Minimum recording levels for studio and field audio recordings? (help settle a disagreement)

[Khashyar Darvich]

P.S…. essentially what the audio engineer was saying was that what really matters is how the audio sounds in the headphone, not what the audio displays on the audio meter (except if it peaks over zero).

[Will Salley]

[Khashyar Darvich] ” it does not matter much how LOW the audio level is, as long as it sounds good on the headphones”

Not true.
Although well-designed and built professional equipment will generally provide a clean signal at less-than-optimum levels, it will need to be boosted (gain added) for proper levels during mixdown. As the level is increased, so is noise.

Audio recorded at the optimum level – which is, theoretically, the highest possible level before distortion due to clipping – will ALWAYS sound cleaner than audio recorded at a lower level.

Headphones are not a very accurate way to monitor audio that is intended to be reproduced on loudspeakers – It’s just the most practical way at this time. For example, the next time you happen to be recording a take and a car with the dreaded boomy bass drives by. Play it back in the field with cans on. Chances are, you won’t hear anything. Then take it in the studio and listen to the same take. Unless a low-cut filter was used, you will probably hear the thumpy stuff.

System Info – G5/Dual 2 – 10.4.7 – QT v7.1.2 – 8GB ram – Radeon 9800Pro – External SATA Raid – Decklink Extreme – Wacom 6×8

[Khashyar Darvich]

Thank you very much for your feedback, Will.

Yes, I have always heard that you should record audio at the highest level without clipping.

The audio engineer also mentioned to me that you want to set the audio level a bit lower during a field recording in case their is an unexpected loud sound that could peak the levels.

I’m going to speak with him again (so that we can receive a clear understanding) after I receive a couple of more comments).

Thanks,

Khashyar

[Thaxter Clavemarlton]

[Khashyar Darvich] “The audio engineer also mentioned to me that you want to set the audio level a bit lower during a field recording in case their is an unexpected loud sound that could peak the levels.
“

I have always done this with recording “dual mono” on a 2-track recorder (or camcorder).

Setting one track 3 to 6 dB lower than the other. That way you have more head-room on one track, and less “less noise” on the other.

[Khashyar Darvich]

Thank you very much for your feedback, thax..

Again, this is confirming what I had always thought to be true about the standard way of setting levels: set the levels as high as you can without peaking or going over “zero.”

My audio engineer also said that with digital recording, especially in a very quiet environment like a studio (and especially with very high end mics like the ones that we used to record the celebrity), that you can increase gain without any hiss or other artifacts.

Are you saying that if audio is recorded low (let’s say, with an average level between -30db and -24 db), that increasing the gain so that the average is -12 to -6 db WILL result in some artifacts or hiss, even if the gain is added through ProTools?

Thanks again for all of your feedback and thoughts on this,

Khashyar
Khashyar

[Thaxter Clavemarlton]

[Khashyar Darvich] “Are you saying that if audio is recorded low (let’s say, with an average level between -30db and -24 db), that increasing the gain so that the average is -12 to -6 db WILL result in some artifacts or hiss, even if the gain is added through ProTools?”

Always.
What ever noise there is will be increased when you bring up the gain.

But it may or may not be noticable in the finished presentation.

A quiet studio with good mics will yield better results than under lesser conditions.
But the big question is WHY would someone INTENTIONALLY record at such a LOW level they KNOW will need to be increased tremendously for the finished production?

Yes, avoid clipping at all costs, but there is a BIG difference between clipping and -30dB on peaks.

[Stephen Muir]

“Never boost when you can cut” as the old recording maxim goes.

Although a low-level recording made with good equipment may not accumulate an appreciable amount of hiss when boosted, the original recording will still have less detail than one made with propper levels. If you’re recording is 16-bit, then an under-mixed will have a very narrow dynamic range (perhaps only 12-bits or even 8-bits). The more you push it, the more digital artifacting will be apparent, because the subtle details just don’t exist in the recording. It’s analagous to under-exposing video footage, and then brightenning it until it reads propperly on a waveform monitor (in which case you can kiss the shadow detail goodbye). Although there’s no tape hiss in a digital recording, you’re still dealing with a finite amount of information.

I’m currently working on the dialogue edit for a film on which the production DATs were badly under-recorded (maximum peaks at around -30dB, averaging around -50 dB), and the dialogue sounds horrible no matter what we do short of ADR.

However, it is true that location sound is a bit more unpredictable, making it prudent to back off production mixing levels slightly (-20 dB rather than -12 dB). That said, between dual-mono techniques and a solid set of limiters (ie. a good ENG/EFP mixer) -12 dB can still be a good level to aim for.

As for the studio recording matter, unless the dialogue was being recorded to perspective (ie. it was going to be that low in the final mix), I don’t understand why the session mixer would want to intentionally provide less than optimum quality recordings.

[Ty Ford]

First, are we talking peak or RMS?
Second, today’s digital audio recorders in 24-bit mode CAN do wonders. Not all of them, but some.
This excludes camera audio.
Third, depends on the sound. If it’s dialog, you want that up front. Peaks at -3dB down to -14 dB on a dedicated professional recorder will be OK, but camera audio is a lot noisier and the lower voices and sound will be in the noise floor. Bring up the level and you’ll also bring up the noise.

I purposely recorded audio with -35 dB peaks on a Sound Devices 744T and normalized the files to 100%. Yes there was some hiss, but less than you’d have gotten with an analog tape recorder. I was pretty impressed.

If the Music and SFX are prominent enough, you’ll probably get away with it because they will mask the system noise and ambient sound around the dialog.

Regards,

Ty Ford

Ty Ford’s “Audio Bootcamp Field Guide” was written for video people who want better audio. Find out more at https://www.tyford.com

[Khashyar Darvich]

Hi guys,

I appreciate very much your feedback and thougtful comments.

This ios helping me have a better understanding of optimum audio levels.

I just want to clarify a couple of things about the circumstance that I am referring to:

1) The narration recording that I am referring to took place at one of the top three recording studios in Los Angeles, with $10,000 to $15,000 mics (two micorphones), using ProTools.

2) the files were saved to my hard drive and to a DVD as Pro Tools .wav files, and I am checking levels after the files have been imported to Final Cut Pro, panned to one side, and checked on Final Cut Pro’s auio meter. (Is it psosible that Final Cut Pro is reading the level of the wav file at a different level than ProTools would?)

3) The average DB on the Final Cut Pro audio meter is from -24 to -30, with peaks at about -18db.

4) the recording studio had their own engineer working the HIGH end control board, and MY audio engineer was sitting with him during the recording. This is part of the reason that it baffles me that a mistake had been made on the levels (if a mistake truly was made).

5) A second audio guy who is working on this project with me laoded the files on ProTools, and the sound waves looked low there also, although the version of ProTools that he had did not show the exact Db level. This second engineer also did say that we could apply gain and the narration recording would sound great.

Here is exactly the point that our main audio engineer was making to me in a 45 minute conversation that we were having:

1) That as long as the audio “sounds” good to the ear (through headphones or through studio speakers), then that is the main judgment of whether audio levels are o.k. (and not a meter). He said that perhaps someone at an academic institution might recommend “going by the book,” and setting levels strictly by an audio meter, but that most audio engineers will go by how the audio sounds to them. (He did say, however, that the audio must NOT peak the meter, however.)

2) He said that with a very good mic and in a quiet setting, that it is possible to raise the gain of a DIGITAL audio signal with no hiss or artifacts. (Personally, this goes against what I have been taught, that you set audio levels to the higehst level possible without peaking– By the way, I like the two mono audio channel approach that was mentioned in this thread, with one channel being set slightly lower than the other).

3) He also said that during a sound mix of the film, that he will not be focused on mixing the audio at a certain level (except that it does not peak), and that during my final level output adjustment when the film is going to take is when I should set the final overall level so taht it does not go past zero.

Thanks again for your thoughts and feedback on this. I am also copying and pasting your comments to him (and the second audio guy) in an email so that they and I can have a good, definitive discussion about this.

Khashyar

[Ty Ford]

1) That as long as the audio “sounds” good to the ear (through headphones or through studio speakers), then that is the main judgment of whether audio levels are o.k. (and not a meter). He said that perhaps someone at an academic institution might recommend “going by the book,” and setting levels strictly by an audio meter, but that most audio engineers will go by how the audio sounds to them. (He did say, however, that the audio must NOT peak the meter, however.)

**Headphones rather than meters can be deceiving, depending on where they are plugged in. Posing that only academics go by the book is pure crap. “Audio Peaking a meter” is an unclear choice of words. My meters show peak all the time. What you don’t want is audio that exceeds 0dB.

2) He said that with a very good mic and in a quiet setting, that it is possible to raise the gain of a DIGITAL audio signal with no hiss or artifacts. (Personally, this goes against what I have been taught, that you set audio levels to the higehst level possible without peaking– By the way, I like the two mono audio channel approach that was mentioned in this thread, with one channel being set slightly lower than the other).

**Yes this is true. I have a Neumann TLM 103. In my studio it’s extremely quiet. Quiet enough to record at a lower level and survive being raised without increasing the noise floor too much.

3) He also said that during a sound mix of the film, that he will not be focused on mixing the audio at a certain level (except that it does not peak), and that during my final level output adjustment when the film is going to take is when I should set the final overall level so taht it does not go past zero.

**Sounds reasonable.

Thanks again for your thoughts and feedback on this. I am also copying and pasting your comments to him (and the second audio guy) in an email so that they and I can have a good, definitive discussion about this.

**If there’s enough other sound to cover any noise that might be there, it’s a moot point, almost. There is an issue of decreased definition because lower record levels mean fewer bits are used and those bits are all about resolution, but as I said before, with really good converters, preamps and low noise mics, you’ll probably not hear any noise if that track is in a mix. If it’s low in the mix, the odds are you’ll never hear the noise.

Ultimately, it’s the ears that count. Different ears are sensitive to different things. It’d bug me to record dialog that low, but not percussion tracks in a music mix that I knew would be low in the mix. Maybe this guy has an airtight, anechoic studio with mics that have a selfnoise of less than 9 dB-a. You can get away wih a lot with that sort of stuff. Most studios DON’T have that luxury.

Regards,

Ty (non-academic) Ford

Ty Ford’s “Audio Bootcamp Field Guide” was written for video people who want better audio. Find out more at https://www.tyford.com