Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Ambler1980]

It only happens during movies with GIGANTIC explosions with SUPER DEEP bass.  99.9% of the time, it's fine.  And it never happens with music, even bass heavy music.

I am in a roommate situation so exclusively use headphones with JRiver, and they are good headphones (Sennheiser HD 558, plugged into external headphone amp + DAC).  I tried looking through some threads regarding this topic but got no solutions and alot of technical jargon that was hard to understand.  

I tried playing around with high pass filter, setting it around 50hz, then 60, then 40 after looking at the analyzer during the distortion/explosion scene, as the level seemed to peak around 50hz (no idea what I was doing)... it was SLIGHTLY better at 50hz high pass, but still distracting (and a little worrisome regarding my headphones, don't want to damage them).  

It's also important to note this distortion happens even when I turn the volume way down... so it is not the headphones that seem to be causing it... if the volume is very low, why would it distort if it's the headphones?  If the headphones can't handle that frequency/volume, lowering the volume should get rid of it, but it doesn't.  FYI it is during lossless DTS-HD/TrueHD tracks... it may also be for normal DTS/DD, but I'm not sure, I only seem to notice distortion on high rez tracks.

For now I just turn the volume down during massive explosions.... clip protection does nothing.  Isn't that what clip protection is for?

[JimH]

Try some of the volume settings in Options for Audio.  Loudness, for example.

Try Night Mode.

Turn off all DSP.

[RD James]

I don't think they've implemented the option to apply a low-pass filter to the LFE channel when using JRSS mixing yet.

I guess none of the dev team uses headphones to watch movies because it doesn't seem to be a priority for them.

[Matt]

I don't think they've implemented the option to apply a low-pass filter to the LFE channel when using JRSS mixing yet.

Of course the LFE gets low-passed.

You configure it in DSP Studio > Output Format > Subwoofer

Or do you mean the LFE channel when it gets mixed into a 2.0 output?

[RD James]

Or do you mean the LFE channel when it gets mixed into a 2.0 output?

This.
The HD formats require a 120Hz 24dB/octave low-pass filter on the LFE channel.
There was a big topic about this a year ago: https://yabb.jriver.com/interact/index.php?topic=94929.0

[Ambler1980]

Try some of the volume settings in Options for Audio.  Loudness, for example.

Try Night Mode.

Turn off all DSP.

I'll try some of this when I get home from work.  But turning off all DSP isn't an option for me, as I have the sonarworks reference 3 headphone plugin and I paid $70 for it, I'm not turning it off.  It's the only DSP I have running.

[Ambler1980]

This.
The HD formats require a 120Hz 24dB/octave low-pass filter on the LFE channel.

Is there a way to do this?  (Sorry, running to work and don't have time to read that thread at the moment)

[mojave]

JRiver's downmixing will clip on bass heavy movies unless you use Analyze Audio, Volume Leveling, and Adaptive Volume > Peak Level Normalize. Below is what I wrote in another thread dealing with just doing bass management. When downmixing to stereo, the clipping issue becomes much worse.

1.  Analyze Audio is recommended for movies because it helps Volume Leveling and Adaptive Volume prevent clipping during bass management.
2.  When Volume Leveling or Adaptive Volume > Peak Level Normalize aren't used, JRiver can clip during bass management. I would say the clipping will be rare.
3.  When Adaptive Volume > Peak Level Normalize is used by itself, it can prevent clipping in most cases.
4.  When Volume Leveling and Adaptive Volume are used, there will never be clipping during bass management.

[JimH]

I'll try some of this when I get home from work.  But turning off all DSP isn't an option for me, as I have the sonarworks reference 3 headphone plugin and I paid $70 for it, I'm not turning it off.  It's the only DSP I have running.

Then it is especially important that you turn off DSP so you can test whether the plugin is a factor.

[RD James]

JRiver's downmixing will clip on bass heavy movies unless you use Analyze Audio, Volume Leveling, and Adaptive Volume > Peak Level Normalize.

It's not clipping caused by downmixing.
If you export the audio to a multichannel track, the LFE channel itself is distorted.
The difference is that when you output multichannel to an AVR, the AVR is automatically low-passing the LFE channel. That's why you only notice it when you use JRSS to downmix.

Then it is especially important that you turn off DSP so you can test whether the plugin is a factor.

It's not a DSP plugin issue.

[JimH]

It's not a DSP plugin issue.

Testing without it might prove that.

[Hendrik]

It's not a DSP plugin issue.

Or so you might assume. The majority of HD tracks are mastered properly without unwanted frequencies in the LFE, so there is no guarantee that its the same particular problem you are referring to. Considering he didnt even name the movie...

[RD James]

Or so you might assume. The majority of HD tracks are mastered properly without unwanted frequencies in the LFE, so there is no guarantee that its the same particular problem you are referring to. Considering he didnt even name the movie...

It's a safe assumption. The LFE track for a lot of movies is distorted.
The specs for the HD audio formats say that the audio is meant to be low-passed by the decoder, which isn't happening in JRiver/LAV Audio.
It's just that you're not likely to encounter it unless you do a downmix which includes the LFE channel because most downmixes discard LFE, an AVR will either be decoding the bitstream and low-passing it, or the AVR will be sending a low-passed output to the subwoofer anyway.

[mojave]

If you export the audio to a multichannel track, the LFE channel itself is distorted.

I've done that with lots and there are hundreds (400+) of Blu-rays that have had the LFE track analyzed at data-bass.com's forum by extracting the audio without using a low pass filter. I or others haven't found any distortion in the LFE tracks.

[RD James]

I've done that with lots and there are hundreds (400+) of Blu-rays that have had the LFE track analyzed at data-bass.com's forum by extracting the audio without using a low pass filter. I or others haven't found any distortion in the LFE tracks.

It's not something that will show up in that type of testing.
Here's the clip which corresponds to this measurement from the data-bass forum: http://www.filedropper.com/tdk-lfe (110MB zip containing "TDK-LFE.mkv")

http://data-bass.ipbhost.com/index.php?/topic/12-the-low-frequency-content-thread-films-games-music-etc/page-7#entry249

Mute every channel except the LFE.
Swap the LFE channel with the center channel and play it through your speakers. (at your own risk)
It has lots of really bad clipping.
Add a 120Hz 24dB/octave low-pass and it sounds perfect.
 
LFE channels are meant to go to subwoofers so it's assumed they will be low-passed.
I don't think most AVRs will even let you pick a crossover higher than 120Hz.
When downmixing lossless 5.1 tracks (DTS-HD, TrueHD, LPCM, FLAC) the LFE channel needs to have a low-pass filter applied.
Compressed formats should already be low-pass filtered so I don't think it's an issue there.
 
Though it should be enabled by default, the low-pass should probably be optional.

[mojave]

It's not something that will show up in that type of testing.

Maxmercy runs all tracks through Audacity or similar to detect clipping in any of the channels.

Thanks for the clip suggestion. I think I have The Dark Knight so I'll check it out directly from the disc, too.

[RD James]

Maxmercy runs all tracks through Audacity or similar to detect clipping in any of the channels.

Audacity will only detect when a track is clipping at 0dB. This is not that sort of clipping.
It's been clipped at some point in the mastering process and then reduced in volume on the disc.
Difficult to detect in software, immediately noticeable to the human ear.

Thanks for the clip suggestion. I think I have The Dark Knight so I'll check it out directly from the disc, too.

The clip I uploaded was straight from my disc. If it's not the same, it's a regional difference.

[blgentry]

Audacity will only detect when a track is clipping at 0dB. This is not that sort of clipping.
It's been clipped at some point in the mastering process and then reduced in volume on the disc.
Difficult to detect in software, immediately noticeable to the human ear.

I've looked and listened to the clip.  Track #4 in Audacity appears to be the LFE channel.  I'm not 100% on this, but I don't think what you hear is clipping.  The waveform doesn't look like it.  I listened to it, and it definitely has high frequency components that sound nasty.  They sound like they shouldn't be there.  Adding the suggested 120Hz 24dB/oct low pass filter, definitely takes away the high frequency nasty sounds.

So, while I disagree with your diagnosis of the source of the problem, I fully agree that I hear it and with your solution to said problem. 

I can't seem to manually configure the DSP studio to not do JRSS mixing *and* let me mix it myself for 2 channel playback.  I seem to not have access to channels other than L and R.  I'm probably missing something.

Brian.

[RD James]

I can't seem to manually configure the DSP studio to not do JRSS mixing *and* let me mix it myself for 2 channel playback.  I seem to not have access to channels other than L and R.  I'm probably missing something.

It can't be done because JRiver sets the output format as the first stage in DSP instead of it being the last stage.
So if you're playing to a stereo device there is no more LFE channel to apply a filter to by the time it gets to the parametric eq.
That's why we need to wait for them to add in the option for a low-pass filter for lossless audio formats.

[Ambler1980]

Okay just got home from work.  I will test it without the DSP but I believe I still heard that type of clipping before I purchased it.

Also, the movie in question is Return of the Jedi, DTS-HD 7.1 downmixed to stereo for headphones.  The scene in question is the very end of Jabba's sail barge when it explodes.  The clipping/distortion/whatever is pretty bad, even at low volumes.  It sounds like crumpling paper.  But really it seems to happen on any movie with very DEEP explosions...

But I will test this out real quick and get back to you.

[Ambler1980]

Okay, disabling the sonarworks DSP did nothing (as If figured it wouldn't).  So I had no DSPs running (except clip protection, which obviously didn't help, and the stereo downmix).

[Hendrik]

In the next build:
NEW: Output Format DSP can lowpass the LFE/Subwoofer when downmixing, or exclude it from the downmix entirely.

This re-uses the existing subwoofer options on the output format DSP, if you go looking for it once its out.
The lowpass set there is now available either when downmixing or upmixing (ie. when JRSS processes the subwoofer signal). When playing an existing LFE signal without mixing, you would want room correction or PEQ to apply any lowpass or other subwoofer corrections.

[RD James]

Thank you. Two questions:

1. What kind of low-pass filter does the subwoofer option use?
2. This only applies to lossless formats, right?

[Hendrik]

1. What kind of low-pass filter does the subwoofer option use?
2. This only applies to lossless formats, right?

1. 48db/octave butterworth, as available through PEQ as well.
2. No, it does not necessarily know what kind of format is being played, especially since in video format that could change without the processor being entirely re-created. If you downmix, you will likely always want to lowpass anyway.

If you are worried about low-passing too much content, you could always increase the cut off slightly. But lossy formats have extremely limited bandwidth LFE anyway, so a lowpass at 120Hz or so is unlikely to remove any actual audio.

[RD James]

Can the low-pass be changed to a 24dB/octave one? I thought that's what the specs called for.

[Hendrik]

Its not meant to follow any particular audio codec spec, they probably disagree anyway.

[RD James]

I did a test with some white noise and it doesn't seem like there is any difference below 120Hz if you use a 24dB or a 48dB filter so I guess it probably doesn't matter.

[mojave]

It's not something that will show up in that type of testing.
Here's the clip which corresponds to this measurement from the data-bass forum: http://www.filedropper.com/tdk-lfe (110MB zip containing "TDK-LFE.mkv")

I tested the file from you and a couple other sources and they all had the "clipping sound" in the TDK when not low passed.

[RD James]

I tested the file from you and a couple other sources and they all had the "clipping sound" in the TDK when not low passed.

Thanks for confirming. Sounds like this should be fixed in the next build.

[Hendrik]

The option is there but it doesn't seem to do anything.

I tried your TDK-LFE sample from above, and I can hear night and day differences between all frequencies, no LFE, or a lowpass setting.
As with nearly all output format settings, you need to restart playback to make them catch on, just to be sure.

[RD James]

I tried your TDK-LFE sample from above, and I can hear night and day differences between all frequencies, no LFE, or a lowpass setting.
As with nearly all output format settings, you need to restart playback to make them catch on, just to be sure.

Two things are happening:

1. I did not realize that I had to stop and resume playback to make a change, and it doesn't show in the Audio Path
2. I have to use a 30Hz low-pass filter for it to not distort. When my output is 5.1 instead of using 2.0 JRSS, a 120Hz 24dB/octave filter was sufficient.

Any idea why I need to use such a low low-pass filter with JRSS mixing?
I am also upsampling to 192kHz and applying volume leveling.

[Hendrik]

120Hz seems fine for me to not hear any of the distortions anymore.

[RD James]

I think it might be the upsampling, as it doesn't happen at 48kHz instead of 192kHz.
It may be an order of operations issue if you're low-passing before upsampling.
I say that because I upsample in my other zone where I'm applying a PEQ filter and it doesn't distort.

[Hendrik]

Resampling is done first, I suppose that changes things? Sounds odd, though. Wouldn't a low-pass work the same no matter the sample rate?

[RD James]

Resampling is done first, I suppose that changes things? Sounds odd, though. Wouldn't a low-pass work the same no matter the frequency?

Well that's the same order it would be if the low-pass is being done by the PEQ anyway.
It only distorts like this when I upsample the audio though.
The low-pass shouldn't be affected by that.
 
Is it possible that the low-pass filter is being scaled up with the resampling somehow?
So a 30Hz LPF applied to a 48kHz file would be a 120Hz LPF at 192kHz?

EDIT: Nope, it isn't that. But I'm setting up some tests that show obvious distortion for some reason.

[Hendrik]

I just tried with upsampling to 192kHz and downmixing to stereo and the obvious distortion in the TDK sample doesn't come back at 120Hz low-pass for me.

[RD James]

I've uploaded a test file that just plays white noise in the LFE channel. http://www.filedropper.com/lfe-test
With a 120Hz Subwoofer setting, it distorts badly at its native rate. (attachment 1)
With a 120Hz Subwoofer setting and upsampling, I get pops and clicks every few seconds.
With a 120Hz PEQ LPF I get clean audio. (attachment 2)

[Hendrik]

I found the problem, should be resolved in the next build.

[RD James]

Seems to be fixed in 21.0.58

The default for a new zone seems to be a 60Hz LPF though, shouldn't that be 120Hz?

[Hendrik]

The default has always been 60 for the low-pass filter, thats not going to change with this. It just introduced a new way to use it.

[mojave]

The default has always been 60 for the low-pass filter, thats not going to change with this. It just introduced a new way to use it.

Because the setting changes its behavior based on channel count, I think the default should change, too.

For everyone that has Output Format set to 2 channels, the old default did nothing to the LFE channel. The entire LFE channel was down mixed. Now the default will put a high pass filter at 60 Hz on the LFE channel prior to down mixing. I can't think of anyone that would would want a 60 Hz low pass on the LFE. Many people don't even read the Release Notes and won't catch that they have to manually make a change to keep things the way they were.

1.  For less than 5.1 channels, the default should be 120 Hz.
2.  For all other channel counts, the default should stay 60 Hz.

This basically keeps existing systems the status quo regarding behavior for downmixing or for adding a subwoofer channel with JRSS.

[Hendrik]

The default can't change based on other options, unless you want your explicit 120/60 choice overwritten every time.
120Hz isn't "as things were" either, that would be disabling the low pass.

Unless I switch the options around depending on the selection, but that sounds fishy, and potentially complex.

[Hendrik]

Changed: The LFE low-pass in the Output Format DSP handles the settings for downmix and upmix separately (with separate defaults as well).

There, I made it store two options, the currently configured value will in the next build only be used for upmix (like before), and downmix will default to 120Hz and be stored separately (so if you want any other value than 120Hz, you need to set it again) - and the options will dynamically switch between the two values depending on if your output channel count includes a sub. Had to modify all things that use the DSP, so hopefully I didn't break anything.

I originally wanted to avoid this since it makes everything much more complex, but oh well.

[mojave]

Never has such complexity taken so little time.  

I was looking at the Marantz AV8801 manual earlier today and saw that they have a Bass Sync feature. Who decided this feature was necessary and how would anyone ever know if the LFE was out of sync with the other channels?

[Hendrik]

Never has such complexity taken so little time.  

Well I investigated this before when I build the feature for the first time, so it was just a matter of getting it all done.
And to be truthful, i hunted a few bugs for some time after I posted here. :p