Why bass management makes my life tedious

[Soundmixer]

Well, you wanted to see evidence. I provided it.

Pretty thin evidence, and certainly evidence without context.

And no, it's not "very weak spikes". The graphs span the entire running time. LFE sounds in over 2 hours worth of data naturally show up as spikes. There is highly audible higher frequency content throughout the whole movie because no LPF was applied to the LFE. That's what the spectrograms show.

First, we don't listen to the LFE in isolation, so your claim of audibility is questionable. There is always content playing in the other channels while the LFE is playing (it is an effects channel) so it is likely content playing in the other channels would mask any content in the LFE above 200hz including these short-term spikes. Based on that alone, I would still recommend turning off the LPF in the AVR because the chances of an LPF being applied at the encoding level is FAR higher than it not being done. Two filters applied over each other will certainly have phase issues, but there is certainly a tradeoff here. This graph does not show an LPF was not applied, but it certainly shows there is garbage in the LFE.

Two movies out of thousands of encodes, and short-term spikes in the LFE do not show an overall trend that warrants calling the fire department. It does show that garbage can get into the LFE channel, and I am not sure SMPTE or ITU(which does not govern this stuff anyway) would be concerned about this, and this is not evidence the sky is falling or the wolf is coming.

 

[Soundmixer]

I see quite a lot of content in the 100-400hz range too which is definitely going to be localizable and probably sound weird in the typical case of somebody with just a single sub.

Are you sure about this statement? Considering there will be content present in the main channels, there is no guarantee the information above 200hz will be audible or localizable. We don't listen to the LFE in isolation, the masking effect would certainly be in play here. The LPF of the Dolby Atmos encoder is 24db per octave above 120hz, so energy up to 240hz will certainly be present. Since there are only small spikes of energy above 240hz, it is likely that is garbage that can (but rarely) show up in the LFE channel. The huge spike is definitely an artifact, as there are no other spikes of that magnitude present elsewhere in the LFE.

Though I'm not sure why this conversation has spanned so many pages. The default 120hz LPF is no doubt in use in 99% of home theatres and if anybody adjusts it it's probably lowering it because they misunderstand the purpose.

You also have to consider that most folks that own that 99% of home theaters don't know there is a 120hz LPF already encoded in the LFE.

 

[Soundmixer]

This is very true and very unfortunate. Not only have I seen 5.1 music releases with full range content in the LFE channel, I've also seen plenty of releases that don't abide by the "+10 dB gain on playback" LFE convention used for movies, which means the LFE will play back way too loud on a system configured for HT. I am baffled as to why the multichannel music producers chose to ignore the well-established and well-defined movie standards. That makes it way harder to play their content correctly.

This is what happens when the industry does not adopt standards. The result is the wild wild west.

(I know nothing about the production side of things, but knowing what I know about the playback side, if I'm ever asked to produce 5.1 music, I know my first decision would be to ignore LFE altogether and keep it silent, leaving all the bass in the main channels. It's impossible to predict how the LFE channel will be rendered in a music playback system, which makes this channel a giant trap to be avoided at all costs. Leaving out LFE makes sense in music anyway, since "low frequency effects" seems like an awkward concept for that kind of content. It's not like you're going to have giant explosions in music.)

I totally agree with your statement here. For multichannel music, the LFE is definitely not needed. A lot of music mixers don't really know how to use the LFE, so ditching it is advisable.

Going back to movies: as others have pointed out in this thread, I also personally experienced "garbage" (i.e. >200 Hz) content in movie LFE channels. It's rare, but it does happen, and for that reason I make sure the LFE is always lowpassed in one way or another in my playback system. I also came across content where the LFE channel was literally just a lowpassed mix of the main channels, which is technically fine but begs the question of whether the producer understands what LFE stands for.

I would love to see how the LPF of the content interacts with the LPF of the AVR or processor. As I have said previously, two cascading filters can have phase consequences unless they have identical characteristics - and that is not certain when it comes to the AVR. Some receivers have 18db per octave LPF on the LFE, and others have 24db per octave. This means you will still see a lot of energy up to 200hz or so even if you engage the LPF on the AVR. If you look at the example above, most of the energy follows the LPF of the Dolby Atmos encoder, but clearly, there is some content (likely garbage) above that. Even if you engage the LPF in the AVR, the energy between 100-200 would still likely be present, but certainly, the spikes would disappear. The flip side is you could have phase issues between the two conflicting filters, which could mode or node between these two frequencies. I try to avoid that conflict, but there is a chance garbage could show up and that is the trade off. The likely hood of me hearing that garbage is pretty slim though.

 

[krabapple]

I see quite a lot of content in the 100-400hz range too which is definitely going to be localizable and probably sound weird in the typical case of somebody with just a single sub. Though I'm not sure why this conversation has spanned so many pages. The default 120hz LPF is no doubt in use in 99% of home theatres and if anybody adjusts it it's probably lowering it because they misunderstand the purpose.

Whether it it localizable is going to depend on (in no particular order)
- level of LFE
- output level setting of subwoofer
- low pass filters setting(s) on the subwoofer (LPF for LFE in the AVR, 'crossover' in the sub)

Lowering the LPF for LFE or subwoofer crossover would not be wrong if we knew the high amplitude LFE content was well in excess of 80 Hz . It is dispiriting to contemplate that we might need to adjust these settings on a per-release basis, in order to hear what as intended, because of lack of standards at the production end.

 

[Sancus]

Are you sure about this statement? Considering there will be content present in the main channels, there is no guarantee the information above 200hz will be audible or localizable.

That's true, I wasn't considering masking. It is possible that it wouldn't be. It is hard to test though as I think 250hz is as high as that setting goes in most AVRs. Some don't even let you change it. I'm sure there are ways around that by not using an AVR, but frankly I'm not willing to put in that kind of effort, lol.

I wonder if they actually tested those movies without any LPF in place?

Lowering the LPF for LFE or subwoofer crossover would not be wrong if we knew the high amplitude LFE content was well in excess of 80 Hz . It is dispiriting to contemplate that we might need to adjust these settings on a per-release basis, in order to hear what as intended, because of lack of standards at the production end.

Content up to 120hz in the LFE is entirely normal though, and it is generally regarded as a bad idea to lower the LPF. The LPF isn't supposed to match your bass management crossover.

 

[krabapple]

Content up to 120hz in the LFE is entirely normal though, and it is generally regarded as a bad idea to lower the LPF. The LPF isn't supposed to match your bass management crossover.

Content up to 120 Hz, yes, but have you seen the actual frequency/amplitude curves?

 

[Sancus]

Content up to 120 Hz, yes, but have you seen the actual frequency/amplitude curves?

Sure, but considering it is the standard, at that point you're just second-guessing what the amplitude of the bass between 80-120hz is supposed to be. If a studio screwed up enough to leave too much bass in above 80hz in the LFE, I don't see how you're supposed to fix that by changing your LPF. You have absolutely no way to know what IS supposed to be there, after all. It's not possible for you to tune it in any meaningful way.

 

[Soundmixer]

Whether it it localizable is going to depend on (in no particular order)
- level of LFE
- output level setting of subwoofer
- low pass filters setting(s) on the subwoofer (LPF for LFE in the AVR, 'crossover' in the sub)

You would be correct if all we listen to was the LFE. But we don't listen to the LFE in isolation, it is accompanied by anywhere from 1-11 other channels output which would mask any output (and direction) above a certain threshold.

 

[Soundmixer]

That's true, I wasn't considering masking. It is possible that it wouldn't be. It is hard to test though as I think 250hz is as high as that setting goes in most AVRs. Some don't even let you change it. I'm sure there are ways around that by not using an AVR, but frankly I'm not willing to put in that kind of effort, lol.

I wonder if they actually tested those movies without any LPF in place?

I've watched both movies with the LPF turned up to 250hz. I could not hear any issues with the LFE, nor could I tell where the subwoofer was. This is why you have to provide context to these graphs. They tell how much content is in the channel (most of it located below 200hz) but it cannot tell you how audible that content when compared to the content in other channels. Presenting graphs in this fashion is misleading, and quite frankly disingenuous.


Most of the energy above 200hz is so low in level when compared to what is happening below 200hz, it is not likely to be audible. That huge spike wouldn't likely be audible either if the other channels were busy. It would all just blend together in the room. The encoder's LPF has a roll-off of 24db per octave at 120hz which means the filter is not a brick wall, and there will be substantial energy until about 200hz and a quick fall off above that. The examples above clearly show that characteristic, but there is clearly some hash above that which is not consistent throughout the movie.

 

[audio2920]

As I stated repeatedly (without much cranial penetration I might add), you cannot use the film industry's standards to judge how the music industry chooses to use the multichannel format. Their goals are different than ours, and they quite frankly don't have any real standards to follow. I have been very clear and consistent about this point, and quite frankly I am the messenger.

Sorry if you got the impression I disagree. I absolutely do not. My only contribution to the specific LFE discussion is that, personally in film mixes, I filter the LFE. What other people do is up to them and it's never bothered me in the end result. Maybe I'm not so discerning as some when it comes to my home playback system, "busman's holiday" and all that...

[For what it's worth, I did briefly get in to the multichannel SACD thing a while ago but I can't remember the last time I listened to music that way. Perhaps subconsciously there was LFE standardisation issue that helped drive me back to stereo presentation for music but if there was, it definitely wasn't a primary factor. My own living room system is probably pretty typical of non-audiophile setups in that it's just got a BM x-over in the AVR and and LPF set about an octave higher than that on the sub itself, plus a little "phase" shift (that seems to get it to integrate just fine). There's no calibration except by ear so I would never know if what I'm hearing is "right" or "wrong" ]

 

[ernestcarl]

I don't listen to enough concert DVDs and bluray audio discs... but just in case, I actually have two presets in my software media player (home theater PC) that allows me to either extend the LFE (to any satellites) or high pass it almost instantly with just a couple mouse clicks.

https://www.audiosciencereview.com/...do-you-use-to-implement-dsp.22659/post-755018

I've tested a few movies with known "garbage" unfiltered noise in the LFE -- quite disturbing to hear, actually -- and so as a preemptive measure I just high pass the stuff above 120Hz in all movies by default.

 

[Soundmixer]

Sorry if you got the impression I disagree.

Actually, I didn't. That comment was not directed at you.

My only contribution to the specific LFE discussion is that, personally in film mixes, I filter the LFE.

So do I, and so do the lion's share of re-recording, and mastering/encoding folks in the industry. My position is why filter it again in the AVR?

What other people do is up to them and it's never bothered me in the end result

As I said previously. You do the best mix you can and let the chips fall where they may. There is no way to bridge the sound quality gap between a soundbar in a living room with poor acoustics and a high-end home theater system in a dedicated room with treatment that is based on perceptual science.

 

[Soundmixer]

I've tested a few movies with known "garbage" unfiltered noise in the LFE -- quite disturbing to hear, actually -- and so as a preemptive measure I just high pass the stuff above 120Hz in all movies by default.

If you look at the typical roll-off characteristics of an AVR's LPF, even if you set it at 120hz, you will still have information up to 400hz in the channel at a much lower level. The Atmos encoder does offer a brick wall filter at 120hz (not the default as far as I know), which would mean you could set the LPF at any frequency above 120hz you want. My question to you is how do you hear the garbage in the LFE using bass management, and in the presence of the output of all of the other speakers? This is blended together either electrically, or acoustically. We don't listen to the LFE in isolation, so the masking of the bass from the main channels along with the midbass/midrange/high frequencies from the other speakers would surely have profound masking effect.

My only concern about applying a filter in the AVR is how it plays with the filter in the encoding system. It may have audible consequences, or it may not have any consequences at all. When it comes to filters, use only what you need...that is my motto.

 

[ernestcarl]

If you look at the typical roll-off characteristics of an AVR's LPF, even if you set it at 120hz, you will still have information up to 400hz in the channel at a much lower level. The Atmos encoder does offer a brick wall filter at 120hz (not the default as far as I know), which would mean you could set the LPF at any frequency above 120hz you want. My question to you is how do you hear the garbage in the LFE using bass management, and in the presence of the output of all of the other speakers? This is blended together either electrically, or acoustically. We don't listen to the LFE in isolation, so the masking of the bass from the main channels along with the midbass/midrange/high frequencies from the other speakers would surely have profound masking effect.

My only concern about applying a filter in the AVR is how it plays with the filter in the encoding system. It may have audible consequences, or it may not have any consequences at all. When it comes to filters, use only what you need...that is my motto.

No we do not listen to the LFE in isolation, but that does not mean that the "garbage" is always filtered out or masked over in all MCH films in my home theater:

example movie clip:
LFE GARBAGE MIDS TEST CLIP - THE DARK KNIGHT

Using my LFE "extended" preset/mode in JRiver, I can definitely hear crap noise above 120Hz perfecly crystal clear without soloing and/or muting any channels.

 

[Soundmixer]

No we do not listen to the LFE in isolation, but that does not mean that the "garbage" is always filtered out or masked over in all MCH films in my home theater:

example movie clip:
LFE GARBAGE MIDS TEST CLIP - THE DARK KNIGHT

Using my LFE "extended" preset/mode in JRiver, I can definitely hear crap noise above 120Hz perfecly crystal clear without soloing and/or muting any channels.

So, I went into both of my home theaters where the LPF on the LFE is set to 250hz and watched this scene. Now I am paid good money to listen to stuff like this, but I could not hear any "garbage" coming from the LFE when all of the action from the other channels were playing. Another question, how can you separate the bass from the mains going to the sub from the LFE going to the sub?

 

[ernestcarl]

So, I went into both of my home theaters where the LPF on the LFE is set to 250hz and watched this scene. Now I am paid good money to listen to stuff like this, but I could not hear any "garbage" coming from the LFE when all of the action from the other channels were playing. Another question, how can you separate the bass from the mains going to the sub from the LFE going to the sub?

I do not use a conventional AVR in the chain.

Ex. playback chain is something like the ff.

Netflix windows app > JRiver WDM driver and DSP processing > MCH DAC > active speakers


"LPF on the LFE is set to 250hz"

My LFE extended preset in JRiver means we assume the signal is left intact with (ostensibly) no effectively applied LPF. Flat magnitude 0 Hz -20 kHz LFE signal sweeps will play and measure flat at the listening position all the way. How is this even possible given most subs have their own internal HPF? Well, the LFE signal is cloned and crossovers are applied -- the cloned high passed LFE signal is sent to your preferred satellite monitor(s) -- e.g. center channel or front L+R mains, for example.

Just to illustrate:

"... how can you separate the bass from the mains going to the sub from the LFE going to the sub?"

There's more than one of way doing this, of course, but it is also pretty easy to do manually in JRiver:

 

[edechamps]

My LFE extended preset in JRiver effectively means we assume the signal is left intact with (ostensibly) no effectively applied LPF. Flat magnitude 0 Hz -20 kHz LFE signal sweeps will play and measure flat at the listening position all the way. How is this even possible given most subs have their own internal HPF? Well, the LFE signal is cloned and crossovers are applied -- the cloned high passed LFE signal is sent to your preferred satellite monitor(s) -- e.g. center channel or front L+R mains, for example.

My Genelec 7060B subwoofer has a feature like this in its bass management system, in which >85 Hz frequencies in the LFE channel are sent to the center channel where they can be reproduced full range. This is precisely the kind of feature I wouldn't use because it is bound to cause trouble with problematic content.

 

[ernestcarl]

My Genelec 7060B subwoofer has a feature like this in its bass management system, in which >85 Hz frequencies in the LFE channel are sent to the center channel where they can be reproduced full range. This is precisely the kind of feature I wouldn't use because it is bound to cause trouble with problematic content.

For the majority of content I would agree. There are some concert discs that shove full-range material (important bass guitar harmonics, for example) in the LFE. One may argue that it’s going to be masked anyway — maybe, or maybe not. But might as well have the option exist to check in case one is wondering. As you saw it’s so easy to mute all channels and quickly solo only the full LFE signal.

 

[Soundmixer]

I do not use a conventional AVR in the chain.

Ex. playback chain is something like the ff.

Netflix windows app > JRiver WDM driver and DSP processing > MCH DAC > active speakers


"LPF on the LFE is set to 250hz"

My LFE extended preset in JRiver means we assume the signal is left intact with (ostensibly) no effectively applied LPF. Flat magnitude 0 Hz -20 kHz LFE signal sweeps will play and measure flat at the listening position all the way. How is this even possible given most subs have their own internal HPF? Well, the LFE signal is cloned and crossovers are applied -- the cloned high passed LFE signal is sent to your preferred satellite monitor(s) -- e.g. center channel or front L+R mains, for example.

Just to illustrate:
View attachment 135728


"... how can you separate the bass from the mains going to the sub from the LFE going to the sub?"

There's more than one of way doing this, of course, but it is also pretty easy to do manually in JRiver:

View attachment 135726

Thank you for your explanation. First, the lion's share of folks do not listen through this kind of chain, and once again you use a single isolation example to prove your point. You muted channels, and by doing so, you eliminated their perceptual contribution to the in-room output. This is much like "Markus" example and does not provide context at all. Lastly, measurements in isolation cannot provide proof of audibility without taking into consideration the masking effects of ALL the channels combined...that is how we listening to movies and music.

 

[Soundmixer]

My Genelec 7060B subwoofer has a feature like this in its bass management system, in which >85 Hz frequencies in the LFE channel are sent to the center channel where they can be reproduced full range. This is precisely the kind of feature I wouldn't use because it is bound to cause trouble with problematic content.

This would fall under the "not needed for great results" category.