Planning a better compressor/limiter for home theatre listening

Seems a familiar complaint. Watch a movie at home these days, you have to turn the volume way up to hear the quiet dialog, only to be blown out of your seats when the sudden action scene starts. So I'm thinking of adding a stereo compressor/limiter. I've experimented with several compressor circuits, and my all time favorite "active" element for music is a home made LDR, typically using a LED and cadmium cell. It "attacks" quickly but not immediately, and has a long 'release", good for both music and live action. But I have two concerns I'd like some "philosophical" guidance on. Please tell me if my thinking is correct, or if not how I should think differently.

1) I'm thinking that since a complex audio signal sounds much louder than most pure tone frequencies, it might be better to use an incandescent lamp in my LDR, so it will better respond to RMS levels rather than peaks?

2) Since the phase and level differences between Right and Left signals is crucial to deriving "surround" channels, I'm wondering whether the control feed-back I apply to the left and right LDRs should be an additive combination of the L and R signals, or should remain totally independent?

3) If neither option in item #2 will properly preserve surround channels, what else can i do? I would just build more independent compressor circuits for the rear (and center) channels, but the only signals I have available from the source are L and R. (In other words, I can't easily get to the point between surround-decoding and the amplifiers, without totally hacking the amplifier/ receiver, and I'd hate to build something that wasn't a standalone box).

Thanks for any ideas
 
Throwing out dynamic range is always last resortish. It's unfortunate how un-hi-fi a lot of AV equipment is, even to the point of ruining low-level intelligibility. But, better than 90dB dynamic range is not the problem if you have a refrigerator in the next room cranking out 45. Certainly start by minimizing your noise floor. You can always grab a decent commercial compressor (most with every parameter in question adjustable) and rebay it when you're done finding out what you need. I know that's not strictly diy but taking your own path might be more fun /easier once you know which way you're going.
 
First, If you have a PC running Windows I would suggest you take a look at the AC3 Filter. This is a swiss army knife audio management tool that decodes up to 7.1 audio and provides a surprisingly good Dynamic Range Control (Compressor/Expander) that may answer many of your questions about the best methods. I use this in my HTPC builds for exactly what you are talking about. Stuff in a BluRay with AC3 Filter running and no more mad dives for the volume control at 1:00am.

Movie sound is not simple. It is recorded in bands that are generally at different levels, and it is split into multiple independent channels. Managing such a scheme is not simple.

Foley sounds, the rustling leaves, the click of heels on pavement, traffic noise etc, is generally recorded at -30db which is very quiet.​
Dialog, people talking is generally recorded between -20 and -10 db. Most conversational dialog runs around -20, yelling and screaming can reach -10.​
Music, typically runs between -15 and -5 db, with some background music recorded in the foley range.​
Effects such as gunshots, explosions, airplane noise, etc. can reach +6db which is about a half a hair short of digital clipping.​

It should be obvious that compressing this is not a simple task. Do you want to just make the dialog louder? Do you want to control the entire multi-channel matrix? Do you want to control the channels individually or manage the matrix as a whole? Each of these decisions will affect your design.

In any event you are probably looking at +20db to -6db of compression to level the sound adequately while leaving some dynamic range between dialog and effects.

If your audio output is stereo, this simplifies things quite a bit. You would get between the stereo output of your TV or Decoder and your pre-amp and compress both channels simultaneously from a control signal that is summed from both channels. This would let you maintain the stereo image and sound balance without weird effects like people talking off the side of the screen or one channel appearing to go dead for a second or two.

For surround playback, it gets a lot more complex. Unless you are only concerned with dialog, you need a control signal that is the sum of all channels and you will need to manage all of them simultaneously to maintain the balance between channels... not an easy task.

The truth is that this kind of sound management is far easier in the digital/software domain than in a hardware device. The multitude of options... Attack time, Hold time, Decay rate, Makeup Gain, Matrix Balancing, Clipping control, etc. are all far easier to manage digitally, in software, than in a circuit which is probably why most AV decoders and TV Sets have a "night mode" or "Automatic Volume Control" function built in.

Also, I'm not sure that a Light Dependent Resistor is going to be your best answer here. You may find yourself having to resort to multiple op-amps with gain control circuitry to accomplish this task.

Best of luck with the project and I hope this has been some help to you.
 
Last edited:
First, If you have a PC running Windows I would suggest you take a look at the AC3 Filter.

Hmmm.. there is a pretty robust Windows computer in the TV area, in fact we use it for movies sometimes. I wonder though where running sound in and through the PC, through a digital filer, and then out, would cause significant delay. I know in my music composition machine, I have a good sound-cared with ASIO drivers which limits most latency to about 6mS... not bad. But I think the PC in our listening area has stock mo-bo sound hardware, is likely driven by standard WDM drivers. I have a feeling it might cause enough latency to make watching a movie awkward. But I'll look into it...thanks!

It should be obvious that compressing this is not a simple task. Do you want to just make the dialog louder? Do you want to control the entire multi-channel matrix? Do you want to control the channels individually or manage the matrix as a whole? Each of these decisions will affect your design.

And thanks for addressing my concern... that the feedback/control path for the compression will have to be a combination of all channels. Especially if (as I mentioned) the compression will have to be done on the stereo signal, before surround sound decoding.

Originally I though I might just design as a limiter... in other words, something that nearly flattened the output during those sci-fi "armegeddon" level action events. But I'm thinking it might be better to strive for an even / overall compression ratio that spans the audio range, maybe with a drop-out point to avoid amplifying noise during silent periods.

In any event you are probably looking at +20db to -10db of compression to level the sound adequately while leaving some dynamic range between dialog and effects.

That is a reasonable starting point.


The truth is that this kind of sound management is far easier in the digital/software domain than in a hardware device. The multitude of options... Attack time, Hold time, Decay rate, Makeup Gain, Matrix Balancing, Clipping control, etc. are all far easier to manage digitally, in software, than in a circuit which is probably why most AV decoders and TV Sets have a "night mode" or "Automatic Volume Control" function built in.

But a fun project just the same. :)

Also, I'm not sure that a Light Dependent Resistor is going to be your best answer here. You may find yourself having to resort to multiple op-amps and gain control circuitry to accomplish this task.

Best of luck with the project and I hope this has been some help to you.

Thanks. Well having experimented with several FET based compression elements, as well as a few of those unwieldy trasconductance amplifier ICs, I've had some surprisingly good results with LDRs... which I might add was an idea I laughed at the first time I saw it. For musical instruments (especially a guitar) they provide a near perfect response, at least to my ears. But as you point out, movie audio may not work as well. And of course the trouble with LDR based compressors is that unlike a FET (or other approach), if you DON'T like the response, its much more difficult to change.

Thanks again! Just the kind of awesome informative response I've always loved about this forum!
 
Throwing out dynamic range is always last resortish. It's unfortunate how un-hi-fi a lot of AV equipment is, even to the point of ruining low-level intelligibility. But, better than 90dB dynamic range is not the problem if you have a refrigerator in the next room cranking out 45. Certainly start by minimizing your noise floor. You can always grab a decent commercial compressor (most with every parameter in question adjustable) and rebay it when you're done finding out what you need. I know that's not strictly diy but taking your own path might be more fun /easier once you know which way you're going.

Well yes, that and the fact that there are other background sounds (such as the monstrous whole-house Air conditioners we've grown used to dealing with in Florida. But I think the problem is simply that movie audio these days is trying to offer you that "theatre" experience, where the dialog level is "normal", and they assume you (as a theater goer) expect action scenes to be as realistic as "live". But the truth is, when listening at home, you usually don't want a gun being shot from 5 feet away to actually sound like a real gun right there in your living room. And the truth is, having a good sound system with plenty of headroom and big powerful speakers is part of my problem. Perhaps movie audio is "expanded" to account for the modern small speaker systems many people have these days. As a musician, I've always opted for more serious amplification and speakers, which probably is working against me now! LOL!
 
Who knows what cinema sound engineers are designing for. 7 out of 10 times I've been in a comercial theater the diolog was loud enough to take out your ears and there was no sub bass. They also had high velociity climate control that was hard to talk over before the movie even started. No wonder home theater became such the thing that it is...
 
Hmmm.. there is a pretty robust Windows computer in the TV area, in fact we use it for movies sometimes. I wonder though where running sound in and through the PC, through a digital filer, and then out, would cause significant delay. I know in my music composition machine, I have a good sound-cared with ASIO drivers which limits most latency to about 6mS... not bad. But I think the PC in our listening area has stock mo-bo sound hardware, is likely driven by standard WDM drivers. I have a feeling it might cause enough latency to make watching a movie awkward. But I'll look into it...thanks!

You would need to play the movie in the computer to get the full benefit. This is why I went the HTPC route. 700+ movies, 5000+ songs... all stored on the hard drive :D

The audio delay is adjustable in the filter. You can lag and lead by quite a bit to get it right. I'm lucky, mine is spot on.

What I generally do is install Media Player Classic then load AC3 Filter in it, assigning it to handle files with the movie extensions (mp4, avi, etc.) Makes the whole thing real simple. Just click on the file and MPC does the rest for you.

And thanks for addressing my concern... that the feedback/control path for the compression will have to be a combination of all channels. Especially if (as I mentioned) the compression will have to be done on the stereo signal, before surround sound decoding.

Isn't that the other way around... the file/disk has multi-channel and you're crunching it down to stereo? (or have I misunderstood?)

Originally I though I might just design as a limiter... in other words, something that nearly flattened the output during those sci-fi "armegeddon" level action events. But I'm thinking it might be better to strive for an even / overall compression ratio that spans the audio range, maybe with a drop-out point to avoid amplifying noise during silent periods.

That's not a bad idea and in many cases that's all these "night modes" really are. True compression would be unnoticeable unless you did an A-B comparison.

Thanks. Well having experimented with several FET based compression elements, as well as a few of those unwieldy trasconductance amplifier ICs, I've had some surprisingly good results with LDRs... which I might add was an idea I laughed at the first time I saw it.

My worry would be a somewhat slow response time... every gunshot would have a much louder click at the beginning and then everything would drop in volume, instead of the gain riding along on the waveform as it should.

Thanks again! Just the kind of awesome informative response I've always loved about this forum!

No worries...
 

billshurv

Member
Paid Member
2014-03-01 11:53 pm
It sounds like you just need to set the centre channel up correctly. You should be able to adjust dialogue relative to Fx so you get the balance you need. And I agree that THX levels can be a little high for late night listening, but that is close to what is considered a 'natural level' for live classical music. It's interesting to note that some people find a full dynamic range too high even for high fidelity listening.
 
It sounds like you just need to set the centre channel up correctly. You should be able to adjust dialogue relative to Fx so you get the balance you need. And I agree that THX levels can be a little high for late night listening, but that is close to what is considered a 'natural level' for live classical music. It's interesting to note that some people find a full dynamic range too high even for high fidelity listening.

Well my receiver, when set to the surround settings, does offer a center level adjustment. But it only works when you have an actual center speaker connected. I'm currently using its "phantom" center speaker setting (no actual speaker), and it works fine for most situations. But strangely, the level adjustment does not actually do anything in 'phantom" mode. Perhaps I'll set up an actual center speaker, if only to see how much it assists the problem. I suspect however, that it won't help a great deal. Some leveling or compression will almost surely be needed.
 
Isn't that the other way around... the file/disk has multi-channel and you're crunching it down to stereo? (or have I misunderstood?)

No worries...

What i mean is that the output from the TV going to the amplifier is a simple L/R stereo signal. Inside the amplifier, the surround channels are decoded and routed to separate power amplifiers, and then to external speaker connections. So the stereo pair is the only place where I could put a pair of DIY compressors. What I was referring to was the suggestion (which I had asked about and you concurred) that the feedback path controlling the output level from each individual compressor circuit should be a summation of the two available (L + R) channels. This would ensure that if, for example, a gunshot was supposed to appear on the right, both channels would be affected.

AND... I'm hoping that by ensuring that the now compressed L and R signals have been processed this way, it will not adversely affect the surround channel decoding process.

Hard to explain this stuff. ;-)
 
My post should have read dialogue level and not dialogue lift (which is used with presence speakers for centre height).

Had a look at my AV Receiver and there is a dialogue level adjustment as well as dynamic range options from max to min (sometimes called night mode).

I do not use a centre channel but use the dialogue level 0-3 settings and I use 1.
 
What i mean is that the output from the TV going to the amplifier is a simple L/R stereo signal. Inside the amplifier, the surround channels are decoded and routed to separate power amplifiers, and then to external speaker connections.

Trust me on this one... if you are working with an analog stereo signal, you don't have any real surround sound going on. What you have is "speaker fill" and while it can sound pretty good, it's not true surround sound.

The good news is that in your case this is actually an advantage since you don't have to design a 6 or 8 channel compressor.

So the stereo pair is the only place where I could put a pair of DIY compressors. What I was referring to was the suggestion (which I had asked about and you concurred) that the feedback path controlling the output level from each individual compressor circuit should be a summation of the two available (L + R) channels. This would ensure that if, for example, a gunshot was supposed to appear on the right, both channels would be affected.

Yep, that's how I would do it.

But first I would check in my TV's menus to see if it has a sound leveling option built in. It might be a simple matter of enabling it.

AND... I'm hoping that by ensuring that the now compressed L and R signals have been processed this way, it will not adversely affect the surround channel decoding process.

Hard to explain this stuff. ;-)

If you are working with a stereo analog signal, it wouldn't likely mess anything up.

Now if you were working with an HDMI feed from the TV to the amplifier, that would be a whole horse of a different colour.
 
Last edited:
Trust me on this one... if you are working with an analog stereo signal, you don't have any real surround sound going on. What you have is "speaker fill" and while it can sound pretty good, it's not true surround sound.

Well I can say that my receiver/amplifier accepts an L/R input, and has a Dolby Surround sound setting. When that setting is enabled, you can clearly hear when sound is being directed to the rear speakers, more than the Left, Right or center. As I understand it, this "surround decoding" is accomplished by amplifying phase differences (for rear), and similarities (for center) between the L and R signals. Thats a simplification, but I fully understand that this is a far cry from actual separate channels. For example, you can't HAVE a phase difference between L and R unless there is audio present in both, so there is no such thing as REAR audio with zero output on the L and R speakers.

But that said, I still would like to preserve that surround decoding process, as limited as it may be. And this is really the central issue I'm wrestling with mentally. Wrestling this way usually means I'm in one of those "no perfect solution" situations, which is why I'm asking here.

You said...

The good news is that in your case this is actually an advantage since you don't have to design a 6 or 8 channel compressor.

Well truth be told, even if I only had a 4 or 5 channel source, it would be better. If that were the case, I could simply buy a couple of those old Behringer MDX1200 two channel compressors (which you can now get on ebay for like $20 USD), connect them all up and adjust them similarly.

But since I just have 2 channels in which surround is "encoded", the thing I'm still wrestling with is whether the two compressors should be fully independent, OR should they share a single feedback network (for I guess what you'd call the "side chain"?), which combines the L and R signals, and applies the mix to two VERY evenly matched gain control elements.

That scheme would SEEM to best preserve the phase-encoded surround information within the signal. After all, it is most equivalent to the human reaction or grabbing the remote and turning down the volume of whole system in response to a loud action scene.

But on the other hand, would that scheme work against a simple loud stereo effect, say a fire engine moving across the sound field from left to right. If the compressors were independent, it might just localize the volume reduction to where it is needed... initially just on the left, followed by reducing the whole sound field (middle) and finally just reducing the right side.

I have a feeling I'm going to have build for both cases, with a master switch to select between independent or combined compression response. And I further have the feeling that in time I'll discover that neither setting is best for ALL cases.

And sadly, my TV's settings for limiting dynamic range don't seem to be doing their job at all.

Oh, and you said...

If you are working with a stereo analog signal, it wouldn't likely mess anything up.

Now if you were working with an HDMI feed from the TV to the amplifier, that would be a whole horse of a different colour.

Oh you just hit on another major pet peeve of mine. My cable box, The DVD player, the Firestick and others, all connect to the TV via separate HDMI ports, and the audio is routed from the TV's L and R outputs to the amplifier. My peeve is that the baseline audio level from all of the HDMI sources is vastly different. And unlike the "old days" of separate analog video and audio connections, there is no way to simply insert a a box with a dual potentiometer to set them all equal. It would be nice if the TV had separate volume settings for each item, or ar least if each item had its own volume settings. NOPE and NOPE! :-(
 
My post should have read dialogue level and not dialogue lift (which is used with presence speakers for centre height).

Had a look at my AV Receiver and there is a dialogue level adjustment as well as dynamic range options from max to min (sometimes called night mode).

I do not use a centre channel but use the dialogue level 0-3 settings and I use 1.

Indeed, as much as I love DIY analog solutions, I may need to consider it time to just buy a new receiver.. Considering decent amplifier/receivers for home listening have not changes a lot in features or audio quality for many years, I typically look in local "flea market" distributors for such things. But live and learn when it comes to what features to look for! When I bought my current Pioneer unit, I looked at the reliable brand, output power, EQ, and for the dolby surround sound decoder capabilities. But at the time I never have thought to look for dynamic range options! And of course since then, the dynamic range of both streamed and disk based content seems to have at least doubled!

And... I'd certainly like to think the engineers who make receivers with dynamic range options have already considered the concerns I've described, and further would like to hope they are more sure of the right way to do it than I am! LOL!