DIGITAL At Least a Dozen 2002/2003 Elliot Scheiner 5.1 Mixes Have a LFE Issue (info/list/fixes inside)

[steelydave]

Thanks for all of this, @Cheezmo - I'll start a new thread for this eventually that's like a master list of all LFE issues, but feel free to keep dropping your findings here in the meantime.

 

[sjcorne]

Some of this stuff is just bonkers. Tarkus' LFE is inverted and delayed 20ms.

Interesting! That might explain some of the comments about a lack of low-end in the poll thread.

 

[paligap]

As far as I can tell, the 2011 Japan remaster of The Nightfly does not appear to have the same issue as the earlier releases. If someone has better waveform analysis ability than me, maybe they can find it.

 

[paligap]

Some of this stuff is just bonkers. Tarkus' LFE is inverted and delayed 20ms.

What tool did you use to detect this? All I've used is Audacity, and on some files, like all those on Tarkus, when I zoom in to a certain point, the waveform(s), especially the LFE channel, kind of pixelates and loses any definition where I can compare it to the others. Maybe it's partly because there's not much going on there on Tarkus.

 

[Cheezmo]

What tool did you use to detect this? All I've used is Audacity, and on some files, like all those on Tarkus, when I zoom in to a certain point, the waveform(s), especially the LFE channel, kind of pixelates and loses any definition where I can compare it to the others. Maybe it's partly because there's not much going on there on Tarkus.

I load them in Audacity and analyze the waveforms by eye. I usually zoom in 9 clicks on the "+" button to get to the right range to look at the LFE. I'm starting to recognize when you need to invert pretty easily. Usually scan through and measure the distance between front/right peaks and LFE peaks at easy to recognize shared impacts. When you sync it up, it looks right.

 

[Cheezmo]

The Genesis SACDs are a hoot. It looks like they flipped a coin on each track and either left the 3ms delay introduced by the low pass filter or inverted the LFE and added another 5ms delay (8ms total). Really bizarre.

 

[ssully]

SpecWeb’s LFE is an infinite slope digital filter (runs in the frequency domain). Any sounds below 20Hz or above 90Hz are rejected. The frequency response between 20Hz and 90Hz is flat.

"""""""""""""""""""""""""""""""""""""""""""""""""""""""

When SpecWeb creates a LFE channel from content that is low pass filtered from the input, is that content also filtered *out* of the other channels?

 

[ssully]

Some of this stuff is just bonkers. Tarkus' LFE is inverted and delayed 20ms.

That's not what I'm seeing. Take a look for example at 2:00.280 and after . This is a section of the tune where there is a big bass presence. Slight delay compared to front left channel, but not inverted. It helps to actually amplify the LFE so you can see the waveform very clearly.

I think you/we all need to be very, very careful before drawing broad conclusions from samples of tracks.

 

[AYanguas]

When SpecWeb creates a LFE channel from content that is low pass filtered from the input, is that content also filtered *out* of the other channels?

I have not found anything related to this in the SpecWeb 2.2 Guide. Only the text I transcribed and the recommendation of not using LFE at all for music. With 5.0 channels the only actors would be the Bass Management implementation in the AVR and the proper calibration.

It would be better if you ask in the SpecWeb thread: SPECWEB (Now 2.2)

 

[Cheezmo]

That's not what I'm seeing. Take a look for example at 2:00.280 and after . This is a section of the tune where there is a big bass presence. Slight delay compared to front left channel, but not inverted. It helps to actually amplify the LFE so you can see the waveform very clearly.

I think you/we all need to be very, very careful before drawing broad conclusions from samples of tracks.

So you don't think that negative peak in the L/R at 2:21.096 matches the positive LFE peak at 2:21.116? My flip/20ms seems to make everything match pretty well through that section. I can't find a simple delay that comes close. And yes, I'm amplifying it to get a clearer picture.

 

[ssully]

OK, I can see that working. I tried your fix and checked the very start end end of where the LFE has any significant content in the track...it does seem to bring it into register at both ends.

 

[paligap]

Some of this stuff is just bonkers. Tarkus' LFE is inverted and delayed 20ms.

So the fix is to invert the LFE and delay all other channels 20ms? And is that for the whole album, or just the Tarkus medley?

 

[HomerJAU]

Maybe @steelydave could do phase check on his phase meter to confirm? (if he has Tarkus)

 

[steelydave]

@HomerJAU I do, I'll try and give it a go in the next couple of days.

 

[ssully]

So the fix is to invert the LFE and delay all other channels 20ms? And is that for the whole album, or just the Tarkus medley?

Clip 20 msec off the start of the LFE and invert it.

I'm A/B comparing the two casually now on 5..2 system and I'm not feeling a blow you away difference, like Blow By Blow was. A more structured comparison might reveal more.

 

[Cheezmo]

Pointing out another extreme weird case.

Lauren Ellis - Feels Like Family (DualDisc) the LFE is inverted and 26ms EARLY.

 

[paligap]

Clip 20 msec off the start of the LFE and invert it.

I'm A/B comparing the two casually now on 5..2 system and I'm not feeling a blow you away difference, like Blow By Blow was. A more structured comparison might reveal more.

Just so I'm clear, is there a difference between delaying the other channels 20ms (using MMH, easy-peasy) and "clipping" 20ms of the start of the LFE (using another tool, I presume)?

 

[Cheezmo]

Just so I'm clear, is there a difference between delaying the other channels 20ms (using MMH, easy-peasy) and "clipping" 20ms of the start of the LFE (using another tool, I presume)?

Not really. It is easier to clip one channel than delay 5 in Audacity. I guess there is the potential of losing 20ms of content on the LFE doing it that way but 99.9% of the time it is just silence being trimmed.

 

[timbre4]

Just a quick note that I downloaded a Sea Change - Fixed (DVD-A) file and burned a DVD-A to see what might be different*.
* this thread indicates that the DVD-A version was "OK' so I'm not sure yet what was fixed?

I only had time to sample Lonesome Tears before I pulled in the driveway (Riverside stereo live disc was so good hard to swap out for this one). All I can say on this short sampling is that something was different as my brain knows from dozens of listens to Sea Change how instruments should sit. This is just a bookmark to bring it up to see if anybody knows about this version and I will elaborate on it as time permits.

 

[Cheezmo]

I finally finished going through my entire library, I don't have extensive notes on every track/release as there were often exceptions and variations on individual tracks and I just wanted to slog through it all. Plus if you are obsessive enough to fix this stuff, I imagine you will want to verify things yourself.

One consistent set of releases of note. All the Steven Wilson XTC and Yes releases have a 10ms LFE delay except for XTC Psurroundabout Ride.