Zoom H4N really low levels but still distorting - why?

sounds like a nice system. When you apply the compression to normalise loud and quiet parts is it automatic or do you need to tell if where the loud parts are that need dropping and where the quiet parts are that need boosting?
 
TheDingo said:
The process I normally go through when working with audio from a video clip is:

1- Export .WAV copy
2- Use Parametric EQ to tweak the overall sound of the audio
3- Use Compression to manually normalize the loud to quiet parts
4- Use a mild form of noise-reduction to remove any annoying background sound
5- Adjust the volume of the finished audio track to -0.1 dB
6- Save the .WAV copy then jump back to Vegas to finish the video edit ( my audio work is automatically synch'd with the video )
Click to expand...


It's important to point out that this is an event videography kind of work flow, not a narrative film workflow. Which is on point with the OP but I just don't want would be filmmakers to start quoting this for narrative work. Those are scorchingly hot levels that will be clipping the D/A converters because of the crest factor. That is going to be in the higher frequencies so it's possible that it won't be noticeable on your clients systems. But this is way over what broadcast would find acceptable and way hotter than any film is released at. If you played that in a calibrated room it would really hurt.

But that is a point some of us have been trying to make, there is no one way that fits all workflows. A few years ago I would probably be pointing out why that workflow is wrong. But that is because I was coming at it from a very specific direction, narrative features. I would still argue for lower final level but the rest is a workflow based on time and budget and seems to be working for you so that is pretty valid.

Since I have been privileged/ forced to work in a number of different areas of audio post I have come to realize that you need to alter what you do, sometimes radically.

If you try to do everything with a single workflow it can kill you. A full narrative film workflow (my favorite) takes time and a crew or lots of time and me. If you try to apply that to a fast turnaround doc your dead.

The global wide brush processes your using I would never use in a narrative post, but I certainly have in other workflows.
 
Noiz2 said:
I don't know Premiere Pro, but if the top of the scale is zero then I would assume yes.
Click to expand...

yes it is zero at the top so I'll try to remember to peak at -3 and not 0.

Ok so in summary, dont record so hot, peak between -20 and -12 on the zoom and then boost in post anything thats not loud level like clapping etc so that its all relatively even for the listener peaking at -3?

That will be my new audio principle then! I think I've learned more in this thread than any other one in the past!
 
Noiz2 said:
But this is way over what broadcast would find acceptable and way hotter than any film is released at. If you played that in a calibrated room it would really hurt.
Click to expand...

Hey Noiz2,

All of my work to date has not been for broadcast. What is the standard audio level for broadcast ?

I am completely ignorant of broadcast audio standards, and would greatly appreciate being brought up to speed.

Thanks.
 
Noiz2 said:
Kinda sorta, not really. 0dBfs is a fixed thing, that's all there is. Analog scales are based on a reference level. So 0dB on an analog meter can relate to just about anywhere depending on what your reference is. It used to be standard on Nagra's to use -18 as your zero reference because there is a nice tick mark at -18. When everything went digital meters started having the tick at -20 so that became the standard. Though in Europe I think -18 is still marked and often used. The only reason it matters these days is because there is still a lot of literature talks about zero (in the analog realm) and folks who don't know the background take it literally with digital meters and they get in trouble.

OK there are a lot of reasons it matters but for this discussion...

In the analog world you picked your zero and as long as you wrote it on the box so others knew how to calibrate the playback deck it was all good. In digital you got what you got and the most you can get is zero. So you need to build in the headroom by keeping your levels low enough that you don't hit the ceiling.

Actually bit depth has nothing to do with headroom. Zero didn't change, you didn't get any more range. It's a good recommendation and your reasoning is sort of close. You get more resolution not more headroom. When you don't use all of the available bits your effectively recording at a lower bit depth. So at 16 bits you may effectively only be using 12 or so bits. By recording at 24 bits your using maybe 20 bits of resolution so it sounds better. You can record quite a bit lower and still have better resolution than 16 bit. If you record lower you give your self more headroom, but the bit depth didn't add the headroom you did by recording lower. The bit depth gave you a better sounding recording.

It may seem like splitting hairs but the distinction is important. Your trading resolution for headroom, 24 bit gives you more resolution to trade with.
Click to expand...

I got -18dBFS (or -20dBFS, depending on who you ask like you said) from a bunch of different sources as well as running a a 1kHz tone at -18dBFS out to my analog mixer. Where did it go when the fader was set to unity gain? 0dB. The +4dBu "professional line level reference" is based off this calibration. It's not an arbitrary reference. It all relates to the nominal level of where audio gear is best operated for I believe the maximum signal to noise ratio.

By 24-bit audio allowing for lower level recording, I meant that the digital quantization noise at 24-bit is much lower than 16-bit. I didn't say it adds extra headroom but that it gives greater headroom potential. I didn't want to get too technical in my reasoning to reduce instances of heads exploding, but here goes...

Source: http://en.wikipedia.org/wiki/Quantization_(sound_processing)

"There is also a good case for 24-bit (or higher) recording in the live studio, because it enables greater headroom (often 24dB or more rather than 18dB) to be left on the recording without encountering quantization errors at low volumes. This means that brief peaks are not harshly clipped, but can be compressed or soft-limited later to suit the final medium."

Think of the nominal audio recording level like your "perfect exposure" on a camera. Yes, you can underexpose a shot but if you bring it up later in post you'll have more noise than if you had exposed the scene properly to begin with. Overexpose (with digital) and you cannot get that detail back. "Overexposing" your audio is like oversaturation (see the massivemastering link below) or worse clipping. Both are forms of bad distortion that you want to avoid.

Also read a lot of info that I put into practice recording audio at http://www.gearslutz.com/board/so-m...s-don-t-sound-good-analog-mixes-restored.html . Cheap analog to digital converters also have issues with phase and frequency when oversaturated. So record at a "clean -20/-18dB level" and then boost digitally if you need more volume in post. The burnt steak analogy that was used on the blog post I linked to earlier ( http://www.massivemastering.com/blog/index_files/Proper_Audio_Recording_Levels.php ) points this oversaturation phenomenon out. Sorry if I split anymore hairs. ;-P
 
adamr316 said:
I got -18dBFS (or -20dBFS, depending on who you ask like you said) from a bunch of different sources as well as running a a 1kHz tone at -18dBFS out to my analog mixer. Where did it go when the fader was set to unity gain? 0dB. The +4dBu "professional line level reference" is based off this calibration. It's not an arbitrary reference. It all relates to the nominal level of where audio gear is best operated for I believe the maximum signal to noise ratio.

.... ;-P
Click to expand...

Just to make it interesting, there are two common calibration standards,

SMPTE: 0VU = +4dBu = -20dBFS
EBU: 0VU = 0dBu = -18dBFS.

But something that often escapes notice is that a lot of newer mixers available is SMPTEland, such as the Sound Devices field mixers and the Mackie compacts, calibrate their meters to the EBU standard rather than the SMPTE, so that an indicated 0VU on the meter equals a 0dBu output from the mixer. That means if you want to go strictly by the SMPTE standard, one should align 0VU tone coming out of those mixers to -24dBFS on a digital recording device. But the only time you actually have to worry about such strict adherence to those standards is when delivering your final mix for broadcast. When making the original recordings in the field, one should be consistent in the aligment levels chosen, so that when one looks at the mixer's meter one knows what's going on inside the camera or recorder, but you can set a personal standard based on the nature of the sound being recorded. Align to -24 and you have 24dB of leeway before your signal clips, align to -18 and you have 18dB of playroom, align to -20 and split the difference. (As a side note, the -24 figure is also reflected in the more recent 'dialnorm' specifications for digital broadcasting, where dialog levels averaged over the duration of the entire program should be -24dBFS +/- 2dB, with music and FX mixed to balance against it, peaks never exceeding -2dBFS (ATSC Recommended Practices, p16).

The idea is that one should record in the field at the hottest level one can without risking approaching clipping on the highest signal peaks encountered (remembering the peaks on a real-world complex waveform such as speech are typically ~10dB above average RMS levels). Imagine an averaging meter calibrated in VU (0VU=0dBu) and a peak-reading meter calibrated in dBu connected in series so you can read the same signal on both meters at the same time. Send them a sinewave tone. When the averaging meter reads 0 VU the peak-reading meter will also read 0dBu. But if you send it speech (or a pink noise test signal, which has approximately the same energy distribution), when the averaging meter reads 0VU the peak-reading meter will read +8 to +10dBu or so. I suspect that this is why a lot of prosumer camera manufacturers put a dot at -12 on the camera's peak-reading meter - if the meter is bouncing around -12 when recording speech, that's about the same level one would see after aligning a mixer to the camera with tone at -20 and then switching the mixer to speech reading ~0VU at the mixer meter. By standardizing at 0VU tone =-18 or -20 dBFS and recording the scene's sound with the mixer indicating 0VU, one can safely record normal speech at high enough level for a good S/N ratio and have a safety margin of 10 to 12 dB before clipping.
 
Last edited:
I got -18dBFS (or -20dBFS, depending on who you ask like you said) from a bunch of different sources as well as running a a 1kHz tone at -18dBFS out to my analog mixer. Where did it go when the fader was set to unity gain? 0dB. The +4dBu "professional line level reference" is based off this calibration. It's not an arbitrary reference. It all relates to the nominal level of where audio gear is best operated for I believe the maximum signal to noise ratio.
Click to expand...

Headroom is just the amount you can go over your defined normal maximum level and still not clip. It is not some set figure. It is generally talked about in reference to your reference level. If yu set a lower reference level for "zero" then you have more headroom. This is more of a historical talking point than a modern one because with digital systems you know where your celling is.

In analog days it was based of your zero VU. Good studio equipment often 26+dB of headroom past zero VU. But it was also a soft celling.

Anyway "zero" was set for what you were doing and it was not completely out of the ordinary to have zero set to something other than -20, though that was the standard. You laid tone and and you noted where that tone was. The next folks to mount the tape played the tone and calibrated the meters to what you wrote on the box.

In general you are correct it is based around where everything is optimal, but it was also a movable reference, which is why you laid tone and marked your boxes.


By 24-bit audio allowing for lower level recording, I meant that the digital quantization noise at 24-bit is much lower than 16-bit. I didn't say it adds extra headroom but that it gives greater headroom potential. I didn't want to get too technical in my reasoning to reduce instances of heads exploding, but here goes...

Source: http://en.wikipedia.org/wiki/Quantization_(sound_processing)

"There is also a good case for 24-bit (or higher) recording in the live studio, because it enables greater headroom (often 24dB or more rather than 18dB) to be left on the recording without encountering quantization errors at low volumes. This means that brief peaks are not harshly clipped, but can be compressed or soft-limited later to suit the final medium."
Click to expand...

Ironically they are saying pretty much the point I was making above. You get more "headroom" by setting your reference level lower.
The problem is that headroom really means something very different in the digital world than it did in the analog world and people are using mixed definitions.
In the analog world it was how far above zero VU you could go and still have an acceptably undistorted signal. It was all about hardware and it varied with the hardware. You then set your zero to optimize your signal and give you enough headroom.

In digital it's pretty much all on you. Zero dBfs is it. If you need more headroom you record lower. Since headroom is defined as the space between you hottest normal signal and the point of unacceptable distortion you have just increased you head room. If you record hotter then you just shrunk your headroom, it doesn't really have anything to do with bit depth.
 
adamr316 said:
By 24-bit audio allowing for lower level recording, I meant that the digital quantization noise at 24-bit is much lower than 16-bit. I didn't say it adds extra headroom but that it gives greater headroom potential.
Click to expand...

Actually in the post I was referencing you didn't say either. You said that the "whole reason" to use 24 bit was so you could record lower.

The reason to record at 24 bit is because it sounds better. Because it sounds better you can record lower if you want to and not risk the graininess of 16 bit when it is recorded low.

And BTW if you record lower you have set your reference level lower so you are no longer at -20 (-18) = 0VU.

To a great extent it probably doesn't really matter what you (the general you not you in particular) understand or don't about it all. The point that comes out is use 24bit and record low enough that you have some good space above your peaks.

If you do that it will sound better and it doesn't really mater if your doing it because you worked out the quantum mechanics of it or because you got the advice in a fortune cookie.
 
The reason to record at 24 bit is because it sounds better. Because it sounds better you can record lower if you want to and not risk the graininess of 16 bit when it is recorded low.
Click to expand...

Well, the end result is the same, but the cause and effect are reversed. The reason it sounds better is because you have better resolution (24 bits vs. 16). And because it has better resolution, you can record at lower levels and have greater dynamic range to deal with real world conditions. Since having the luxury of 24-bit recording capability, I dread having to record in 16 bit, especially live musical events where you can't reliably predict peak levels.
 
Noiz2 said:
Headroom is just the amount you can go over your defined normal maximum level and still not clip. It is not some set figure. It is generally talked about in reference to your reference level. If yu set a lower reference level for "zero" then you have more headroom. This is more of a historical talking point than a modern one because with digital systems you know where your celling is.

In analog days it was based of your zero VU. Good studio equipment often 26+dB of headroom past zero VU. But it was also a soft celling.

Anyway "zero" was set for what you were doing and it was not completely out of the ordinary to have zero set to something other than -20, though that was the standard. You laid tone and and you noted where that tone was. The next folks to mount the tape played the tone and calibrated the meters to what you wrote on the box.

In general you are correct it is based around where everything is optimal, but it was also a movable reference, which is why you laid tone and marked your boxes.




Ironically they are saying pretty much the point I was making above. You get more "headroom" by setting your reference level lower.
The problem is that headroom really means something very different in the digital world than it did in the analog world and people are using mixed definitions.
In the analog world it was how far above zero VU you could go and still have an acceptably undistorted signal. It was all about hardware and it varied with the hardware. You then set your zero to optimize your signal and give you enough headroom.

In digital it's pretty much all on you. Zero dBfs is it. If you need more headroom you record lower. Since headroom is defined as the space between you hottest normal signal and the point of unacceptable distortion you have just increased you head room. If you record hotter then you just shrunk your headroom, it doesn't really have anything to do with bit depth.
Click to expand...

OK cool, sorry I was just going on what I have heard over and over again from engineers who are way smarter than me. Indeed YOU decide the headroom...I think I was referring moreso to hardware headroom like you said. Most hardware mixers can handle about +20dB over 0 without distorting. The good ones a tad bit more but levels at 0 is where your signal is clean.

Going back to the original post, applause will absolutely cause distortion. Gear sometimes has a built-in limiter but they usually aren't good on cheaper equipment. So in order to maximize your audio quality you want to record loud enough to avoid the noise but not too loud to avoid the distortion. The standard is somewhere around -20dBFS. This will give you 20 decibels of headroom so that when the applause happens your gear doesn't distort. You have to also set your transmitter so that it too doesn't distort prior to getting to your H4N. Due to HDX compander technology you have a lot of headroom to work with before distortion. See this video http://www.youtube.com/watch?v=TyVAjZq9HF8 starting at 2:40 is where he sets the levels for the transmitter. Coupled with a 24-bit recorder you are good to go and shouldn't have to record so loud to get clean audio in the end.

Indeed I believe we are saying the same thing now Noiz. Yes?
 
Richard Crowley said:
Well, the end result is the same, but the cause and effect are reversed. The reason it sounds better is because you have better resolution (24 bits vs. 16). And because it has better resolution, you can record at lower levels and have greater dynamic range to deal with real world conditions. Since having the luxury of 24-bit recording capability, I dread having to record in 16 bit, especially live musical events where you can't reliably predict peak levels.
Click to expand...

Oh I feel you on the 16-bit comment. Except for delivery formats I don't even want to see that number. Hehehe.
 
the last wedding was a raging success from an audio point of view based on the advice here. thanks very much
 
tikigod19,

I too have this same problem and am VERY new to sound. I am using the Zoom H4N and, for the moment, an Azden SGM 2X which I will be upgrading to a Sennheiser ME66 soon, and recording in WAV48kHz/24bit and editing in CS5. The H4N's record level is around 60-80% and my peak levels never go past -12db or -6db. However, when I bring my mono audio into CS5 the levels are super quite! Dialouge hover's around -30db. I've been trying to figure out the best way to fix this and I'm going to try what was said in this post. However, I'm still abit confused as to a few things...

1. What is the best tool,(in ethier CS5 or Soundbooth) to "Boost" my dialouge volume?
2. What would be a good db level to have my dialouge at?
3. Should I "center" my mono audio by using ethier the left or right fill effect in CS5? Or is there a better way?

Thanks so much.
 
Does anyone know the answers?

Also I came across this bit of information:2) In PP, you go to Clip>audio options>break out to mono clips. This creates a left mono and a right mono. Delete the right mono and the left mono will still play on both sides.

Is this a viable way of "Phantom centering" my mono dialogue tracks?

Any help at all would be GREATLY appreciated.

:)
 
i just apply a (varying) audio gain to each clip. Sometimes on dialogue its +30dB and sometimes with ambient sound and disco music it only needs +15dB but it always needs something
 
You must log in or register to reply here.
Back
Top