Audio Levels Technique

[tirapell]

Simple question but something I haven't heard from all of the board pros (the guys making money doing sound work). Everyone just mentions 'riding the levels'.

Think spontaneous documentary-style for a second, rather than scripted. If you're capturing dialogue (or any relevant sound really but mostly dialogue), are you looking to get the average audio level at -20dbfs and keep the majority between -20dbfs and -12dbfs? So on a Sound Devices mixer/recorder, basically keeping the the audio at the top of the green and into the orange LEDs? That would give you 12dbfs of headroom.

Or do you keep the majority of dialogue peaking at -20dbfs allowing 20dbfs of headroom?

This is really a simple question and I apologize if it's been beat to death, but I just wondered what standard sort of protocol was for experienced sound guys. I know scripted stuff will be different as you know what the lines are in advance and can plan ahead accordingly.

 

[Mitch.Kear]

Depends on what you're doing - -20, -18, -12 or L-12 R-20. If you're doing a seated interview you're not going to have too much to worry about so get the best signal you can. You are doing spontaneous shooting, so maybe it's a good idea to set your left channel for average -12 and right for -20 incase something unexpected comes.

 

[tirapell]

Depends on what you're doing - -20, -18, -12 or L-12 R-20. If you're doing a seated interview you're not going to have too much to worry about so get the best signal you can. You are doing spontaneous shooting, so maybe it's a good idea to set your left channel for average -12 and right for -20 incase something unexpected comes.

Well, what about say following a high school sports team around guerrilla doc style ? You're going to have talking, coaching, noise related to the sport, etc. Some is going to be quite a bit louder than others. No gunshots or anything really loud but a range of volume.

It seems like doing -12/-20 might be a good option. How easy is it to work with the two channels in post to select the one you want?

I guess there is no real standard. I just wondered what setup professionals used the majority of the time without really having a firm grasp on headroom needs.

 

[Mitch.Kear]

Very simple in post, just work with the -12 track and take from the -20 when the -12 isn't usable.

 

[Steve House]

Simple question but something I haven't heard from all of the board pros (the guys making money doing sound work). Everyone just mentions 'riding the levels'.

Think spontaneous documentary-style for a second, rather than scripted. If you're capturing dialogue (or any relevant sound really but mostly dialogue), are you looking to get the average audio level at -20dbfs and keep the majority between -20dbfs and -12dbfs? So on a Sound Devices mixer/recorder, basically keeping the the audio at the top of the green and into the orange LEDs? That would give you 12dbfs of headroom.

Or do you keep the majority of dialogue peaking at -20dbfs allowing 20dbfs of headroom?

This is really a simple question and I apologize if it's been beat to death, but I just wondered what standard sort of protocol was for experienced sound guys. I know scripted stuff will be different as you know what the lines are in advance and can plan ahead accordingly.

-20dBFS is the nominal alignment level for a 0VU sinewave tone. For a sinewave, the distribution of energy is such that an average reading meter indicates the same as a peak reading meter. But speech isn't a pure sine tone. For real-word waveforms, the average level is about 8 to 10 dB lower than the peak level. Thus a signal that reads -20dBFS on an averaging meter will read -12dBFS (more or less) on a peak reading meter. Most digital cameras and recorders are equipped with peak-reading meters. In other words, speech at an average level of -20dBFS on an averaging meter with classic VU ballistics will read around -12dBFS on the typical camera's peak reading meter. If you record keeping the peaks down at -20 on your peak reading meter, you'll be recording about 8dB lower than you ought to.

With the SD mixer, send 0VU tone to your camera or recorder and adjust the camera input pots so the tone reads -20dBFS. Lock off the camera input pots with tape so they can't accidently be moved. Tone off. Place the channel fader for the mic at the zero point, mid-rotation. While talent is speaking into the mic at the volume they'll be speaking during the scene, increase the mic channel trim control until the mixer's meter bounces around in the orange. If you look at the camera's meters they will be bouncing around -12 to -10. This is the correct nominal level for dialog. Engage the mixer's limiters. Use the channel fader to make fine adjustments to the gain during the take. Even in run-and-gun situations where you can't take as much time to adjust the mic channel in the mixer, you should still use the same process with tone to align the mixer to the camera, lock off the camera level controls, and make ALL level adjustments while shooting at the mixer, using its controls to kee the mixer running around 0VU.

 

[tirapell]

-20dBFS is the nominal alignment level for a 0VU sinewave tone. For a sinewave, the distribution of energy is such that an average reading meter indicates the same as a peak reading meter. But speech isn't a pure sine tone. For real-word waveforms, the average level is about 8 to 10 dB lower than the peak level. Thus a signal that reads -20dBFS on an averaging meter will read -12dBFS (more or less) on a peak reading meter. Most digital cameras and recorders are equipped with peak-reading meters. In other words, speech at an average level of -20dBFS on an averaging meter with classic VU ballistics will read around -12dBFS on the typical camera's peak reading meter. If you record keeping the peaks down at -20 on your peak reading meter, you'll be recording about 8dB lower than you ought to.

With the SD mixer, send 0VU tone to your camera or recorder and adjust the camera input pots so the tone reads -20dBFS. Lock off the camera input pots with tape so they can't accidently be moved. Tone off. Place the channel fader for the mic at the zero point, mid-rotation. While talent is speaking into the mic at the volume they'll be speaking during the scene, increase the mic channel trim control until the mixer's meter bounces around in the orange. If you look at the camera's meters they will be bouncing around -12 to -10. This is the correct nominal level for dialog. Engage the mixer's limiters. Use the channel fader to make fine adjustments to the gain during the take. Even in run-and-gun situations where you can't take as much time to adjust the mic channel in the mixer, you should still use the same process with tone to align the mixer to the camera, lock off the camera level controls, and make ALL level adjustments while shooting at the mixer, using its controls to kee the mixer running around 0VU.

EXCELLENT explanation Steve. I've gone this route before -- into camera with an HDV camcorder. Partly why I'm just recording double system this time with a 702 and a DSLR. I won't have to worry about the input to camera.

You answered exactly what I was asking. Thank you. If I understood correctly, the meters should bounce in the orange (-20dbfs to -12dbfs) which will average about -20dbfs for correct nominal level dialog capture. If I'm wrong, please clarify.

I appreciate all the help.

 

[Mitch.Kear]

-20dBFS is the nominal alignment level for a 0VU sinewave tone. For a sinewave, the distribution of energy is such that an average reading meter indicates the same as a peak reading meter. But speech isn't a pure sine tone. For real-word waveforms, the average level is about 8 to 10 dB lower than the peak level. Thus a signal that reads -20dBFS on an averaging meter will read -12dBFS (more or less) on a peak reading meter. Most digital cameras and recorders are equipped with peak-reading meters. In other words, speech at an average level of -20dBFS on an averaging meter with classic VU ballistics will read around -12dBFS on the typical camera's peak reading meter. If you record keeping the peaks down at -20 on your peak reading meter, you'll be recording about 8dB lower than you ought to.

With the SD mixer, send 0VU tone to your camera or recorder and adjust the camera input pots so the tone reads -20dBFS. Lock off the camera input pots with tape so they can't accidently be moved. Tone off. Place the channel fader for the mic at the zero point, mid-rotation. While talent is speaking into the mic at the volume they'll be speaking during the scene, increase the mic channel trim control until the mixer's meter bounces around in the orange. If you look at the camera's meters they will be bouncing around -12 to -10. This is the correct nominal level for dialog. Engage the mixer's limiters. Use the channel fader to make fine adjustments to the gain during the take. Even in run-and-gun situations where you can't take as much time to adjust the mic channel in the mixer, you should still use the same process with tone to align the mixer to the camera, lock off the camera level controls, and make ALL level adjustments while shooting at the mixer, using its controls to kee the mixer running around 0VU.

Much more informative than what I said haha, good stuff!

 

[Steve House]

EXCELLENT explanation Steve. I've gone this route before -- into camera with an HDV camcorder. Partly why I'm just recording double system this time with a 702 and a DSLR. I won't have to worry about the input to camera.

You answered exactly what I was asking. Thank you. If I understood correctly, the meters should bounce in the orange (-20dbfs to -12dbfs) which will average about -20dbfs for correct nominal level dialog capture. If I'm wrong, please clarify.

I appreciate all the help.

Didn't pick up that you were using a 702 but yes, it's essentially the same. The default setting for the 702's meter ballistics is VU/peak so it shows average levels as a continuous line of illuminated LEDS with a flickering dot above it to indicate the peak level. I'd ride gain so the peak dot causes the last orange LED or the first red LED above the orange to light frequently and the one above that to occasionally (very rarely) flicker. That should put the continuous bar of illuminated LEDs bouncing in the middle of the orange. You might find it easier to monitor the peaks if you change the meter in the setup so it is VU/peak-hold and the peak indicator LED doesn't flash quite so briefly.

 

[tirapell]

Didn't pick up that you were using a 702 but yes, it's essentially the same. The default setting for the 702's meter ballistics is VU/peak so it shows average levels as a continuous line of illuminated LEDS with a flickering dot above it to indicate the peak level. I'd ride gain so the peak dot causes the last orange LED or the first red LED above the orange to light frequently and the one above that to occasionally (very rarely) flicker. That should put the continuous bar of illuminated LEDs bouncing in the middle of the orange. You might find it easier to monitor the peaks if you change the meter in the setup so it is VU/peak-hold and the peak indicator LED doesn't flash quite so briefly.

Great advice. This plus recording the single mic input on both Tracks A & B with Track B gain set 8dBFS lower than Track A should cover 90% of what I'm doing. I should end up with a WAV file with two tracks -- one perfect for normal dialogue and the other making up for the times when loud yelling, sounds, etc occur without warning.

Thanks Steve and Mitch.

 

[tirapell]

Depends on what you're doing - -20, -18, -12 or L-12 R-20. If you're doing a seated interview you're not going to have too much to worry about so get the best signal you can. You are doing spontaneous shooting, so maybe it's a good idea to set your left channel for average -12 and right for -20 incase something unexpected comes.

Is there a way to set TWO different levels for 1 input on the 702? I've tried to set Input 1 (my shotgun and only source) to record to both Track A and Track B but can't seem to adjust the gain individually. The front panel knob only works on the input side, causing both tracks to be recorded at the same level. Ideas? Or not possible?

 

[Steve House]

Is there a way to set TWO different levels for 1 input on the 702? I've tried to set Input 1 (my shotgun and only source) to record to both Track A and Track B but can't seem to adjust the gain individually. The front panel knob only works on the input side, causing both tracks to be recorded at the same level. Ideas? Or not possible?

Easy when you add a mixer such asa SD302 to the mix. Mic feeds mixer. Mixer's input is panned centre so the mic appears at equal level on the left and rigt output channels. Left out to input 1 to track A on the 702, right out to input 2 to track B with the level set 6dB lower than 1.

 

[tirapell]

Thanks Steve. Going to test the waters with that setup on the 302 with 702.

702:
Input 1 set to +2dB gain going to Track A
Input 2 set to -6dB gain (because -6 is as low as line input will allow) going to Track B.
This gives -20dBFS (Track A) and -28dBFS (Track B) when using the 1k tone on the 302.

That gives -8dB of headroom on Track B. Am I missing anything? I assume you would monitor the recorder headphone out in this setup (not the mixer)?

 

[Steve House]

Thanks Steve. Going to test the waters with that setup on the 302 with 702.

702:
Input 1 set to +2dB gain going to Track A
Input 2 set to -6dB gain (because -6 is as low as line input will allow) going to Track B.
This gives -20dBFS (Track A) and -28dBFS (Track B) when using the 1k tone on the 302.

That gives -8dB of headroom on Track B. Am I missing anything? I assume you would monitor the recorder headphone out in this setup (not the mixer)?

It's not 8dB of headroom, it's an 8 dB track difference, but other than that semantics quibble it seems fine. An 8dB difference seems to me to be a little more than you would actually need; 6dB would be fine. Track A needs to be set for +2 to align the 1kHz tone from the mixer to -20dBFS on the recorder? Check the 302's setup menu to see if someone has changed any of the factory defaults - output levels and such. For monitoring, you could use the recorder headphone out or give yourself the most flexibility and bring a return from the recorder's headphone out back to the mixer's monitor return in. Phones plug into the mixer. Then you can switch select on the fly at the mixer between monitoring the inputs or the recorder. Remember the confidence monitoring in the recorder introduces about a 12 second delay.

 

[tirapell]

It's not 8dB of headroom, it's an 8 dB track difference, but other than that semantics quibble it seems fine. An 8dB difference seems to me to be a little more than you would actually need; 6dB would be fine. Track A needs to be set for +2 to align the 1kHz tone from the mixer to -20dBFS on the recorder? Check the 302's setup menu to see if someone has changed any of the factory defaults - output levels and such. For monitoring, you could use the recorder headphone out or give yourself the most flexibility and bring a return from the recorder's headphone out back to the mixer's monitor return in. Phones plug into the mixer. Then you can switch select on the fly at the mixer between monitoring the inputs or the recorder. Remember the confidence monitoring in the recorder introduces about a 12 second delay.

Ok Steve. I reset the 302 to factory defaults then went through the menus. However, tone still shows up just under -20dB on the 702. I went into the setup menu and raised the Tone Oscillator Level +1dB (setup menu -14) and that has it reading correctly on the 702. Why the tone is off 1dB...I have no idea. It shouldn't effect recording levels though since I reset the outputs to factory defaults, correct?

Also backed Track B down to -6dB difference. I'm not sure I understand the benefit of running Track B -6dB lower since the signal has already passed through the mixer and the limiters. Like you said, it's not really headroom is it? It's just -6dB quieter. Any real world advantage here? Or am I just doing a long exercise in looking like a professional (when clearly I am NOT) by setting up an extra step in the process?

I set up the RTN as suggested. Works great. Although, I believe you meant a 12 millisecond delay right? Not 12 seconds. That might drive you insane...

As always, I appreciate the generous help. Thank you sincerely.

 

[Alex H.]

I'm not sure I understand the benefit of running Track B -6dB lower since the signal has already passed through the mixer and the limiters. Like you said, it's not really headroom is it? It's just -6dB quieter. Any real world advantage here? Or am I just doing a long exercise in looking like a professional (when clearly I am NOT) by setting up an extra step in the process?

Analog devices have much more headroom than digital devices. In the analog world, the signal can cross far above 0dB quite gracefully without clipping/distortion (sometimes as far as +18dB, +22dB, or more... depending on the device). With digital, 0dB is the top and cannot be crossed. At 0dBFS (Full Scale), all the bits are full (they're all 1s) and no more data can be added, so you'll get nasty clipping in the form of square waves as the system tries to cram more garments into an already-stuffed suitcase. This is why the reference level for Full Scale (digital) isn't 0dB but is far below that.

Feeding a 302 to a 702 and creating the -6dB safety track in the 702 does have advantages. It can save the audio from clipping if the analog mixer hits a level that the digital recorder cannot handle. It won't protect you from clipping within the mixer (though Sound Device's "unclippable" limiting is pretty reliable).

With mixers like the MixPre and the 302, panning is hard-switched L-C-R. There's no way to create the safety track within the mixer itself. In the old days, when Shure mixers were the industry standard (FP32, 32a, and 33), the safety track could be created two ways: rotary pan on the input set to 10:00-ish would send more signal to the left than to the right (this is the ideal method), and the master rotary level control was tiered and could control left and right output levels independently (not ideal, but still effective). For Sound Devices, one must move up to the 442 or 552 in order to have rotary pan control. That is really the best way to set a safety track, since it can protect from clipping on the mixer's output.

Still, creating the safety track in-camera or within the recorder's controls can save you in the field.

 

[tirapell]

Alex, great explanation of the systems. Now it makes more sense why the 302's LEDs are measured in dBu (-30dBu to +20dBu in 20 segments) and the 702's LEDs are measured in dBFS (-50dBFS to 0dBFS in 19 segments). Analog vs. digital, duh...

As you can see, I have just enough knowledge to be dangerous, but I'm willing to listen and learn.

 

[Steve House]

Ok Steve. I reset the 302 to factory defaults then went through the menus. However, tone still shows up just under -20dB on the 702. I went into the setup menu and raised the Tone Oscillator Level +1dB (setup menu -14) and that has it reading correctly on the 702. Why the tone is off 1dB...I have no idea. It shouldn't effect recording levels though since I reset the outputs to factory defaults, correct?
....

When tone is turned on, the 302 shouyld be puttiing out 0dBu. That in turn should align to -20dBFS on the 702. Not sure if you meant you set the tone to output +1dBu or you had to increase it 1dB to get it up to 0dBu. In any case, tone should be at 0dBu.

 

[tirapell]

When tone is turned on, the 302 shouyld be puttiing out 0dBu.

Correct. It is. As displayed on the 302 meters. However, on the 702 meters, it shows -18dBFS (which is likely somewhere between -18 and -20 since there is no -19).

That in turn should align to -20dBFS on the 702. Not sure if you meant you set the tone to output +1dBu or you had to increase it 1dB to get it up to 0dBu.

Had to increase tone output +1dBu (in 302 mixer setup menu) from 0dBu to get it to read -20dBFS on the 702 meter. Could the mixer be "off" that much?

 

[tirapell]

302 Setup Menu
–14​

Tone Osc. Level ​

OFF (–30), –20 dBu (–26) –10 dBu (–24), 1 dB increments to +8, 0 dBu is default​

Selects the output level of the tone oscillator in dBu at LINE level​

 

[Steve House]

Nothing wrong with the 702 reading -18 dBFS when presented with a 0dBu tone, I'd leave the 302 at the default tone output level rather than mucking around with it. In fact 0dBu = -18dBFS is the EBU alignment standard and works just fine. If you really want to pik nits, the SMPTE standard is that 'nominal analog studio line level' of 0dBVU aligns to an output level of +4dBu which in turn aligns to -20dBFS on a digital device, which means 0dBu actually should align to -24 dBFS. But all we really need to be concerned with is when you look at the mixer meter and see something registered there, you know what it implies is going on in the recorder. What it tells you is that when the mixer's 0dBVU light is on you have 18 db of wiggle room before the recorder will clip. Set your mixers output limiters to kick in at, say, +16dBu, if your meter is set to VU mode keep dialog level around the 0dBu mark on the mixer, and you don't have to worry about clipping as any signal excursion trying to drive the recorder past -2dBFS will get limited before it leaves the mixer.