The Bass EQ for Movies Thread

[SME]

It's not really a proper bass EQ, but I decided to try getting a bit more low end out of "the Martian".  First, a disclaimer.  I don't know what's going on in the individual channels or how much headroom was there to play with, so I can't guarantee that this won't push one of the channels over full-scale or push the overall level above a 7.1 WCS.  The way my BEQ system works, the only headroom limitation I have is at the amps.

 

Looking at the PvA, it looks like there might be a slight shelf around 25 Hz, so I applied this to all channels:

 

        LowShelfFilter(f0=25.0, gain=+7.0)
        LowShelfFilter(f0=15.0, gain=+1.0)
 

How did it sound?  Well, the boost was enough to make at least some of the 10 Hz "tension effect" perceivable.  It definitely added some extra heft to the doors and quite a few other effects.  On the flip side, it robbed one of the disco songs of some punch.  Maybe if I could have reduced the amount of boost at ~40 Hz that'd not be a problem, but sadly that's an area that could have used a bit more output on the launch scenes.

 

The launch scenes were definitely improved.  The latter one with the heavy ULF is a lot more exciting with the boost, which pushed the single digits high enough to get a bit of floor wobble going.  My wife was apparently standing at the back of the room, near the hall, and said she felt her hair moving.    Also, the few big explosions got just an extra bit of heft.

 

I'm not sure if it was really worth it though.  Next time, I'll probably watch the movie without the EQ, but I don't know if I'll remember this experience well enough to fairly compare.

[Kvalsvoll]

Wanted to see Titanic, never saw it before.

Watching the opening scenes reveals something is missing, and from what I can hear - such a pity. Can this be improved..

And it can. There is a steep 25hz high-pass, and doing the bass-eq magic retrieves some of the weight and natural balance that was lost in the studio because they had mediocre monitoring speakers.

The exact numbers are on a different computer, but something like this:

L,C,R:

2x sfm 22hz q=2.2 gain=+9dB, + eq to adjust for the inevitable midbass dip.

LFE:

2x sfm 22hz q=2.2 gain=+9dB, 1x sfm 22hz q=2.2 gain +6dB + eq to adjust for the inevitable midbass dip.

 

Much better, the helicopter in an early scene now sounds great, you can sense your hair moving as it comes around to land.

[maxmercy]

Kong: Skull Island may be amenable to BEQ.....

 

JSS

[SME]

3 hours ago, maxmercy said:

Kong: Skull Island may be amenable to BEQ.....

Way cool!  But does it need it?  (I haven't watched it yet FWIW.)

[SME]

I watched "Thor" with BEQ tonight.  It's the first time I've watched "Thor" on the new system, and I'm glad I waited.  This was a very loud track with a lot of wide-band upper mid / high frequency effects.  And of course there is clipping in a lot of places.  However, I was pleasantly surprised by how good it sounded at my usual "-4" playback level and without any roll-offs applied, even at the top.  The loud effects were never uncomfortable or harsh.  In some sense, this was a great mix for the mixers, but not so much for those with inferior systems, which would tend to not handle the harsh, overdriven waveforms as gracefully.  Certainly the last time I watched this on mere "92 dB/2.83V/1m" speakers, I found it to be uncomfortable at "-7".

The bass here with BEQ was a lot of fun.  I think a lot of sequences in this are demo-able.  At one point, I noticed very slow sweep down in the infrasonic range.  Very subtle, but very effective for the scene it was in.  I also noticed some nice ULF ambiance (wind) in the some of the desert scenes.  Perhaps this was just "noise" picked up by the mic, but it fit the picture very well.  In fact, nothing sounded out of place with the BEQ.

I think about the people on AVSForum who argue that extra extension is worthless because "the director didn't hear it that low" on the stage.  True enough, but that doesn't mean that it isn't a great track with full extension.  I'd argue that I heard it the way the director would have wanted it to be heard, if he/she had known better.

Big thanks again to JSS for putting this together.  I have to say that I've enjoyed the ULF in BEQ films more than in just about any of the naturally extended films I've watched.   Your work is high quality. 

[SME]

So about "Kong".  One guy says it challenged his Raptor 3 system.  But it could benefit from BEQ?  How does that happen?  Maybe it's content above 15 Hz that pushes them so hard.  If so, I'm going to have a real fun time with this one. 

[SME]

Continuing the conversation about GOTG2 BEQ from here:

@3ll3d00d posted PvAs: https://imgur.com/a/mjxeS

...

First off, can you tell me which channel is which?  I believe the standard order for 5.1/7.1 is: FL/FR/SL/SR/FC/LFE/RL/RR, but it looks like LFE may be 4th in the sequence in the images you posted.  I'm going to take a wild guess: FC/FL/FR/LFE/SL/SR/RL/RR?

It's remarkable how different the channels look.  The 30 Hz notches in LRCS are especially weird.  I wonder if that's common.  Perhaps it helps keep the 30 Hz bump in the LFE from being too peaky on the down-mix?

The one thing I can't do with just the PvAs is check the final levels.  So it's possible that whatever solution I come up with could clip when played at reference if the output from each channel must be kept under "0 dBFS" relative to the input level.  My DSP doesn't have that limitation, but I'm pretty sure it will be an issue when using an HDMI MiniDSP upstream of the AVR.

Hey @maxmercy, if you're reading, feel free to drop any tips you have for how you would approach this.  My inclination is to try the following:

• shelf on LRC of roughly 18 dB @ 37 Hz
• shelf on S of roughly 36 dB @ 32 Hz
• shelf on LFE of roughly 12-15 dB @ 15 Hz

I'm thinking it may take multiple filters to get the right shape on some of these, particularly the LRC.  I'm worried about weird things happening to the score, but I guess that's what listening is for.  Thoughts?

[3ll3d00d]

It is L R C LFE RL RR SL SR (which is pretty standard afaik)

[SME]

6 hours ago, 3ll3d00d said:

It is L R C LFE RL RR SL SR (which is pretty standard afaik)

Thanks!  It looks like several standards exist.  I did some research, and it looks like the order you quote here is very common but FR and FC are often interchanged.

The order I gave was not correct for the "standard" I intended to quote.  The correct order based on that "standard" would be: FL/FR/RLS/RRS/FC/LFE/SLS/SRS.  However, for 5.1, the SL and SR channels occupy the RLS and RRS positions.  I think the idea of this order is that it allows 2.0, 4.0, 5.1, and 7.1 configurations to co-exist without remapping.  It may be mostly a Linux-ism, as that is the order that the channels appear in for my HDMI/PCM audio interface under Linux, and it may be doing remapping to whatever HDMI/PCM actually uses.

 

[maxmercy]

Let me look at it.  I'll come up with what I think will be a decent solution tomorrow.

JSS

[SME]

52 minutes ago, maxmercy said:

Let me look at it.  I'll come up with what I think will be a decent solution tomorrow.

Excellent!  And please, if you don't mind, share your thought process.  I'd like to learn how to do this.  It'll still be a while before I can analyze the soundtracks, but once I 'm there, I'll be eager to do this with other titles. 

[maxmercy]

I basically look at each individual channel, and see what may have been filtered/shelved and then try to get a decent solution to reverse the initial shelving.

That 30Hz dip is strange.  I'm separating the audio channels into individual wavs and graphing them this evening to ensure I get the same FR before I begin.

JSS 

[SME]

19 hours ago, maxmercy said:

I basically look at each individual channel, and see what may have been filtered/shelved and then try to get a decent solution to reverse the initial shelving.

That 30Hz dip is strange.  I'm separating the audio channels into individual wavs and graphing them this evening to ensure I get the same FR before I begin.

JSS 

Is that all?  That was basically my thinking.  The remaining question is whether the PVAs posted above are accurate or consistent with what you work with.  BTW, I just picked up the movie last night, so I'll be able to do some testing once we have some trial filters.

[3ll3d00d]

FWIW I used jriver convert format to generate the input wav file and then these are run through my app in a way that is equivalent to the approach used around here in speclab (which decimates it to 1000Hz and then runs it through the fft with a nuttall window of either 0.5Hz or 1Hz resolution, I forget which). I used a 0.5Hz resolution in my case, this is pretty much all done by scipy (for the fft bits) and librosa (for the resampling)

I thought I'd double check so used eac3to to generate the wavs instead (though they both use the same underlying dts decoder anyway - https://github.com/foo86/dcadec), same result. 

If you get a different result then it would be interesting to know how you're performing that extraction and conversion.

[maxmercy]

3 hours ago, SME said:

Is that all?  That was basically my thinking.  The remaining question is whether the PVAs posted above are accurate or consistent with what you work with.  BTW, I just picked up the movie last night, so I'll be able to do some testing once we have some trial filters.

No magic or mystery, this all started trying to undo the amazing Avengers shelf filter....after I BEQ a film, I generally watch it before I say it's any good, as I have bloated out a few mixes.  I now basically just try to get back whatever was shelved out while trying to preserve the midbass (not easy to do with very steep shelves, like Pacific Rim, that one took me many tries to get right).  I'll  post up my LCRS graphs this evening.

JSS 

[maxmercy]

I hope the mixers just pulled down the 30Hz to keep the LCRS from distorting with high level playback (most theater LCRS are tuned to 40-45Hz), as the days of the twin BagEnd 18" units under the LCR in a soundstage are no longer around, AFAIK.  FilmMixer used to work on a stage that had twin sealed 18s under each LCR in order to monitor what was encoded down low, along with sixteen 18" sealed units along the front wall floor to monitor LFE.  

Another BEQ I have done had a similar dip in the 25-30Hz range and sounded better with the dip filled in, IIRC.  I would be interested in hearing what you think filled vs non-filled....

JSS  

[maxmercy]

Center and LFE:

At first glance, Center will be probably easy.  Shelve up the 10-30Hz range to nearly meet 40Hz and up.  Then look at the 30Hz Dip.

The LFE will prove similarly easy from 30-10Hz, but below that diminishing returns, as it looks like a highpass was added under 10Hz, around 7Hz if I had to guess.

JSS

[maxmercy]

Right/Left:

 

 

Probably will receive same correction as Center, as is usually the case.  Shelve up 10-30 to meet 40+ and then look at the dip.

I like it when tracks use the upper bass, like that 140Hz hit at -7dB...it generally makes a BEQ better if there is already a lot of upper bass present 

JSS 

[maxmercy]

Surround Left and Right

 

Similar treatment to these , but maybe a bit more de-shelving from 40-60Hz, depending on how the back surrounds look.

I basically try to match the PvAs from my favorite, unfiltered films, and it usually does the trick very well (but sometimes fails).

JSS

[maxmercy]

Back Surrounds:

 

Looks like only 2-3 solutions needed, one for LFE, one for LCRS and if you want, one for LCR and a separate one for Surrounds.

The more I look at these graphs, the more I think someone heard the LCRS 'farting' with the content and pulled ~30Hz down to keep it sounding 'clean'.

JSS

[maxmercy]

Center pre-post, looks like a decent solution:

The shelves and dip cleaned up nicely.  I usually shoot for a flat peak graph, I have found it sounds better than correcting for the avg graph.  Others may differ.

JSS

[maxmercy]

Left Channel Pre-Post:

 

 

I think that is a pretty good solution, at least one worth trying out for LCRS.  I'll work on LFE later today, and have a tentative full solution.

 

LCRS tentative solution:

Gain -7dB

Shelf Filter 30Hz, Q=0.707, Gain=6

Shelf Filter 30Hz, Q=0.707, Gain=6

Shelf Filter 30Hz, Q=0.707, Gain=6

Shelf Filter 12Hz Q=0.707, Gain=6

Shelf Filter 12Hz Q=0.707, Gain=6

PK EQ 31Hz,  Q=2.871, Gain=6

PK EQ 31Hz,  Q=2.871, Gain=6

 

I have found in practice I get smoother, more predicatble curves with multiple +6dB or lower steps than with one big filter,  especially when using higher Q values, so in this case, this solution could be  made into just 3 filters and a gain change.  YMMV.

 

JSS

[maxmercy]

Finally, LFE pre-post:

 

About 7 iterations to get this one right, as headroom was at a premium.  Turns out my first guess was closest to my final solution.  As I expected, <10Hz is highpassed away, only able to be brought back by ever increasing negative gains before BEQ.

Tentative LFE Solution:

Gain -7dB

Low Shelf 18Hz Q=0.707 +5.25dB

Low Shelf 18Hz Q=0.707 +5.25dB

Low Shelf 18Hz Q=0.707 +5.25dB

 

SME, if you can input that into your incredible DSP, I'd be interested in what you think of the solution.

 

JSS

[Ricci]

John how much time does it take you to go through the process above?

[3ll3d00d]

If the approach is based on a target curve which is based on other known good bass films then t seems like this should be amenable to automation, at least as a first cut anyway.

For example, create a minimum phase representation of the actual response and a target curve, load into MSO and create a configuration that has a load of adjustable Q shelf filters & some PEQ available to it then give it some time to work out what combination of filters produces the target response.