Print

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

[stewart_pk]

Hi, I have a 4.0 setup but want to mix LFE into my mains but not the surrounds.
This does not present a problem for video with audio tracks (e.g. 5.1, 6.1, 7.1).
I can high pass the left surround channel and add it to the left channel and high pass my right surround and add it the right channel.
In this scenario I believe that the JRSS mixer is adding 10dB to the LFE (.1) channel which it should.
All good?

When it comes to playing audio only tracks (e.g. 5.1, 6.1, 7.1) which I have playing in a seperate zone are these tracks also getting the LFE (.1) track boosted by 10dB?
And if so should they?
Is the same standard of -10dB on the LFE (.1) channel to be boosted later by 10dB applicable to audio only tracks (music)?
I think JRSS is boosting the LFE (.1) by 10dB for audio only tracks (music) and it shouldn't be.

What I might be potentially asking for is a switch on the JRSS mixer called "Don't add 10dB to LFE channel when mixing".

[Hendrik]

The nature of the LFE channel means that it should generally always be boosted, so that one single channel can compete with potentially 5 or 7 other channel in volume.

[stewart_pk]

The nature of the LFE channel means that it should generally always be boosted, so that one single channel can compete with potentially 5 or 7 other channel in volume.

If so would that mean that the maximum volume the LFE channel is mixed to is -10dB below full scale?

[mattkhan]

If so would that mean that the maximum volume the LFE channel is mixed to is -10dB below full scale?

a film soundtrack is mixed knowing that the LFE channel will be played back +10dB louder than the main channels after calibration. In an av processor, this is typically an analogue gain stage but in MC it would have to be achieved by attenuating the main channels instead. It should be easy to verify if JRSS is doing this by recording the output from MC.

[stewart_pk]

a film soundtrack is mixed knowing that the LFE channel will be played back +10dB louder than the main channels after calibration. In an av processor, this is typically an analogue gain stage but in MC it would have to be achieved by attenuating the main channels instead. It should be easy to verify if JRSS is doing this by recording the output from MC.

Hi, I know that the LFE channel needing to be played back +10dB louder for VIDEO tracks.
My question was about AUDIO surround tracks which is still unaswered.

I don't think JRSS is attenuating the other channels when it mixes in the LFE, but I could be wrong, it would be good to know.
And I suspect that JRSS doesn't just mix the LFE into the main channels it mixes it into all channels.
If so, this has long been one of my please change requests.
It really should just mix them into the main L and R channels (by default) because I think it's fair to say that the majority of 4.0, 5.0, 6.0 etc setups have larger mains than the other speakers which are better suited to LFE.

[stewart_pk]

The nature of the LFE channel means that it should generally always be boosted, so that one single channel can compete with potentially 5 or 7 other channel in volume.

I'm not sure this is quite right, I think the LFE needs to be boosted by 10dB bacause the standard says it should be attenuated by 10dB in VIDEO audio content to be 10 dB lower in output than the other channels.
From memory there is a standard (THX?) that says the LFE channel must be able to hit 115dB peaks while the other channels only need to be able to hit 105dB peaks, but that doesn't have any relation to mixing that I can see.

[mattkhan]

As far as I am aware there is no standard in audio so you can only deal with it on a case by case basis

Reference level for cinema is 115dB from the lfe alone and 105dB for main channels calibrated at the listening position(s)

It is relevant to mix downs because you want to end up with content being at the right relative level in the final mix

[AGAWA]

are we talking about .1 channel in surround recordings or LFE channel?

pls read the following:

"Tips on using the LFE Channel in 5.1
The LFE is a channel and is not the same as the subwoofer. It’s actually a 120Hz bandwidth limited channel (80Hz for DTS) that has an additional 10 dB of electrical headroom above the 5 main channels. It was created as part of the film version of Dolby Digital to enhance low frequency sound effects, such as large explosions and other LF content which exceeded the headroom of the main channel. Hence the name LFE: Low Frequency Effects and not low frequency bass channel. So, unless you are recording Tchaikovsky’s 1812 Overture with REAL cannons, it doesn’t need to be used in most music applications, but can be useful in film and multimedia applications.

An important note about the LFE channel: Never put anything critical to the mix exclusively in the LFE channel. This is because when a Dolby Digital decoder folds down a 5.1 source, it drops the LFE channel completely (poof!)."

DVD-A and DVD discs have .1 at -10dB and need boost while being played. Perhaps same with BR. SACD rarely use .1 channel.

[mattkhan]

The reference level comment refers to films which are encoded in a standard way. Music has, to my knowledge, no such standard.

[stewart_pk]

are we talking about .1 channel in surround recordings or LFE channel?

pls read the following:

"Tips on using the LFE Channel in 5.1
The LFE is a channel and is not the same as the subwoofer. It’s actually a 120Hz bandwidth limited channel (80Hz for DTS) that has an additional 10 dB of electrical headroom above the 5 main channels. It was created as part of the film version of Dolby Digital to enhance low frequency sound effects, such as large explosions and other LF content which exceeded the headroom of the main channel. Hence the name LFE: Low Frequency Effects and not low frequency bass channel. So, unless you are recording Tchaikovsky’s 1812 Overture with REAL cannons, it doesn’t need to be used in most music applications, but can be useful in film and multimedia applications.

An important note about the LFE channel: Never put anything critical to the mix exclusively in the LFE channel. This is because when a Dolby Digital decoder folds down a 5.1 source, it drops the LFE channel completely (poof!)."

DVD-A and DVD discs have .1 at -10dB and need boost while being played. Perhaps same with BR. SACD rarely use .1 channel.

The LFE channel and .1 are the same thing.
I have numerous DVD-Audio discs and when I inspect the content in Audacity there is plenty of content on the LFE (.1) channel and not one of my DVD-Audio discs is the 1812 overture.
And two DVD-Audio discs I recently looked at had content peaking about 5dB below full scale.
I think this means that if they were boosted by 10dB by JRSS then they could end up being digitally clipped and hazardous and not to mention just too loud and hazardous.

[stewart_pk]

The reference level comment refers to films which are encoded in a standard way. Music has, to my knowledge, no such standard.

This is exactly what I understand to be the case too, thank you.
At the moment I think JRSS is always boosting the LFE by 10db on downmixing and this could be dangerous with Audio only content (music) and the reason why I started this topic.
I was listening to DVD Audio the other day downmixed by JRSS to 4.0 including the LFE (.1) and the bass did seem too loud.

At the moment I handle this with two zones, one for VIDEO and one for AUDIO with more than 2 channels.
For both, I have JRSS downmix to 5.1 with the option of the center channel moved to the left and right and then manually mix the LFE (.1) channel into the L and R channels. For VIDEO audio I apply the 10dB LFE (.1) boost but for AUDIO only content with more than 2 channels I do not.

[Hendrik]

The LFE channel and .1 are the same thing.
I have numerous DVD-Audio discs and when I inspect the content in Audacity there is plenty of content on the LFE (.1) channel and not one of my DVD-Audio discs is the 1812 overture.
And two DVD-Audio discs I recently looked at had content peaking about 5dB below full scale.
I think this means that if they were boosted by 10dB by JRSS then they could end up being digitally clipped and hazardous and not to mention just too loud and hazardous.

JRSS downmixing will never result in clipping, if it needs to then the mains will be reduced in volume, but thats an unlikely scenario, as the volume is already being reduced as the data is spread over different channel, and the risk of clipping is reduced as data is not necessarily correlated.

Note that its entirely expected that even the LFE channel will have data close to the maximum the channel can support - the 10dB boost would serve no purpose if it was quieter, why would you not just make it louder in the file afterall.

[mattkhan]

At the moment I think JRSS is always boosting the LFE by 10db on downmixing and this could be dangerous with Audio only content (music) and the reason why I started this topic.
I was listening to DVD Audio the other day downmixed by JRSS to 4.0 including the LFE (.1) and the bass did seem too loud.

FWIW I played a 7.1 track that has silence on all channels except the LFE channel, mixed down to 4.0 in JRSS and looked at the resulting output. All channels were silent. It says the LFE channel is discarded when mixing down to 4.0.

[mojave]

The reference level comment refers to films which are encoded in a standard way. Music has, to my knowledge, no such standard.

ITU BR.1384-1 provides the "Parameters for international exchange of multi-channel sound recordings
with or without accompanying picture" which states that "the LFE channel is recorded with a level offset of –10 dB." DTS and DD also required the same level offset, even if used for just music.

There were a few multi-channel audio recordings on DVD-Audio/HD DVD and maybe some SACD's that had the level too low on the LFE channel. Also, some SACD players played back the LFE incorrect, but that isn't relevant here.

[mojave]

FWIW I played a 7.1 track that has silence on all channels except the LFE channel, mixed down to 4.0 in JRSS and looked at the resulting output. All channels were silent. It says the LFE channel is discarded when mixing down to 4.0.

I tried to duplicate using JRiver's Audio Calibration for 7.1 and downmixed to 4.0. It plays fine for me. Did you make sure Subwoofer isn't set to Silent (don't use subwoofer in downmix) in Output Format?

[mattkhan]

ITU BR.1384-1 provides the "Parameters for international exchange of multi-channel sound recordings with or without accompanying picture" which states that "the LFE channel is recorded with a level offset of –10 dB."

I can't say I follow such standards (or have lots of multichannel audio content) so I'm curious whether releases are applying this standard consistently, do you know?

[mattkhan]

I tried to duplicate using JRiver's Audio Calibration for 7.1 and downmixed to 4.0. It plays fine for me. Did you make sure Subwoofer isn't set to Silent (don't use subwoofer in downmix) in Output Format?

thanks and well spotted, I did have that option set.

[AGAWA]

https://www.dolby.com/uploadedFiles/Assets/US/Doc/Professional/38_LFE.pdf

[mattkhan]

repeated the same test

on the LFE only down to 4.0 -> -12dB down on each channel
on the LFE + L down to 4.0 -> L is 8.386dB down, other channels are -12

it's not immediately obvious to me what JRSS is doing in this case but i can't see how these numbers equate to there being a +10dB boost applied

fwiw in the 2.0 case, the same level difference appears to apply (i.e. L in ~3.7dB higher than R) which suggest some LR only balancing here?

I don't recall the behaviour of JRSS ever being documented so probably you just take it as is or be happy doing it yourself

[stewart_pk]

JRSS downmixing will never result in clipping, if it needs to then the mains will be reduced in volume, but thats an unlikely scenario, as the volume is already being reduced as the data is spread over different channel, and the risk of clipping is reduced as data is not necessarily correlated.

Note that its entirely expected that even the LFE channel will have data close to the maximum the channel can support - the 10dB boost would serve no purpose if it was quieter, why would you not just make it louder in the file afterall.

That is good to know that JRSS won't clip, but would it not be played 10dB too loud at typical volumne levels?

I don't understand your point in the second paragraph.
But regardless, I think there are two potential whorthwhile enhancements.
1) Provide a dropdown relative to downmixing with options of "Mix LFE to all channels" and "Mix LFE to the main L and R channels".
2) Provide a checkbox option of "Don't apply 10dB boost to LFE" which if checked would apply to both non-dowmixed and downmixed output.

[Hendrik]

If anything this thread shows that the 10dB boost for audio is not very well understood, and options resulting from such a situation are more likely to add to the confusion, rather then solve any problems.
All I see here is theoretical thoughts, no actual data that something is wrong.

From what I understand, properly mastered multichannel audio should always incorporate that boost, in videos or music, this includes DVD-A and Blu-ray Audio, typical sources for multichannel audio, as it serves a real purpose to be able to have the LFE compete with all the other channels in volume, otherwise the LFE would be too quiet.

Downmixing is a special situation as well, and the 10dB boost is handled differently here to account for the fact that its being downmixed. I would recommend to actually listen how it sounds, if the LFE would be 10dB too loud, that would be an easy situation to hear or measure.

[AGAWA]

more interesting reading:
https://www.audioholics.com/audio-technologies/0dbfs-blu-ray

[mattkhan]

Much of the confusion would disappear if there was a written statement of how JRSS actually (up/down)mixes.

Taking the 7.1 to 4.0 case, I can see that putting the same signal into L and LFE results in the LFE channel being equally distributed across all 4 channels so that the sum of those 4 channels is the same level as the LFE input while the L channel comes out with a signal that is 3.89dB higher in level.

This does not seem to equate to (LFE + 10dB split into 4 equal strength signals) + L as it seems this would result in a signal +5.06dB higher
if there there were no 10dB boost applied then it seems like you'd get a signal about +1.94dB stronger

The actual output is somewhere inbetween the two so the Q is what does it mean exactly when you say the 10dB boost is handled differently here?
In addition, as far as I can see, MC does not apply that 10dB boost at all and expects it to be handled downstream (e.g. by an AV processor).

[mattkhan]

more interesting reading:
https://www.audioholics.com/audio-technologies/0dbfs-blu-ray

this is not really relevant to this thread, that is about the fact that consumer equipment generally has problems performing bass management without clipping internally. It's a different thing to how MC up or down mixes.

[Hendrik]

In addition, as far as I can see, MC does not apply that 10dB boost at all and expects it to be handled downstream (e.g. by an AV processor).

If you actually output a dedicated LFE channel, then you should control its volume ideally in analog, and not in digital, so you can actually make use of the full range of the signal for all speakers, and don't need to reduce the digital level to allow for the 10dB boost. This is why its not applied in this case, because thats the most reaonable and most expected way to handle it. This is (hopefully) what happens when you let your AV processor or other multi-channel DAC handle it. Active subwoofers typically also have their own volume control to allow you to do this - which you need to calibrate to match your system regardless.

It would not make sense for MC to be the only source to pre-apply the 10dB boost, when every other source won't, and your pre-pro applies it as well, giving you inconsistent results between different sources.

As for technical details on JRSS, we won't detail its exact inner workings, sorry. Down and upmixing is not an exact science, there is some room for interpretation. If you find some real world audio where it doesn't work well, we're happy to listen, but artificial theoretical scenarios just won't get you there.

[mattkhan]

Fair enough. It does seem the premise of the OP is incorrect though, i.e. MC does not appear to be applying this boost at any time and hence if the OP wants to achieve what they want then they should continue to do it manually (and/or share content clips of samples where they think jrss is doing something undesirable).

[stewart_pk]

It does seem the premise of the OP is incorrect though, i.e. MC does not appear to be applying this boost at any time and hence if the OP wants to achieve what they want then they should continue to do it manually (and/or share content clips of samples where they think jrss is doing something undesirable).

Hang on I didn't get that from Hendrik at all!
Unless Hendrick will categorically state right here right now that JRSS won't effectively and I emphasise effectively boost the LFE by 10dB when downmixing to an output where there is no subwoofer, Hendrick?

[stewart_pk]

If anything this thread shows that the 10dB boost for audio is not very well understood, and options resulting from such a situation are more likely to add to the confusion, rather then solve any problems.
All I see here is theoretical thoughts, no actual data that something is wrong.

From what I understand, properly mastered multichannel audio should always incorporate that boost, in videos or music, this includes DVD-A and Blu-ray Audio, typical sources for multichannel audio, as it serves a real purpose to be able to have the LFE compete with all the other channels in volume, otherwise the LFE would be too quiet.

Downmixing is a special situation as well, and the 10dB boost is handled differently here to account for the fact that its being downmixed. I would recommend to actually listen how it sounds, if the LFE would be 10dB too loud, that would be an easy situation to hear or measure.

As mentioned there is seemeingly no standard for mastering music which includes the LFE channel.
Hence my request of:

Provide a checkbox option of "Don't apply 10dB boost to LFE" which if checked would apply to both non-dowmixed and downmixed output.

The way I see it, this wouldn't "hurt" anything and could potentially stop equipment from being damaged.

Also, what do you think about my request regarding where the LFE can be mixed to? This is actually not directly related to the request above but seperate.

Provide a dropdown relative to downmixing with options of "Mix LFE to all channels" and "Mix LFE to the main L and R channels".

I know there have been other previous and similar threads regarding this from other users.

[stewart_pk]

From what I understand, properly mastered multichannel audio should always incorporate that boost, in videos or music, this includes DVD-A and Blu-ray Audio, typical sources for multichannel audio, as it serves a real purpose to be able to have the LFE compete with all the other channels in volume, otherwise the LFE would be too quiet.

I don't follow you and I think you are misunderstanding something.

The LFE channel needs to be boosted somewhere, somehow for VIDEO audio because the standard says it must be mastered 10dB less than the other channels. Boosting the LFE has nothing to do with bringing 'balance' just because there are likely to be more speakers than subwoofers. The LFE must be boosted by 10dB to restore it to the same average level as the other channels. From there the consumer can do whatever they want with it e.g. down-mix it into the other channels, not downmix it into the other channels and ignore it, not ignore it, make it louder, make it quieter, high-pass it, low pass it, daisy chain 50 subwoofers in a room from the same LFE signal running full-tilt with it if they want etc.

[Hendrik]

Provide a checkbox option of "Don't apply 10dB boost to LFE" which if checked would apply to both non-dowmixed and downmixed output.

The way I see it, this wouldn't "hurt" anything and could potentially stop equipment from being damaged.

I already commented on that above - theoretical considerations are not enough reason to offer an option we don't have any confirmed data if its needed or helpful.
Please share any data or files where this is a real problem so we can look into it.

PS:
Equipment damage should always be impossible in your setup. If your equipment gets damaged when playing a 100% digital signal (and more is not possible), then it'll get damaged eventually, and you should limit it in hardware to prevent that.

[mattkhan]

Hang on I didn't get that from Hendrik at all!
Unless Hendrick will categorically state right here right now that JRSS won't effectively and I emphasise effectively boost the LFE by 10dB when downmixing to an output where there is no subwoofer, Hendrick?

It followed from my reading of previous test with synthetic signal however I realised the previous test was confusing so I tried again with 100Hz sine wave in L and 50Hz sine wave in LFE, with this it's clear that the LFE channel is boosted but the signal from the input L channel is also reduced by ~6.67dB. This is just for the 4.0 downmix btw.

Provide a checkbox option of "Don't apply 10dB boost to LFE" which if checked would apply to both non-dowmixed and downmixed output.

re the bit I put in bold... MC does not apply this boost, you can test this yourself and Hendrik said the same in https://yabb.jriver.com/interact/index.php/topic,133272.msg923355.html#msg923355

[stewart_pk]

It followed from my reading of previous test with synthetic signal however I realised the previous test was confusing so I tried again with 100Hz sine wave in L and 50Hz sine wave in LFE, with this it's clear that the LFE channel is boosted but the signal from the input L channel is also reduced by ~6.67dB. This is just for the 4.0 downmix btw.
re the bit I put in bold... MC does not apply this boost, you can test this yourself and Hendrik said the same in https://yabb.jriver.com/interact/index.php/topic,133272.msg923355.html#msg923355

Thanks for performing the test especially as Hendrik didn't answer.

I know that JRSS does not boost the LFE by 10dB for actual .1 channel output although I can see what I wrote might be confusing.

[stewart_pk]

I already commented on that above - theoretical considerations are not enough reason to offer an option we don't have any confirmed data if its needed or helpful.
Please share any data or files where this is a real problem so we can look into it.

If I sent you a multi-channel 5.1 FLAC or MP3 that has the LFE at a level within -10dB of full scale e.g. -4dB or -5dB would that suffice?

[stewart_pk]

Provide a dropdown relative to downmixing with options of "Mix LFE to all channels" and "Mix LFE to the main L and R channels".

Hendrik, if you are not interested in this enhancement than can you please say NO?
If so, I will simply leave it at that.

[stewart_pk]

Here I have a sample within Audacity.

Note in the first screen-shot the LFE channel (4) appears to peak at about the same level as the centre channel (3) which at this stage this is fine.
In the second screen shot I am applying a 10dB boost to the LFE channel.
The third screen-shot shows the LFE channel after it has been boosted by 10dB and it's become very loud (and clipped).

Now JRiver may have clip protection and also lower channels other than the LFE (L, R, C, LS, RS) to appropriate levels but the LFE will be too loud as far as I can see.

[stewart_pk]

Apologies for the size of the images above.
I tried to make them smaller but nothing seemed to work.
One can zoom out with their browser to see them completely.

[Hendrik]

As I said above, the LFE channel being within 10dB of full scale would be expected in a low-frequency heavy piece (that is mastered relatively loudly, most video content uses a lower reference level these days, because standards). The 10dB boost would serve no purpose if there always would be enough headroom to just boost the signal digitally - because then why would you not just encode it at the intended level and forget about this complexity?

Just a brief glance at my video library, the LFE channel being a similar volume to the main channels is entirely normal. Its not supposed to be 10dB quieter.

[stewart_pk]

As I said above, the LFE channel being within 10dB of full scale would be expected in a low-frequency heavy piece (that is mastered relatively loudly, most video content uses a lower reference level these days, because standards). The 10dB boost would serve no purpose if there always would be enough headroom to just boost the signal digitally - because then why would you not just encode it at the intended level and forget about this complexity?

Just a brief glance at my video library, the LFE channel being a similar volume to the main channels is entirely normal. Its not supposed to be 10dB quieter.

I repeat my request has nothing to do with VIDEO audio and everything to do with AUDIO audio.

But to answer your question, I think and this is just my take on the standard for VIDEO audio to master LFE 10dB less than all other channels was decided upon for safety, so that the end user has some protection for their subwoofer although this is of course negated by AVR's and processors automatically boosting the LFE by 10dB anyway. Perhaps more likely it was decided upon for the very reason of downmixing because if the downmixing algorithm doesn't factor in the mutiplication of the LFE being mixed into multiple channels the LFE will become too loud. JRSS is doing the correct thing for VIDEO audio, no problem.

But and to repeat again there is seemingly no standard for AUDIO audio and it seems some content creators will use the VIDEO audio standard as a guide but some have ignored it as shown in my previous example. I inspected another of my DVD-Audio discs and this time the LFE peaked about -13.471dB and significantly less than the L and R channels which peak at about 0dB. So this particular disc very much appears to be mastered expecting the LFE channel to be raised by 10dB so that it would peak at -3.471dB.

[EnglishTiger]

But and to repeat again there is seemingly no standard for AUDIO audio and it seems some content creators will use the VIDEO audio standard as a guide but some have ignored it as shown in my previous example. I inspected another of my DVD-Audio discs and this time the LFE peaked about -13.471dB and significantly less than the L and R channels which peak at about 0dB. So this particular disc very much appears to be mastered expecting the LFE channel to be raised by 10dB so that it would peak at -3.471dB.

How do you know it was mastered expecting the LFE channel to be raised by 10dB so that it would peak at -3.471dB? Maybe it was mastered that way because either the mixing or mastering engineer wanted the LFE values to be lower than for the other channels.

Media Players can only react to the information fed to them from the actual media and since their is no Relevant Standard for AUDIO nobody is capable of developing a platform that alters the input data based on assumptions made by anyone other than the people who were responsible for the creation of that Media.

[stewart_pk]

How do you know it was mastered expecting the LFE channel to be raised by 10dB so that it would peak at -3.471dB? Maybe it was mastered that way because either the mixing or mastering engineer wanted the LFE values to be lower than for the other channels.

I don't know and this is the whole premise of this thread.
JRSS assumes it does know and well and truly applies that 10dB boost when downmixing. 

[stewart_pk]

Media Players can only react to the information fed to them from the actual media and since their is no Relevant Standard for AUDIO nobody is capable of developing a platform that alters the input data based on assumptions made by anyone other than the people who were responsible for the creation of that Media.

I can think of perhaps some ways to raise the LFE to an acceptable level if required based on JRiver's audio analysis. Perhaps it could effectively adjust up or down the LFE channel to peak at a standard level say 0dB, -3dB, -6dB or perhaps even let the user specifiy. It could be great and safe, I think, I'm just brainstorming now.
But I didn't ask for anything as full-on as that in my initial request. 

[Graham Hitchcox]

I used JRiver or MC from version 16+, but somehow everything went south?

I hope the JR team could revive this, or give a valid reason why its not supported any more!
Respect Graham