels in post for dramatic effect.

Your voice over issues are a matter of learning and practicing the craft of voice acting. You need to practice using a microphone so that you're constantly aware of the working distance and your voice pitch and intensity. Take a look at on-screen reporters and performers like comics who are working a hand held mic - they've nailed the technique of keeping the mic within a fraction of an inch of a precise contant position with respect to their mouth no matter how they move the head or body. That's a matter of practice, practice, practice.
 
Last edited:
Lots of misinformation above, but I won't get into most of it...

Basically level metering is there to help you get a clean, usable signal recorded, nothing more. The actual level you'll need to provide the desired dynamics in your audio mix will be judged subjectively. If the gun shot isn't loud enough, grab a fader and *turn it up*. If the whispered voice is too loud, *turn it down*. Meters will come into play here in the same way as above, but much more importantly on the upper end of the scale - you can't turn that gunshot above a certain level or it will clip, and the meter will tell you that if your ears don't (which is not uncommon with light clipping). So basically you're creating a "top down" gain structure, with the loudest sounds being safely under clipping, and everything else being mixed to subjectively sound correct relative to that.

Regarding compression and "all the peaks (from whispers to gunshots) are relatively similar", this is extremely misleading at best. Even recorded alone, most 'whispered' audio is typically recorded well below the level that a gunshot would be. It'll be ok like that (as long as it's not *too* low) because it's not going to be very loud in the mix. And no, "most things" are not compressed, particularly in movies. Pop records in many cases yes, most assuredly (but not all), but movies are much more typified by the reverse - an extreme range of dynamics. Many's the time I've had to "ride gain" on my home theatre system to watch movies late at night in my apartment. A level appropriate for intimate dialog produces eviction-notice levels of volume in the explosions and gunfight scenes. Leave it where that's ok and the soft dialog is inaudible. This is done for dramatic effect, with the film being mixed for theatrical purposes. They'll have the level loud enough that the intimate dialog will be fine, and when the car chase happens it's gonna be LOUD. But it's a theatre - it's ok to do that there.

But the simple answer to the OP's question is that it's basically about two things: use meters to get proper initial levels to the recorder, but use *your ears* (along with the meters, mainly for peak levels) to get the mix right for the film.
 
Ditto Ted Spencer above.

I'll add that meters are important in setting up your mix levels, at least once in a while.. and you need a few types of meters to do it.

As a rough start: get an SPL (Sound Pressure Level) meter from Radio Shack. Put it at your mixing position, and play some pink noise through your speakers. Set the output level of your NLE/DAW software so that the pink noise level is around -12dB (dBFS). Then, adjust the monitor level of your speakers so that this signal reads around 82dB (dBSPL) on the SPL meter. Your monitor level should always remain at this setting so that your mix(es) will be consistent. This approximates levels in a professional mix theater.

This is your "standard level." From here, use your ears to make everything sounds right. Use the output meters in your DAW/NLE just to make sure you have no peaks. When you think you're done editing dialogue, set your timeline so you can see your whole project, and jump around between different scenes rapidly to make sure that all the levels seem consistent and appropriate.

Do this with the final mix, as well.

------

For recording, the ideal is to have the recorded signal as hot as possible without clipping - this gives maximum quality. However, clipping is so bad in digital audio that its very important to be conservative. If you have a chance for a good sound check, try to listen to the loudest sound you expect to be recording in a given shot/scene, and set your levels so that you have a few dB of headroom over that.

----

From a technical point of view:

The decibel scale has nothing to do with sound. It expresses the ratio between two numbers. All decibel scales have a specific reference.... which is why you need to pay attention to the letters right AFTER "dB" - as in dBu, dBv, dBFS, dBSPL, etc. They're all very different things.

Most importantly, you need to know where the different scales are pegged, and where they are limited.

Sound pressure levels (SPL) are relative to a rough approximation of the quietest sound a human can perceive. Thus, typical sound pressure levels are POSITIVE. 0dB-SPL is the quietest thing you can hear; ~60dB-SPLis normal conversation, ~130dB-SPL will cause hearing damage, etc. These are the numbers used in the typical "jet plane is x dB, pin drop is y dB" charts.

Analog audio has a reference level that indicates where distortion (fairly pleasant distortion, compared to digital) begins; this is 0 dBu. Since our job as audio engineers is (generally!) to avoid distortion, audio meters are set up relative to this level of 0 dBu, and geared towards keeping us below it. So, most numbers are NEGATIVE, meaning "smaller than." Average levels for dialogue tend to be either -12 or -20 dBu; average levels for pop music, higher. In analog, going over 0dBu occasionally is OK, and is often used as an effect with analog gear - it causes a compression-like effect, e.g., on analog tape.

Digital audio is measured in dBFS - decibels relative to Full Scale. Full scale is the absolute loudest amplitude that can be recorded. There's nothing over 0dBFS.

For A/D conversion, you generally set an analog gain level so that the maximum expected sound is converted to about 0dBFS (or, better yet, leave a little leeway!). If some unexpected sound in the analog domain goes above this level, the digital representation of the waveform is distorted into a squarish-wave (actually, more of a flat-top sine wave) by being limited to 0dBFS. Squarish waves sound very ugly. Even mixing in the digital domain, it's possible to do the same thing - combine to nice, round waveforms, and if any part of the result would be higher than 0dBFS, it gets chopped off to an ugly square wave.

So, for digital systems, it's essential to NEVER try to go above 0dBFS. (Every rule has its exceptions, and for this, it's short, percussive sounds - VERY brief peaks like the initial attack of a drum hit sometimes don't really sound distorted).

-Barry
 
Last edited:
Here are a few tips to get you on-track, Rfink:

Always (alwaysalwaysalways) monitor your recording with a good pair of professional closed-back headphones. The industry standard are the Sony MDR-7506. Earbuds, Bose noise cancelling `phone, or other consumer `phones simply will not cut it. You need monitoring headphones.

Once you set your headphone levels at the start of the day (eg. using 1kHz tone at -20 dB), leave the knob alone. Occasionally, you may wish to play some more tone to remind your ears where that objective reference level is.

VU and Peak meters are useful diagnostic tools, but they do not give you the whole picture; for that you'll nead headphones. Relying on a meter to judge the quality of your audio is like relying on a vectorscope to judge a shot (in this analogy, headphones are like video viewfinders).

You will have the opportunity to boost or cut levels during post-production, however: record too quiet and you'll get mired in noise and 8-bit samples, record too loud and you'll run out of 1's and 0's (causing nasty square-wave distortion).

You'll be in good shape level-wise if your dialogue sits between -20dB and -12dB, with the tall peaks not exceeding -6dB. However, you need to practice until you can achieve this consistantly without even glancing at the meters. The meters are a measurement tool, they're not an "everything is OK" indicator.
 
...but don't use headphones for post-mixing, except as a cross-check, or even better, for noise-hunting. Use a good quality pair of speakers (or set, for multichannel) that, most importantly, you *know the sound of* intimately. Like, you know what every one of your favorite CDs and/or DVD movies in your collection sounds like on...

Headphones have a tendency to mask mixing errors - they usually make things sound better than on speakers, which is bad since most movie/TV viewers will be listening on speakers only.
 
I would add that 82dB is a strange choice. 85dB is the mix stage standard and near field mixing is usually done at around 80dB (smaller space, closer speakers). Also you need to use properly filtered pink noise or your numbers will be off. Blue Sky and Dolby have clips of calibrated pink noise that you can download. A lot of built in pink noise generators don't create noise you can calibrate with.

I will second Ted somewhat about compression. Compression is common in film work but not in the way ProfessorU is talking about. Dialog often gets some light compression and IF your going to an optical track on 35mm the Dolby processor does some compression of the overall mix. But film work uses very little compression. POP music as a rule (these days) uses a LOT of compression. Broadcast TV uses very heavy brick wall limiters and if you send them a hot mix it will get heavily compressed because of that, though more likely it will just get rejected. TV commercials are as a rule pretty heavily compressed.

RE: meters

For broadcast you need them because you are to some extent mixing to them. Broadcasters have rather strict delivery specs and you have to hit the numbers or your mix will get rejected, and you get to do it again for free :~). But films are mixed by ear in a calibrated space. Which is what the talk about pink noise and SPL meters is all about.
Read this thread

the short version is you set your speakers to a standard level and lock them. Then you mix to make it sound the way you want. It's tricky to do if your working out of your apartment, but a semi standard way of working (in less than ideal locations) is to work at a lower level and switch back to your standard level to check. At some point though you will have to watch the whole thing at a calibrated level to make sure it all works the way it should.
 
Noiz2 said:
I would add that 82dB is a strange choice. 85dB is the mix stage standard and near field mixing is usually done at around 80dB (smaller space, closer speakers).
Click to expand...

82 is probably a compromise used by various people... in any case, I certainly defer to NoiZ on any film mixing standards... my experience is mostly in the music field (and classical music, at that!).

I'd also add - always listen in headphones for edit smoothness/glitches! Some things magically bypass your speakers in the studio and stand out as glaringly obvious in the theater - but headphones emphasize these things more than anything.

You get the idea... you're listening to everything a few times, on different playback setups, to get as many perspectives as possible!
 
ullanta said:
I'd also add - always listen in headphones for edit smoothness/glitches! Some things magically bypass your speakers in the studio and stand out as glaringly obvious in the theater - but headphones emphasize these things more than anything.

You get the idea... you're listening to everything a few times, on different playback setups, to get as many perspectives as possible!
Click to expand...

Correct on all counts...
 
Wow! This is a ton of info. Way more than I currently have the experience or knowledge to understand...But after reading through it several times, I've picked out a few "milestones" that I can achieve at my current level.

Sincere thanks!!.. to all that spent the time to post. I greatly appreciate your help.

-Rob
 
You must log in or register to reply here.
Back
Top