Sound ?!

[Elizabeth Lowrey]

Here's a question: how difficult would it be to actually add a limiter/AGC circuit to the camera? I'm assuming it would call for a hardware modification to the main board in the camera and some minor firmware tweaks. The motto of the RED enterprise is to render "obsolescence obsolete", and Jim has repeatedly stated that this camera will continue to evolve. So is this the kind of thing we could see as an upgrade option once all current orders are filled? Or is space and configuration of the current camera board such that it would call for not just an add on module inside but a completely new, redesigned main board?

Also, Philip, I own both a Sony DSR 300 and a DVX. Both name their auto-dynamic range solutions "AGC" (automatic gain control). My understanding is that these circuits are in effect compressors, not limiters, as the base gain level is set automatically and is different/variable depending upon the sustained sound levels in the environment while the manual knobs have absolutely no effect when AGC is turned on. I have never worked with a 2/3" or large format HD broadcast camera, but the DSR300 is certainly considered on the lower end of "broadcast quality" and does not offer a strict "limiter", where, presumably, you could still use knobs to manually select your base gain level and depend upon the limiter to only control the peaks. What sorts of cameras offer a true "limiter" as opposed to AGC circuit? And, even more curiously, which offer a choice of AGC or a limiter?

When I've used or viewed material shot on my brother's Sony VX2000, I am often acutely aware of the AGC functioning (much like its hideously annoying -- and unoptional -- automatic video gain compensation!). It's that kind of aesthetic that turns me off to the use of AGC. Maybe if I had a camera with a true limiter, that let me set the bottom and only touched the top, I would be more inclined to use it for interviews.

 

[10s]

The DVX is a limiter, but is called an AGC...those panansonic engineers? bless their little minds, but it's a limiter and that's why it's so useful. The Sony is probablly an AGC, those terrible units that raises the noise floor level and creates hiss and everything you never wanted to hear.

 

[jbeale]

Not really on topic, but I shot for several years on a VX2000, and it does indeed have fully manual exposure and manual audio gain, both are selectable options. The audio quality is nothing to write home about, though. There is actually a guy who offers a rework service bypassing some of the noisy audio chips used in the Sony VX2k and the earlier TRV900 cameras.

 

[S. Um]

What sorts of cameras offer a true "limiter" as opposed to AGC circuit?

The DVX for one.

An AGC (automatic gain control) circuit will try to raise or lower the volume of the recorded sound (by adjusting the gain) depending on whether the camera thinks the audio is too quiet or too loud. Many consumer cameras have this, but pros generally do not use this feature.

The DVX100 does NOT have AGC. It has an Automatic Limiter Control or ALC. This limiter tries to clamp down on high volumes to prevent the audio from clipping. It kicks in at about -6dB. It doesn't modify the overall signal (like AGC), just the part over -6dB. I believe most professional cameras and audio recorders have some type of limiting feature.

For more info, please refer to this post by Barry Green on DVXuser, http://www.dvxuser.com/V6/showthread.php?t=8489&highlight=limiter+gain+dvx100 (there are other posts, but I can't find them right now) or refer to the DVX Book page 77.

BTW, a compressor and a limiter actually perform similar functions. With a compressor, you set a threshold or knee (where the compression kicks in), and a compression ratio. With a limiter, the threshold is usually set just under the limiting level, and the ratio is usually very high. So essentially, a limiter is a type of compressor.

 

[david farland]

I can feel a new reduser section a coming.

Couple of things,
1. So Red doesn't have a limiter/compressor/AGC etc.
Hopefully you can gang 2 channels (even 3!!) to the one input.

What db difference would people use for say dialogue when someone may suddenly screams?
I guess using 2 channels at different db levels on the one physical input is the substitute for a limiter?

2. forgot the send thing!

Dave,

 

[10s]

So the big question is, Can RED match or surpass the DVX audio?

Yes , it's sad to say, but this seems to be a real valid question. I'll assume the RED mic preamps are decent and don't create undue nosie or color.

So far RED is not cutting it in two areas: 1. limiter, 2. manual level controls.

Red surpasses the DVX 16 bit with 24 bit and has four channels of audio instead of two, but what features are missing and needed?

Answer: manual controls and a limiter.

Here's a challenge to users of the new build release.

Hook up a mic to your new RED camera and go into a crowd, let's see what's useable. I'll bet right now you'll run across all sorts of major audio difficulties with peaking and trying to set levels as you move about. Now try the same situation with your DVX and have the ALC engaged.

Which set-up will give superior results?
What's easy and intuitive?
What allows you to focus on artistic expression rather than technical settings?

Shouldn't RED's design allow getting good audio footage w/o a sound man?

Another thing to remember about audio devices, they make noise. So the fewer items chained together the better the audio. All sorts of boxes and cords are a nightmare and evidence of poor design.

 

[LEON]

Here's a question: how difficult would it be to actually add a limiter/AGC circuit to the camera? I'm assuming it would call for a hardware modification to the main board in the camera and some minor firmware tweaks

.

You can have a software limiter YOU won't even hear and you do not need hardware modification.

 

[S. Um]

You can have a software limiter YOU won't even hear and you do not need hardware modification.

A software limiter works well only if you have good data to start out with. If the data is clipped at the A-D converter, then there's no way to recover it. I think Red would have to implement a hardware fix for this.

But come to think of it, if they build in some headroom at the A-D converter stage (say 6dB), then they might be able to do some limiting in software. I think they do the same thing (extra headroom) with the video A-D converters.

 

[Elizabeth Lowrey]

Not really on topic, but I shot for several years on a VX2000, and it does indeed have fully manual exposure and manual audio gain, both are selectable options. The audio quality is nothing to write home about, though. There is actually a guy who offers a rework service bypassing some of the noisy audio chips used in the Sony VX2k and the earlier TRV900 cameras.

I borrowed it a couple of times for a multicamera shoot where I needed a static third camera, and for the life of me I could not find any setting in any menu that allowed me to turn the video gain compensation off. I don't know if turning auto iris off would have made any difference, but I HAD to use auto iris as it was an unmanned camera. I just didn't want auto video gain to go with it. Every time house lights went down the video noise just bloomed like a nasty weed.

 

[Elizabeth Lowrey]

BTW, a compressor and a limiter actually perform similar functions. With a compressor, you set a threshold or knee (where the compression kicks in), and a compression ratio. With a limiter, the threshold is usually set just under the limiting level, and the ratio is usually very high. So essentially, a limiter is a type of compressor.

Exactly. As I stated earlier, a limiter is a uni-directional compressor, a less drastic tool for accomplishing the same goal as a compressor: reduction of dynamic range.

I have never used the AGC in the DVX because I assumed it was . . . well . . . an AGC and not a limiter and also because I truly haven't felt the necessity of using it for the staged interviews and live board feeds that I run to it on most jobs. Knowing that it functions as a limiter only might make me less wary of using it for interviews in the future.

 

[Steve Sherrick]

This was a hot topic early in the camera's development. I pushed for a few priorities. Clean Mic Pres, 48V phantom, physical gain controls on the camera, and high quality limiters. Also analog to digital conversion was discussed. I have not received a camera to test with yet, so I can't comment on any of the audio implementations yet. However, I think lack of gain control will be potentially addressed with the supergrip and other 3rd party gear. I still would have put it on the camera. I pushed for recessed pots, and all you needed was two of them with a small switch that could be used to select which channels you are controlling. Red went the other way, but they may have had reasons for this, so again if they can come up with a good solution, I'm okay with that. But omitting limiters will pose problems for a segment of their user base. Some serious tests are now needed to see just how good the audio circuitry is in the camera. I will go out on a limb here and venture a guess that the Red team had some discussions with some quality audio vendors or consultants and they used components that achieve better than average audio recording within a price point that makes sense in the context of the camera's overall price. In other words, I expect clean audio recording, but not on the level of a high end digital audio recorder.

So, let's give Red the benefit of the doubt here and run some tests over the next month and get a baseline for what the camera can and can't do in terms of audio and then we can forward our thoughts to the Red Team for improvements.

Steve

 

[S. Um]

I have never used the AGC in the DVX because I assumed it was . . . well . . . an AGC and not a limiter and also because I truly haven't felt the necessity of using it for the staged interviews and live board feeds that I run to it on most jobs. Knowing that it functions as a limiter only might make me less wary of using it for interviews in the future.

Why not give it a try, especially if you connect your microphones directly to the camera. If your audio doesn't get too hot, then the ALC doesn't kick in at all. If the audio does get hot, then the limiter helps keep it from clipping. It's a win-win situation.

 

[S. Um]

So, let's give Red the benefit of the doubt here and run some tests over the next month and get a baseline for what the camera can and can't do in terms of audio and then we can forward our thoughts to the Red Team for improvements.

Steve

Good suggestion. There are a bunch of smart people at Red, and I'm sure they've given this lots of thought. If the camera has 24-bit A-D converters, then they probably have a strategy to get good audio. Let's see/hear some test results.

 

[philper]

Exactly. As I stated earlier, a limiter is a uni-directional compressor, a less drastic tool for accomplishing the same goal as a compressor: reduction of dynamic range.

I have never used the AGC in the DVX because I assumed it was . . . well . . . an AGC and not a limiter and also because I truly haven't felt the necessity of using it for the staged interviews and live board feeds that I run to it on most jobs. Knowing that it functions as a limiter only might make me less wary of using it for interviews in the future.

A limiter is not a compressor. Please look more closely at the functions of these tools, how they operate and what those terms mean in the audio world. Again, a limiter only functions on the peaks of a signal, while a compressor works on the whole mix according to its settings, which look at the average level of the mix. There are places where the functions blur together, but only in the most sophisticated of studio processors, not on inexpensive video camcorders. An AGC is a different animal--it is triggered like a limiter (in response to peaks in the mix) but functions like a crude volume control w/ no frequency or level response slope like a real compressor. One can say that an AGC "compresses" audio, but that is a misleading expression since the resulting effect on the sound is very different from that of a limiter or a compressor, esp. if properly deployed.

I don't like the sound of AGCs, and don't know anyone who does. That said, they have their place, mostly in usage by shooters struggling to get anything usable in unpredictable situations. Better cameras (in my opinion) have AGCs which really are peak limiters (higher-end Sony cameras for instance) that don't sound half-bad in certain situations. In any case, relying on an automatic device of any kind will almost always produce lower-quality audio that will have to be more massaged in post, and this is especially true w/ AGCs, or limiters if over used. While not being able to completely verify this, I suspect that many video camcorders at all price levels have some degree of AGC or limiter that is not user-controllable. I say this because even when the controllable onboard AGC/limiters are set to off I often hear limiting going on in my return from the camera that is not there in the mixer feed. This may be something they do to keep big audio peaks from affecting the picture--that was the opinion of at least one knowledgable field engineer I asked about this.

Philip Perkins CAS

 

[philper]

This was a hot topic early in the camera's development. I pushed for a few priorities. Clean Mic Pres, 48V phantom, physical gain controls on the camera, and high quality limiters. Also analog to digital conversion was discussed. I have not received a camera to test with yet, so I can't comment on any of the audio implementations yet. However, I think lack of gain control will be potentially addressed with the supergrip and other 3rd party gear. I still would have put it on the camera. I pushed for recessed pots, and all you needed was two of them with a small switch that could be used to select which channels you are controlling. Red went the other way, but they may have had reasons for this, so again if they can come up with a good solution, I'm okay with that. But omitting limiters will pose problems for a segment of their user base. Some serious tests are now needed to see just how good the audio circuitry is in the camera. I will go out on a limb here and venture a guess that the Red team had some discussions with some quality audio vendors or consultants and they used components that achieve better than average audio recording within a price point that makes sense in the context of the camera's overall price. In other words, I expect clean audio recording, but not on the level of a high end digital audio recorder.

So, let's give Red the benefit of the doubt here and run some tests over the next month and get a baseline for what the camera can and can't do in terms of audio and then we can forward our thoughts to the Red Team for improvements.

Steve

As a soundman, I would say that if it can record audio as well as an F900 etc then they've done their job. I'm not thrilled with the sound of those cameras and their competitors, but it gets the job done most of the time. If higher fidelity is needed then the production should be recording double-system. It is extremely difficult to get clean preamplification and digital conversion of a low-level signal like a mic feed done within a device that is also doing a lot of high speed data processing, as the Red is. In audio we avoid sending low level analog signals (or even line level ones) directly into computers--there is too much hash, RF, grounding issues etc etc to get a really clean recording. That Sony, and presumeably Red, can get a decent quality audio signal recorded that matches a 16 bit DAT fairly well is a great accomplishment to have done within such a small box, and will serve most users quite well I think.

Philip Perkins CAS

 

[DNomer]

Audio Preamps, etc.

Audio Preamps, etc.

My best wishes to you all in what promises to be a fine new year for RED users.

This was a hot topic early in the camera's development. I pushed for a few priorities. Clean Mic Pres, 48V phantom, physical gain controls on the camera, and high quality limiters.
Steve

I believe that Steve has a good take on this. The weak link in location audio is the microphone, its cable, and how it is held (not counting aspects of the location itself). You absolutely must have a balanced XLR input to even be in the game, unless perhaps nowadays you might digitize outboard and run the audio in digitally. To get every last dB of signal to noise, you want very clean 48 volt phantom power, so you can power a very nice amp that exists in the (condenser) microphone itself, giving you a nice high level output from the mic.

However, a problem comes when the signal enters the camera: you have a very noisy environment, with gigahertz signals radiating RF all over and power and ground lines are getting slammed a lot. So special shielding has to be done to protect the delicate analog audio signal and its chips up to the A/D. You have to use good mic preamps, and the limiter has to be BEFORE the A/D, in the analog part. You want a nice soft limiter to get rid of that gawdawful sound of a digital clip. A good limiter will cut in only near the clipping point, and it will INCREASE its gain reduction as the signal gets closer to clipping, to give you that nice 'soft' limit. The wizardry of modern op amps (used in the preamp and limiter) is one of the really amazing things that we get to take advantage of these days. Since the true signal to noise of the analog part of things will not exceed 16 bits even in an ideal situation, talking about 24 bits is mostly just fluff. Storage and processing of 24 bit audio is fine; just don't think that you really have 24 bits of data.

Re: AGC, I am amazed that manufacturers have the nuts to install them even in consumer cameras. The way that an AGC boosts the signal when no one is talking is nothing short of ruinous. The basic concept is fundamentally flawed.

Re: noise reduction in post, I have been very impressed with the ability of modern noise-reduction software to get rid of hiss-type noise and other relatively constant noises, but realize that if you use that, you have destroyed the ambient sound of your recording, and you will have to add it back in, using a separate recording of 'room tone', to avoid a sterile location sound. When you add in room tone later, you are now 'faking it' a little, sort of like using Photoshop to drop a nice sky behind a still shot with a grey sky. Run and gun people, who need every help they can get, often cannot replicate the ambient sound later like pros on a set. The ambient sound may be very jumbled and varying, with people talking all over, etc.

The difficulties of getting quality location sound with proper ambience is why pros will use a really top notch short shotgun in a state of the art mount and boompole, held by a boom-op who can aim it just right and not add handling noise. A good camera must support THAT, as mentioned above. In many situations, there is no need for dual system sound, in my view.

Re: restoring clipped audio, I have been very impressed that some software can do a tolerable job of dealing with this. Of course, it does not truly recreate the lost data, but it tries to reconstruct the lost peaks, including the various frequency bands that were riding on those peaks. Obviously, you are far better off if you can avoid clipping.

I would like to say that I am very appreciative of Philip's input in this thread. The way that I know that he knows his stuff is not because of 'CAS', but because of his adroit attention to a number of important issues, always touching on just the right points to elucidate some aspect of what we need. There are several other fine contributors, as well, and Philip may tend to emphasize the 'high end' perspective, but we can surely benefit from quality input from people with hard-won field experience who know what it takes to get good location audio, despite a whole slew of situational problems that actually occur.

Dennis Nomer

 

[Alexis Hanawalt]

For terminology and theory, I don't think it gets better than this site:

http://www.filmsound.org/

The Apple support document for audio filters in Soundtrack Pro is also very helpful and concise.

...However, I'm very interested in a lot of the discussion Phillip has brought up regarding the highly technical issues of different audio processing situations... sound floors, cpu noise... these are things I know little about and do not have an ear for. For me, 24bit audio is 24bit audio, however it's been recorded - whether in camera or on a DAT. Phillip, in your wisdom, what are the primary concerns with regards to how audio is processed and recorded (limiters, etc. aside.) What might one sacrifice recording into an F900, for instance, and what would one gain in fidelity from recording dual system - or adding a mixer to the chain before the audio gets to camera?

(This has already been a very educational thread - thanks to all for contributing to the discussion.)

 

[10s]

Nice contribution Dennis, and I would add the area of greatest challenge once the electronics are worked out is microphone placement: where the mic is placed in consideration of the source, reflections, secondary sources and microphone characteristics. The EMI studio engineers of the 50s and 60s were very good at this as well as many here in the USA. There are some incredible masters of the microphone out there.

Side note: I bought the original DVX100 and was pleasantly surprised with it's sonic ability. So a while back I used someone's DVX100A model and the audio was echo-ey. I thought there must be something wrong with this, but was then informed that the later models (A & B) were made this way so video shooters would know their camera was getting a signal. Is that correct? To me it seems that Panasonic should have suggested operators get better headphones and learn how to listen, because this new echo feed to the headphones distroyed any sort of ability to correctly place the mic and navigate reflective surfaces, etc. So miking with this handicap severly reduced any sort of effective mic technique. If anyone knows more about these A & B models, PM me and let me know. I just may not have understood it's operation correctly.

Moral of the Story, tools need to be designed for intelligent users, not hacks.
I'm hoping RED has the audio right. They done a great job so far, and we all want excellent cameras so we can tell our stories as well as the big boys tell theirs.

PS: as far as the noise inherent in the analog realm in a camera, get to digital ASAP. Every box, cord, tinker toy, etc is just another chance to collect more noise. Direct simplicity goes along way. I thought I read where someone, DPA or some other quality company, is now making a mic with decent A/D conversion in the mic, that would be interesting!

 

[philper]

For terminology and theory, I don't think it gets better than this site:

http://www.filmsound.org/

The Apple support document for audio filters in Soundtrack Pro is also very helpful and concise.

...However, I'm very interested in a lot of the discussion Phillip has brought up regarding the highly technical issues of different audio processing situations... sound floors, cpu noise... these are things I know little about and do not have an ear for. For me, 24bit audio is 24bit audio, however it's been recorded - whether in camera or on a DAT. Phillip, in your wisdom, what are the primary concerns with regards to how audio is processed and recorded (limiters, etc. aside.) What might one sacrifice recording into an F900, for instance, and what would one gain in fidelity from recording dual system - or adding a mixer to the chain before the audio gets to camera?

(This has already been a very educational thread - thanks to all for contributing to the discussion.)

The "filters" that provide noise reduction etc in SoundTrack Pro are about in a par with what comes in other low cost audio programs like Sound Forge, Audition etc. They are not on a par with higher cost plugs available for ProTools etc or dedicated outboard systems like Cedar. What I mean by this is that they can do less work for you reducing noise and distortion before the artifacts they cause in the resulting audio file start to make the whole exercise not worth it. It is VERY easy to improve a piece of noisy audio into a ruin with these tools.

The fact that an audio file is 24 bit tells you nothing about how it really sounds. I can record 24 bit via the analog audio input of my laptop or on a low-cost file recorder and end up with audio that is inferior in every way to that recorded by a used DAT machine @ 16 bit. The bit width is like a videotape format--just a container. You can record very nice looking video on DVCAM and you can record unwatchable crap with a low end camera--the format is the same. As I've mentioned, there is not yet available a 24 bit A-to-D that gives you a real 24 bits of resolution, even in the ultra high end audio world. And as I've also opined, the audible difference between a good 16 bit recording of most field production audio and a 24 bit recording of the same scene would be nil. Why? 1: The background noise floor of this noisy world we live in. 2: the analog audio electronics in all video camcorders and most audio recorders under $2000 or so have so much noise and distortion themselves that the added resolution is wasted--the only real difference in a 16 bit and a 24 bit recording they'd make is that the 24 bit one would take up more storage. (By the way, there are no 24 bit DAT machines--the DAT spec is 16 bit.)

For taking the signal of a microphone and recording it on a camcorder (F900 was mentioned), using an external mixer makes for an audibly better recording. There are two major reasons for this: first, the preamps in even a small mixer like a SD 302 or even a Shure FP33 are far cleaner and lower distortion that those in the F900, and there is the previously discussed matter of low-level analog audio amplification going on within the chassis of a device also doing high speed digital image processing--the proximity to those circuits is a real problem. The control of the audio signal in the mixer is more precise, low roll-off filters are available and easily auditioned, the headphone circuit of the mixer is much cleaner and capable of higher levels w/o distortion. Metering is immediately available. Dynamics control is usually included, either on output (FP33) or both (302, 442). The 2nd reason for using a mixer is that having worked the audio as discussed within the mixer, the user can now feed it to the camera as a much more easily recordable signal, at a good level (stay out of the noise floor mud), without big peaks (momentary distortion) and with a way, via a return feed, to immediately evaluate if there is a problem with the mix feed or how the camera is dealing w/ the sound by switching between the two on the mixer. Recording a mic directly to the camera will always have its place on certain sorts of shoots, but it will never make for as high fidelity and easily controlled and monitored track as one done via a mixer.

Philip Perkins CAS

 

[brandon herman]

I bought the original DVX100 and was pleasantly surprised with it's sonic ability. So a while back I used someone's DVX100A model and the audio was echo-ey. I thought there must be something wrong with this, but was then informed that the later models (A & B) were made this way so video shooters would know their camera was getting a signal.

I'm pretty sure this was a matter of changing the 'headphone out' signal from the record heads to the playback heads a tiny bit further down the tape track, so that you're hearing a very slight audio delay. You can change this in the menu. Not sure why they did that, though.

I think.