If recording a filmed dialogue scene that has a sound level range between whispering and shouting (or similar high range beyond acceptable limits for the equipment available), what is the right way to get good sound, particularly in a resource-constrained shoot?

Do you try to ride the levels according to the lines after perhaps memorizing them?

Do you ask to break the dialogue up into sections of different loudness?

Do you have the boom operator pull the microphone away at the right moments?

Multiple microphones, multiple recorders or multiple tracks set to different levels?

Or do you do something else?

4 Answers 4


Personally, I don't ride the faders too much between single words. Only if 'longer' passages are quieter/louder I turn up/down the gain. Moving the faders within words should be avoided.

Until now, I've never asked the boom operator to pull away the microphone as it colors the sound.
Since I'm recording at 24 bit, I can leave quite some headroom for unexpected things. However, you should already know the scene, especially if the actors do some practice before actually shooting. As scenes are mostly broken up into different shots, the dialogue is often split into sections by itself.

Also keep in mind, that in very quiet passages, turning up the gain to its maximum might be counterproductive as preamps have a point where their self noise rises non-linearly.

At last, I still have limiters for safety. Here it is important that you can 'trust' your limiters. Sound Devices are nearly 'unclipable' while other limiters often behave poorly.

However, production sound isn't really my expertise.


In this situation, two of your proposals are commonly used in conjunction :

  • ride the levels appropriately
  • have the boom operator pull the microphone at right moments.

Hopefully the boom operator knows what they are doing and are trying to capture the best sound and levels, letting the director know when there are noises that will spoil the shot, and coordinating on shooting locations and blocking challenges with respect to the sounds.

Along with that, there should be a dialog between the location sound person/people/team and the post-production sound person/people/team, such that the location team is aware of what the post team considers to be ideal audio. If the post team wants no major distance changes between lines, then the boom operator should not pull the mic back, but should have the levels on the preamp set to avoid clipping for the loudest sounds. In this situation, the post team A) has a good reason for wanting consistent distance (e.g., consistent background noise level) and B) has a method they plan to use to make it sound right. If the post team wants the most even level or the hottest levels possible for all dialog, then the boom op/location team need to be aware of that and work accordingly.

Note that if you have a two-person location sound team (boom op and recordist), it might be much better to have the recordist (or a very skilled and strong-armed boom op on a one-person team) actually ride the preamp gain a little during recording, rather than moving the boom around too much. When I've been on a two-person team, there was definitely at least a little gain riding for a lot of shots.

If I were doing post, I would want the levels to vary, but not too much. I want more level when people are shouting than when they are whispering, since that will lend the most verisimilitude to the audio. At the same time, in real life people don't suddenly move four feet into the air when someone starts shouting, so major boom moves might sound very fake or unusual. I'd want a compromise between keeping it level with the boom position versus not moving the boom at all. It's good when a post person/team has run location sound and vice/versa so they understand each other's challenges.

During post, I would focus on one of two major ways to deal with varying levels. First off, everything will get some compression on it. Even if I'm trying to deliver a wide dynamic range, I can't think of a time when I've not wanted to compress it at least gently to help smooth things out. After that, it depends on the situation and the shot. If it's two characters and one is shouting and the other is whispering (or something similar), then I'll probably separate the location sound into two tracks, one for each character. This also depends on the video editing, since if the shot is cut back and forth between each character's point of view, I might want to actually set up one track for character A's POV and another for character B's POV. It's an art, not a science.

If separating the audio doesn't make sense or isn't really possible, then my primary way of dealing with varying level is to automate fader moves. Too much compression is not appropriate for sound for video, where the crest factors are kept much higher and viewers have learned to expect that. Even in music engineering, there's no substitute for fader automation on vocal tracks. It's something about the human voice, I guess.

Regarding the recording and engineering of shouting in general, I think right away about a scene from Lawrence of Arabia. Lawrence and his Arab "army" have just crested a rise and have spotted a line of Turkish soldiers falling back after massacring some Arabs. Lawrence was horrified at the massacre and has a chance to exact retribution on the tired Turks, who are unaware the Arabs have come up behind them. He sits up in his stirrups, raises his sword, and screams "No prisoners! No prisoners!". It's a chilling scene, and on the third or so viewing I realized that the audio for the scream of "No prisoners!" is fairly well overdriven and distorted. I feel like that adds to the intensity of the moment, even though it's subconcious for most viewers who are naive about sound production. Of course the distortion created by overdriving the equipment used at that time had a warm and more subtle quality that wouldn't automatically happen with today's tools. But if you have a very hot level and compressing it just makes it sound... compressed, maybe clipping it in a subtle way can maintain the intensity while also taming the level. It's still an art.

  • Thank you very much for your detailed answer, Todd. I marked another answer as the solution because it applies more directly to my situation and experience level, but your answer provides a lot of other things to think about as well.
    – optic
    Oct 22, 2015 at 23:53

I am primarily mixing music - but I've had the same type of problem with vocals with extreme variation (scream vs. whispering).

I'm doing three things to get around it:

  1. Parallel Compression/Limiting (also called "new yorker comp" or "N.Y. style comp"): Duplicate the track (or create an aux send prefader copy) and compress/limit that track/channel quite hard (slow release, fast attack) and bring up that compressed

  2. Automation (volume curves): with the wave curve of the recording visible as aid, it is rather easy to manually compensate where needed. The problem with this method is if the variations happen too quickly.

  3. Compensation during recording: With a little guidance most vocalists can help reduce the problem by controlling distance and how loud the sing. I guess your equivalent here is the boom operator. But again if variations happen too quickly this may not work (and actually do more damage than good).

  • I think he is speaking of recording on set. Personally, I would think NY style compressions is harsh specially with film dialogue. Oct 22, 2015 at 13:06
  • @LukeFarroh - yeah, you're probably right about the "on set" angle here. Regarding the parallel comp - I don't think you can really generalize it as being too harsh a method: just lower the blend level (the level of the compressed track/send) and/the threshold if you find what you hear too harsh. But in general I wouldn't apply dynamics during recording - it should of course be a realtime non-destructive effect or applied in post production. Oct 22, 2015 at 17:44
  • Thank you for your answer, Michael. Luke is right, this is for on set recording of a film scene. Thank you both.
    – optic
    Oct 22, 2015 at 23:45
  • @MichaelHansenBuur I see your point but wouldn't you still get the pumping and breathing effect even with a lower blending level? It seems it would sound a little bit un-natural. Although, I never tried or experimented with this so, I will have to try it out! :D Oct 24, 2015 at 12:51
  • @LukeFarroh - It can definitely be set to gentle intervention, 2:1 - 4:1 ratio, Release > 300 ms (attack just as fast as possible). Threshold somewhere around the low level parts. Blend that with the original to your satisfaction. Cheers! Oct 27, 2015 at 9:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.