Recording and mixing screams and whispers.

[j1clark@ucsd.edu]

I'm really confused now. When RECORDING, there is no reason to not record whispers loud, peaking at the same levels as when you record the actors shouting - you are attempting to capture maximum dynamic range not reality! A recording using the full available dynamic range without ever hitting the full scale digital maximum is as good as it can be, quality wise. Common sense means that it's usually a bit less - but you are not attempting to balance out in the field, just capture. Back in the studio or in the edit, you balance the shout against the whisper - and dropping level is so much better than trying to raise it when the levels were poorly controlled when it was recording. Many people under record, level wise and spend time normalising in the edit - but sometimes this really brings up the noise floor!

Here's where I think the use of 'noise floor' is misleading, given that it appears that audio people use different terms for 'noise floor'...

So perhaps 'room tone' is a better word to describe the 'nose', from which any 'signal' is measured, at least in the capture of production sound. The contribution to this maximum level of 'noise' by 'electronics' is down in the -90 or -100 dB level, way below 'room tone', and only 'hearable' by someone wearing headphones in an audiologist's sound booth.

"Room tone" seems to also be divided up into 'things that are not distinctly heard'... so the traffic from a street via a window is a 'room tone contributor', but the 'chain saw massacre' in the next room is 'ambiance'...

In any case it is clear that 'adding gain' will raise the level of all noise contributors, whatever the source, along with the desired 'audio signal', the dialog.

However, for some set of 'noise' contributors, filters can attenuate the 'noise', leaving the desired signal less attenuated and thereby 'stronger' relative to the now filtered out 'noise', effectively raising the SNR.

 

[Steve House]

The OP used the words "noise" and "hiss" interchangeably, talking about hiss increasing when he advanced the gain to record whispers. That doesn't sound like the environmental sounds or room tone to me.

 

[paulears]

Room sound vs wanted sound is simply wrong mic, or right mic in wrong place - and is only a problem when you've got the selection of location wrong. I've never heard anyone refer to 'room sound' as noise, because it's not. It could be unwanted of course, especially if the mic choice and placement are wrong, but it's not the kind of noise I mean. On the other hand pre-amp noise does not need an audiology set-up to detect. A cheap in ear pair of headphones plugged into the camera is easily able to determine the audio is noisy (read as hissy for accuracy). I'd argue that room tone/room sound is more difficult to hear. The critical point is that we're talking about selecting an input to output route that introduces as little noise and distortion as possible. If you are recording too much room sound, you're doing it wrong!

 

[ironpony]

Okay I did some more scream tests and got myself some better headphones to play it backthrough. And yes it sounds better to have the gain turned up very high, and the fader, turned down lower, rather than any other way. If this is the case, why even have a gain knob? If it is to be kept at 75% all the time, then why not just invent the sound recorder to record at 75% gain all the time? 75% seems to be the perfect number since there is it's the least amount of hissing, so is their any other reason to turn it up or down?

 

[Jema]

I believe traditionally there was a lot more "mixing" in production. I'm not sure if new mixers are still doing a lot of level riding on the set. With analog it was an important tool to reduce hiss and noise but with digital recorders and clean preamps I'm not sure it's much of an issue today.

This depends a lot on the type of produktion, but from what I've picked up it's very much still the praxis in bigger motion pictures.

Okay I did some more scream tests and got myself some better headphones to play it backthrough. And yes it sounds better to have the gain turned up very high, and the fader, turned down lower, rather than any other way. If this is the case, why even have a gain knob? If it is to be kept at 75% all the time, then why not just invent the sound recorder to record at 75% gain all the time? 75% seems to be the perfect number since there is it's the least amount of hissing, so is their any other reason to turn it up or down?

This result sounds very strange to me. Screaming into a mic not several meters away with that much gain must introduce heavy distortion - the fader on the FR2-le is digital, and as such does not prevent preamp overload. Does the peak warning light go on? Are you using a mixer with a limiter?

Also, I feel like pointing out an important fact about preamps. Several times you've said that you get audible hiss when the gain goes over a certain amount. That is simply because the whole signal is louder, not necessarily because the preamp gets noisier. Most preamps have something of a jump in gain the last bit, which creates this "illusion". Generally, preamps are actually quieter when giving max gain. Record something twice with different gain, put it in a workstation and bring the clips to the same percieved level then see which is noisier.

 

[ironpony]

Well there is no clipping in the scream, with a high gain, as long as I keep the fader way down. Is that not the way to do it? And by noise I mean two things, hiss and also this other noise when it gets too loud. It's not clipping but it just sounds too loud, like something else... not sure what.

 

[Steve House]

Well there is no clipping in the scream, with a high gain, as long as I keep the fader way down. Is that not the way to do it? And by noise I mean two things, hiss and also this other noise when it gets too loud. It's not clipping but it just sounds too loud, like something else... not sure what.

Hiss heard in the headphones but not recorded in the file is irrelevant, it's inconvenient for monitoring but the only hiss that really matters is hiss that is recorded along with the desired signal.

Not sure which "gain" control you're talking about here but I'll assume by that term you mean the microphone trim controls.

The monitor level is irrelevant to the recording, it's purely to adjust the headphones for your listening comfort and has no affect on the recorded signal.

Different mics produce different output levels in response to sounds of the same loudness - the purpose of the trim control is to allow you to use all of them with the recorder, adding more gain when using mics with low outputs and lowering the gain if you're using mics with high outputs. That's why they don't just take away the knob and lock the gain at 75% at the factory - 75% might be good for mic A but not good for mic B.

The big recording level controls are the faders. The faders should start out set at mid-scale while you adjust the trim. Set the trim so your normal voice level causes the -12 LED to light constantly and the -6 LED to flicker rarely. As mentioned above, with mic X it might end up at 50%, mic Y at 75%, and mic Z at 30%. For example, an EV RE50 only puts out 2mv/Pa so with it you'd need to turn the gain all the way up (and even that might not be enough without an external mixer/preamp). An Oktava is about 10mv/Pa so you might run it at 60%. A Senn MKH-416 puts out 25mv/Pa so you'd might find it needs <50%. A Rode NTG-3 puts out almost 32mv/Pa so it would need even less gain than the Senn. As you record, if the voice gets louder and drives the signal up too high, turn the faders down a bit. Likewise when it gets softer and you're worried about it getting lost, turn them up a bit.

 

[paulears]

There's a simplicity to our explanations when boiled down. If the microphone is assumed to not be producing distortion itself - as condensers often can when subjected to high SPLs - then to get maximum dynamic range, the pre-amp needs to be close to maximum level before distortion when presented with the mic's output. Then, the mic pre-amp self-noise will be the best it can be. In a traditional mixer, this would be done by simply pfling the channel and adjusting it with an example of the loudest source - which really could be the presenter saying one, two. The amount of this preamp output signal is then sent elsewhere via the channel fader - which is how the almost full scale whisper is brought back to the quieter level it needs to be. In a portable recorder, it's this output that is recorded, and if you just wish to record, there's no reason why a second fader would even be needed - apart from the historical requirement to be able to fade things in or out. In reality, if we are recording just a single channel, then there really is no need for a second fader. On a multi-input mixer, where there is a requirement for the operator to balance multiple inputs, then we are presented with multiple snr optimsed sources (by virtue of the preamp gains) and then the faders, working over the majority of their travel can do the blend properly. This is best for noise and distortion. I have a good idea that the zooms and Tascams we often use, with their built in mics are designed to allow a live recording to be topped and tailed on site, with no post production - a nice start and finish by virtue of the second fader, leaving the gain tim pot set for take two, or three.....

Worst case is a pre-amp set poorly - so the s/n is compromised at the start. Bringing up the gain in a later stage cannot improve the snr.

The people in post want low noise, full range audio. They do not want undistorted but compressed range audio they have to boost, and they do not want over range distorted audio they have to spend time on trying to make it sound right. Given decent audio recordings, they can then balance and eq without worry. Providing this material is the responsibility of the recordist. Back a way - the use of the normalising function was mentioned. before the people in post can do much with a low level recording, normalising is a perfectly acceptable and useful system of getting maximum levels up to the working level while retaining the original dynamic range. I can't see any technical or production reasons for not doing it - because the people in post need their files to be in the working range of the system.

 

[ironpony]

Okay thanks, that helps a lot. I am recording the screams right though, aren't I? High gain, and low fader, till it reaches below 3db around? I have the AT4053b.

 

[Steve House]

Okay thanks, that helps a lot. I am recording the screams right though, aren't I? High gain, and low fader, till it reaches below 3db around? I have the AT4053b.

The 4053b is rated for 20mv/Pa so it would need a little more gain than a 416 but not as much as an Oktava, all other things being equal. What do you mean by "recording the screams right through"? I'd aim for medium gain, middle faders. Set your levels so that normal voice levels hit -12dBFS and reduce the faders on the screams only if they're so loud they threaten to clip on peaks. If the whispers are so soft you're tempted to raise the faders, have the talent project their voices a bit better - an actor should be able to "whisper" and still make himself heard.

 

[ironpony]

Okay thanks that helps a lot. How do I tell how much gain for a particular mic though? Is there a way of telling? I know I should get a pro, but none of my film school collaborators took audio, so until we find one, we are kind of stuck on our own.

 

[Steve House]

Okay thanks that helps a lot. How do I tell how much gain for a particular mic though? Is there a way of telling? I know I should get a pro, but none of my film school collaborators took audio, so until we find one, we are kind of stuck on our own.

You get a rough idea of how much you'll probably use with each mic by glancing at the specs to get an idea what to expect but then when you're on set, you run a "dress rehearsal" with sound check and set up the trims etc while watching the meters. The is no formula you can apply to get the right settings - every case is different.

 

[ironpony]

Okay thanks. There are listings in the manual such as Maximum Output, and Maximum Input, and Noise and Dynamic Range. Which of these would it fall under? I'm guessing for the AT4053b it falls under Open Circuit Sensitivity? If that's the one, that it says -34 is the one to be at. But that would just be way too loud, wouldn't it, so what does that mean?

 

[Steve House]

Okay thanks. There are listings in the manual such as Maximum Output, and Maximum Input, and Noise and Dynamic Range. Which of these would it fall under? I'm guessing for the AT4053b it falls under Open Circuit Sensitivity? If that's the one, that it says -34 is the one to be at. But that would just be way too loud, wouldn't it, so what does that mean?

No, that -34dBv is NOT a setting or recording level to be at! Rather it is an indication of the output voltage of the mic when exposed to a standard sound level of 94dBSPL, which is a very loud sound (much louder than normal speech). Such specs are useful for comparing one mic to another and will tell you that an Oktava, for example, will need more gain than a 416 when exposed to the same sound, but it will NOT tell you where to set the gain for a particular recorder and recording situation with any particular mic. That has to be done by experimentation as part of the setup for each recording session or take.

 

[ironpony]

Okay thanks. I just don't know how to tell since every mic it seems sounds the same once you get below 75%, at least with the FR2-LE, they do with me.

 

[paulears]

It's getting too scientific to make sense. I have a recorder with 0, -10, -20 and -30 db settings on the preamp. I start on 0, and if the levels with the recording level control set to 70% or so are off the scale, I try again with -10, and then if necessary with the other settings until I get a useful range on the meter, with the big gain control at a practical setting. I've never had a need to read up on a mic's sensitivity or output level to make it work.

 

[Steve House]

It's getting too scientific to make sense. I have a recorder with 0, -10, -20 and -30 db settings on the preamp. I start on 0, and if the levels with the recording level control set to 70% or so are off the scale, I try again with -10, and then if necessary with the other settings until I get a useful range on the meter, with the big gain control at a practical setting. I've never had a need to read up on a mic's sensitivity or output level to make it work.

Yep, that's why I keep trying to tell our OP - with the faders centered, adjust the trim control for a proper meter reading with the specific setup in question. The METER, not some number on a control knob's dial, is the arbiter of the "right settings." Only introduced the notion of the mic outputs to get across the point that different mics will behave differently is a given situation and there is no "one size fits all" setting to use.

 

[ironpony]

Okay thanks I'll do that. I am going to record some actor's screams tomorrow. I have gotten the hang of normal dialogue, but loud sounds and quiet sounds are still tricky. Now that I know I am suppose to center the fader, that helps. With normal dialogue volume I just did it with the faders turned all the way up, then moved them back. Once the fader is centered, and I then get the trim for proper reading. But do I move the fader at all, once the proper trim has been established?

 

[Steve House]

Okay thanks I'll do that. I am going to record some actor's screams tomorrow. I have gotten the hang of normal dialogue, but loud sounds and quiet sounds are still tricky. Now that I know I am suppose to center the fader, that helps. With normal dialogue volume I just did it with the faders turned all the way up, then moved them back. Once the fader is centered, and I then get the trim for proper reading. But do I move the fader at all, once the proper trim has been established?

You try not to. The trim is your coarse gain control, the fader fine tunes it, so the trim gets you into the ballpark with the faders centred and then the fader is adjusted a bit up or down to set the final recording level. "Gain riding", turning the faders up on quiet passages and down on louder ones, may sometimes be necessary if the dynamic range of the performance is especially wide but in general you try to avoid that. It also difficult to do smoothly with rotary knob faders like your recorder. Larger scale productions that are doing multitrack recording may use a flat mixing desk with sliding faders that are easier to adjust 'on-the-fly'.

 

[ironpony]

Okay thanks. I think the screams recorded better this time. I was the one screaming, not listening, but the levels were at 12 and 6 pretty much the whole time. They sound good so far and will have to play them back on my computer later to be sure. Thanks, now I have an idea of how to record louder and softer sounds. Thanks.