Print

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

[mojave]

My hard drive crashed recently and I had to start from scratch. This was a work computer so personal stuff, like MC and music, was not backed up. I'm finally getting around to testing the latest Bass Management crossovers.

I downloaded and installed the Voxengo Span VST plugin to check the frequency response and levels. I also downloaded this Pink Noise file from Ethan Winer's website. I set it to repeat in MC and then check things out in Span while playing with Room Correction. The default settings in Span have a tilted frequency response. To get a flat response one needs to click the Edit button in the top right corner of the plugin and change the slope to 2.90. I change the Freq Low to 60 and the Freq High to 2.00K so the window is narrower. Also, the pink noise starts to roll off at 60 Hz. I then change the Type (still in Edit in Span) to Avg so I get a smoother line in the spectrum. I do this for both the L/R group and the LFE group.

First I played the pink noise with the L & R channels combined and Room Correction off to make sure I had a flat frequency response. Then I enabled Room Correction with the Bass Management set to the following:  Crossover of 200 Hz (to better see what is going on), Move, 24 dB/octave for both High and Low Pass filters. The bass is rolling off and is not as loud.

I then kept increasing the Subwoofer channel in Room Correction until I was back to a flat frequency response. I had to increase the gain by 6 dB in order to have a flat frequency response. Remember, this is the electronic signal and has nothing to do with the speakers, subwoofers, and room. You are trying to reproduce the original signal, but with it split to the the mains and the subwoofer.

It may sum correctly with a multichannel source, but it isn't summing correctly with a stereo source. I'll check the multichannel stuff next.

I also noticed that all crossover settings were lower in frequency than what was set. When the Subwoofer channel had the gain increased, then the crossover was closer to its actual setting.

[mojave]

Here are some more charts showing the crossover.

[mojave]

Here is the final chart with L, R, LFE combined and the Subwoofer with a 6 dB gain.

[Matt]

I then kept increasing the Subwoofer channel in Room Correction until I was back to a flat frequency response. I had to increase the gain by 6 dB in order to have a flat frequency response. Remember, this is the electronic signal and has nothing to do with the speakers, subwoofers, and room. You are trying to reproduce the original signal, but with it split to the the mains and the subwoofer.

Remember that the moved bass gets attenuated by 10 dB.  This is done because the LFE channel should be 10 dB louder at equivalent signal levels.

But since there's bass coming from the left and right, together they're shifted roughly +6 dB.

The math looks like:
LFE = ([Left bass] - 10 dB) + ([Right bass] - 10 dB)

So the results you're posting seem reasonable to me.  Could you explain in more detail if you disagree?

Thanks.

[Matt]

I also noticed that all crossover settings were lower in frequency than what was set. When the Subwoofer channel had the gain increased, then the crossover was closer to its actual setting.

Here's a quote from AlexB that might explain this:

This is probably obvious and you may already have it correct, but at the crossover frequency the curves are supposed to cross at -6 dB.

[Matt]

Also, thank you for all your testing and help with this feature.  I really appreciate it.

[mojave]

Remember that the moved bass gets attenuated by 10 dB.  This is done because the LFE channel should be 10 dB louder at equivalent signal levels.

But since there's bass coming from the left and right, together they're shifted roughly +6 dB.

The math looks like:
LFE = ([Left bass] - 10 dB) + ([Right bass] - 10 dB)

So the results you're posting seem reasonable to me.  Could you explain in more detail if you disagree?

Thanks.

Let's solve the math.

LFE = ([Left bass] - 10 dB) + ([Right bass] - 10 dB)
LFE = ([Left bass @ 75 db] - 10 dB) + ([Right bass @ 75 db] - 10 dB)
LFE = (75 dB-10 dB)+(75 dB-10 dB)
LFE = (74.99 dB) + (74.99 dB)
LFE = 81.015714067 dB

L @ 75 + R @ 75 also equals 81 dB so the bass matches the mains (at least on paper). Therefore, I agree completely with what you are saying. When I electronically take the L + R + LFE channel, though, the bass is 6 dB too low as can be seen in the chart in the first post.

I couldn't find another multichannel spectrum analyzer. I double checked my settings and ran everything again. The bass is still 6 dB too low. This is confirmed two ways. First, the crossover is at the wrong frequency until the the bass is increased. Second, the combined channels show a drop off in volume well below the crossover region.

Here's a quote from AlexB that might explain this:

Yes, the curves do cross at around -6 dB as can be seen in my charts. The problem is that the bass does not come back up to the correct volume.

Again, the goal is to have the original signal look exactly the same after the stereo channel has been split to the LFE channels and then all are combined again in the spectrum analyzer. This gives an indication of what it will sound like with a properly calibrated system (subwoofer and speakers set to the same volume level).

Either I'm measuring wrong or you are summing wrong.

[Matt]

Let's solve the math.

LFE = ([Left bass] - 10 dB) + ([Right bass] - 10 dB)
LFE = ([Left bass @ 75 db] - 10 dB) + ([Right bass @ 75 db] - 10 dB)
LFE = (75 dB-10 dB)+(75 dB-10 dB)
LFE = (74.99 dB) + (74.99 dB)
LFE = 81.015714067 dB

L @ 75 + R @ 75 also equals 81 dB so the bass matches the mains (at least on paper). Therefore, I agree completely with what you are saying. When I electronically take the L + R + LFE channel, though, the bass is 6 dB too low as can be seen in the chart in the first post.

I got a little lost there.

Shouldn't it be:
75 dB - 10 dB = 65 dB
65 dB + 65 dB = 71 dB

It almost seems like we should average the redirected bass instead of sum it, so that it would always be -10 dB instead of, in this case, -4 dB.

[mojave]

Just like 65 plus 65 equals 71 dB instead of 130, you can't just subtract 10 dB from 75 to get 65 dB. You have to convert to Pascals and then add/subtract and then convert back to decibels. Here is one for doing the conversion. There are others available, too.

Another thing to consider is that I am talking about 2 channel to 2.1. When you combine with an already existing LFE channel the level mixing is different.

[Matt]

When we attenuate the LFE by -10 dB, it means we multiply by about ~0.32 in the PCM space (which is analogous to voltage / pressure).

So with one volt:

LFE = ([Left bass] - 10 dB) + ([Right bass] - 10 dB)
LFE = (1.0 * 0.32) + (1.0 * 0.32)
LFE = 0.64

1.0 volt becoming 0.64 is about a -4 dB shift, which matches my previous post.

So I'm still not clear on the problem.  Could you spell it out in a little more detail, maybe with math?

[Matt]

you can't just subtract 10 dB from 75 to get 65 dB

Maybe -10 dB means something different in different contexts.

To me, it means the output signal is ten decibels less than the input signal.  So you can "just subtract" as that's the point.  Of course, you have to take care to do the math correctly to ensure a -10 dB shift.

Is it possible you're thinking about adding a signal that's 10 dB quieter or something?  Is there some other angle to this?

[Alex B]

I think you would need to do

(0.707946 x L) + (0.707946 x R)

to get a flat frequency response curve when the bass from 2.0 is redirected to 0.1.

Before the redirection you had two output channels that added to each other. After the redirection you have only one output channel so it should receive a louder signal than one of the two outputs channels received before (the above math results ~1.41).  [edited this part a bit]

Of course you can't go over 0 dBFS in the digital domain so you would need to attenuate all channels to compensate.

This is done because the LFE channel should be 10 dB louder at equivalent signal levels.

Matt, isn't this the Dolby standard for movie theaters? I don't think it is normally assumed that the subwoofer speaker is set to be 10 dB louder than the other speakers. For instance, when a 5.1 channel audio recording is mixed, I think the assumption is that the speakers are configured to produce a flat response curve.

I suggested the LFE mixer so that the balance between the redirected bass and the bass that comes directly from the source 0.1 channel could be freely adjusted.

Here's a quote from http://www.hometheaterhifi.com/technical-articles/163-the-misunderstood-01-lfe-channel-in-51-digital-surround-sound.html :

Stop!  Before you run home and set your subwoofer 10 dB higher than you've already set it, you need to read on and find out why this is not necessary.

[Matt]

Matt, isn't this the Dolby standard for movie theaters? I don't think it is normally assumed that the subwoofer speaker is always set to be 10 dB louder than the other speakers. For instance, when a 5.1 channel audio recording is mixed, I think the assumption is that the speakers are configured to produce a flat response curve.

I suggested the LFE mixer so that the balance between the redirected bass and the bass that comes directly from the source 0.1 channel could be freely adjusted.

Here's a quote from http://www.hometheaterhifi.com/technical-articles/163-the-misunderstood-01-lfe-channel-in-51-digital-surround-sound.html :

So does anyone know if AC3Filter or FFDShow already boost the LFE by 10 dB relative to the other channels in analog output mode? 

It didn't seem to me like they would because then you either have to worry about clipping or have to attenuate all the other channels -10 dB.

[Matt]

Of course you can't over 0 dBFS in the digital domain so you would need to attenuate all channels to compensate.

Isn't this another argument for the 10 dB shift on the subwoofer channel?

Otherwise, we're going to have to really crank down the volume on all the other channels if we're redirecting from 5.1 or 7.1 so the subwoofer can't clip.

[Alex B]

So does anyone know if AC3Filter or FFDShow already boost the LFE by 10 dB relative to the other channels in analog output mode? 

It didn't seem to me like they would because then you either have to worry about clipping or have to attenuate all the other channels -10 dB.

Google search:  +10dB LFE site:.ac3filter.net

Seems like it has been discussed there. I have not yet read the discussions.

[Alex B]

Otherwise, we're going to have to really crank down the volume on all the other channels if we're redirecting from 5.1 or 7.1 so the subwoofer can't clip.

I think you have to do that. Here's an example from: http://ac3filter.net/guides/mixing_matrix

Matrix normalization

Note that in the example above we mix several input channels into one output channel. But imagine a loud 'boom' sound reproduced by all channels. In this case when we sum several channels into one, overflow may occur.

Overflow means distortions (see 'Loudness and dynamic range' article for details). What to do if we don't want distortions? We should decrease the sound level. To do this we may decrease Master gain, or lower all values in the matrix. To do this automatically, use matrix normalization. See the result of normalization on 5.1 to stereo downmix matrix:

Not normalized:



Normalized:



Normalized matrix looks more complicated, but in fact it does exactly the same thing as unnormalized, just a little quieter. Normalization is enabled by default. For clearness we will use unnormalized matrix afterwards, but for real use, it is better to use normalized matrix.

Maximum level of an output channel equals to the sum of the row elements:

Code: [Select]

MaxLevel(channel) = Sum(Row(channel))
To avoid overflow we must make the level of all output channels <= 1. To do this we should divide all matrix elements to the maximum level among all channels:

Code: [Select]

MaxLevel = Max(MaxLevel(channel)) NormalizedMatrix[i][j] = 1/MaxLevel * Matrix[i][j]
That's exactly what matrix normalization option does.

[Matt]

I think you have to do that. Here's an example from: http://ac3filter.net/guides/mixing_matrix

We use normalized mixing matrixes for JRSS downmixing, and I think you're right that we need to do the same here as well.

[Matt]

Google search:  +10dB LFE site:.ac3filter.net

Seems like it has been discussed there. I have not yet read the discussions.

From my reading, it sounds like it assumes either your amp or subwoofer will make the LFE 10 dB louder.

So that means when mixing from 2.0 to 2.1 with bass redirection, the LFE should be attenuated by 10 dB so it matches the levels you would get when playing a movie.

I suppose we could make additional options for "Move / Copy bass to subwoofer (LFE) at -0 dB".

[Matt]

I just sketched out the updated math, and I think I have a handle on how to proceed.

We'll preserve energy when redirecting bass (AlexB's Sqrt(2) example), which is a little different than what we're doing now.

We'll then add an "avoid clipping" stage that attenuates all channels equally based on the maximum possible gain. 

This raises a question: if you turn on room correction and just pick "+3 dB for the center speaker", is it confusing if all the other speakers turn down in volume instead of boosting the center?  This would be the safe way to do it, but I think it makes calibration complicated.

We could also make the sliders only allow negative gains, but this makes it a pain to boost a speaker.

[Alex B]

From my reading, it sounds like it assumes either your amp or subwoofer will make the LFE 10 dB louder.

I have not done my reading yet, but since in this case MC is the preamp/processor I think it should perform the +10 dB trick unless the used decoder filter has already done it.

So that means when mixing from 2.0 to 2.1 with bass redirection, the LFE should be attenuated by 10 dB so it matches the levels you would get when playing a movie.

I suppose we could make additional options for "Move / Copy bass to subwoofer (LFE) at -0 dB".

Could the option be: "Adjust a separate LFE source +10 dB when it is mixed with redirected bass (needed to conform Dolby specs)"? Naturally you would also then need to prevent overflow.

[Alex B]

I still think that it would be good to have a way to adjust the balance of the redirected bass content and the original 0.1 LFE content freely. Following strictly the standard (or standards?) doesn't necessarily produce a well-balanced, enjoyable result.

Here's my old "submixer" graph:




Here is an attempt to visualize its UI:




However, perhaps the same could be achieved with a single LFE boost adjustment. You would still be able the adjust the master subwoofer level separately and thus fine-tune the balance:

[Matt]

I still think that it would be good to have a way to adjust the balance of the redirected bass content and the original 0.1 LFE content freely.

You may be right.  I think this whole area is complicated. 

So when possible, it's good to make the program smart so the user doesn't need an audio engineering degree to get ideal sound.

Following strictly the standard (or standards?) doesn't necessarily produce a well-balanced, enjoyable result.

I am finding this to be true.  The LFE level between popular television shows or different movies can be quite disparate.

I think putting the LFE gain on the OSD so I could adjust it from the couch with my remote would be nice.

[mojave]

It looks like there has been some good progress on this over the weekend. I really don't know the answers to your earlier questions, Matt, but it sounds like to might have come up with someone else for the next build.

Bass management has caused problems for many companies. There is an old thread at AVSForum called LFE, subwoofers and interconnects explained which pointed out the problems caused by receivers having to decode SACD, DVD-Audio, DVD, Blu-ray, and HD DVD and have consistent bass output levels.

Audioholics had an article called 0dBFS & Bass Management of DVD / Blu-ray Players. I just found it today, and it discusses the very issue of summing of channels and then shows how they tested the Oppo BDP-83SE. The measurement of the Oppo looks very similar to my Crossover +6dB chart above. My chart is showing two channels instead of one and I was using 24 dB/octave filters for both the High and Low Pass instead of 12/24 like Oppo uses.

[Matt]

It looks like there has been some good progress on this over the weekend. I really don't know the answers to your earlier questions, Matt, but it sounds like to might have come up with someone else for the next build.

A coming build will give you the flat curve you're looking for.  We'll do an energy preserving accumulation of redirected bass.

We have to do a little internal plumbing first so that we know that the signal coming into Room Correcting is really 2.0 or 2.1 for music even though there are six or eight output channels.

There's still the issue of whether the LFE should be offset by 10 dB, but I think we'll start by making it possible to select a 0 dB or 10 dB shift since these are the only two "standards" I'm aware of.

[mojave]

A coming build will give you the flat curve you're looking for.

I'm looking forward to seeing what a flat curve looks like. 

[Matt]

We've got the updated algorithms mostly squared away.  However, I have one remaining question:

Imagine playing 5.1 from a movie.  There is redirected bass adding to the 0.1 from 5 channels.

To preserve energy, you would sum the five channels and multiply the sum by (1 / Sqrt(5)) or 0.45.

However, I'm not sure how this should combine with the existing 0.1.  I can see three different arguments:

1) All six channels are the same -- sum the LFE and the 5 redirected bass values then multiply by (1 / Sqrt(6))

2) Combine the LFE and redirected bass in an energy preserving manner -- add the two values then multiply by (1 / Sqrt(2))

3) Simply add the LFE and the redirected bass together.  No mixing done.


I think #3 is probably best or else the 0.1 channel would get attenuated as a result of redirected bass, which I don't think would be expected.

Any second opinions?

[flac.rules]

I think receivers do number 3? They just take everything below crossover in all channels and send it to the sub. (after boosting the LFE 10 dB), so if you add bass to LFE from orther channels, it shoudl be attenuated 10 dB? BTW, afaik, music or movies shouldn't make a difference for a receiver, I think most receivers will boost the LFE-channel 10 db if receiving a 5.1 signal, so default should(?) to attenuate it by 10 db.

[Alex B]

I'd say #3.

A 0.1 source is a single source channel. One channel in => one channel out.  No calculations for preserving its energy are needed.

The +10 extra gain is a separate issue. I've got the impression that typical HT receivers do it when Dolby* multichannel source with 0.1/LFE channel is decoded to PCM, but not when analog 5.1 or 7.1 signal is fed directly to the receiver's multi-channel inputs and digital signal processing is bypassed.

(* ...and perhaps also DTS, but I have not seen any documentation that would explain how LFE in DTS encoded signal should be handled.)

[Alex B]

I have been thinking about a situation in which low bass in the 5.0 audio content (i.e. not in the 0.1 channel) is mixed to come from a certain direction instead of mixing it more or less evenly to all five channels.

I created a sample in which only the front left channel of the five contains low bass (a combination of 30 Hz and 60 Hz sine waves). The audible volume level of this content should be preserved when it is redirected, but I don't think that will happen when the equation is (1 / Sqrt(5)).

I added also 0.1 aka LFE content to the sample (the same signal as above), but as already said, this part should be a no-brainer. One LFE in => one LFE out, except if the optional +10 dB gain is applied to this signal.

The 3 second sample in the attached zip package is in the 5.1 FLAC format.

[Matt]

I have been thinking about a situation in which low bass in the 5.0 audio content (i.e. not in the 0.1 channel) is mixed to come from a certain direction instead of mixing it more or less evenly to all five channels.

I created a sample in which only the front left channel of the five contains low bass (a combination of 30 Hz and 60 Hz sine waves). The audible volume level of this content should be preserved when it is redirected, but I don't think that will happen when the equation is (1 / Sqrt(5)).

I added also 0.1 aka LFE content to the sample (the same signal as above), but as already said, this part should be a no-brainer. One LFE in => one LFE out, except if the optional +10 dB gain is applied to this signal.

The 3 second sample in the attached zip package is in the 5.1 FLAC format.

Thanks.  What are you recommending?

A simple stereo addition example looks like this:

L = 1 volt = 1^2 units of power = 1 unit of power
R = 1 volt = 1^2 units of power = 1 unit of power
Simple addition: L + R = 2 volts = 2^2 units of power = 4 units of power
Power preserving addition: Sqrt(2) * (L + R) = 1.4 volts = 1.4^2 units of power = 2 units of power

Doesn't this mean, by definition, that if we attenuate to preserve power (not voltage) we would attenuate any single channel?  Is there something we're missing?

[Matt]

Alright, I think I have it.  Let me know if this makes sense:

5.0 source, redirecting bass to the 0.1 channel (assume 0.1 channel is silent for now) should use this equation:
L^2 + R^2 + C^2 + SR^2 + SL^2 = LFE^2

Assuming values from 0 to 1.0, the maximum LFE is then Sqrt(5) or 2.24.  This means to avoid clipping, all values must be multiplied by (1 / 2.24) or 0.45.

So, in AlexB's example of a single channel with full intensity sound:
L^2 + R^2 + C^2 + SR^2 + SL^2 = LFE^2
1.0^2 + 0^2 + 0^2 + 0^2 + 0^2 = LFE^2
1 = LFE^2
LFE = 1

Then, to adjust for attenuation:
LFE = 1 * 0.45 = 0.45.

The original left speaker also has a range of 0 to 0.45 after the attenuation.  This means the movement of the bass from AlexB's example results in no difference in voltage or power.

Does this all check out?  Basically we need to do squared summing with the root at the end.

[Alex B]

I think you are correct. Naturally also the original (now high passed) channels will get the same attenuation.

A practical test for e.g. L + R could be:

When the output is measured with an SPL meter in a listening room which method will produce the same reading from the single speaker as from the two original speakers? (Assuming all speakers are identical full range speakers and the room is an anechoic chamber.)

[Alex B]

The +10 dB LFE gain issue may still need to be considered. I found some indications that actually the subwoofer (or its poweramp) should be set +10 dB louder and when bass is redirected the final resulting mix should be corrected by -10 dB. The article that I referenced earlier made me believe that the Dolby decoder adds the +10 dB gain, but that may not be the case after all.

Here are a few more links that may be helpful:

http://www.ultimateavmag.com/howto/805bass/index.html
http://www.avsforum.com/avs-vb/showthread.php?t=748147
http://www.hometheatershack.com/forums/home-audio-subwoofers/1736-bass-effects-downmixing-dd-stereo-spectrum-labs-information.html

[flac.rules]

I think you are correct. Naturally also the original (now high passed) channels will get the same attenuation.

A practical test for e.g. L + R could be:

When the output is measured with a dB meter in a listening room which method will produce the same reading from the single speaker as from the two original speakers? (Assuming all speakers are identical full range speakers and the room is an anechoic chamber.)

Assuming uncorrelated sound from the speakers (which is the most corret assumption i guess), it would be having 3dB less sound power in the two speakers. That is half the sound power. So the "correct" thin is to just add togheter all the sound power into the LFE-channel. However as mentioned, it needs to be normalized (all channels, full bandwidth), to make sure that the LFE-channel dont get clipped.

[flac.rules]

The +10 dB LFE gain issue may still need to be considered. I found some indications that actually the subwoofer (or its poweramp) should be set +10 dB louder and when bass is redirected the final resulting mix should be corrected by -10 dB. The article that I referenced earlier made me believe that the Dolby decoder adds the +10 dB gain, but that may not be the case after all.

Here are a few more links that may be helpful:

http://www.ultimateavmag.com/howto/805bass/index.html
http://www.avsforum.com/avs-vb/showthread.php?t=748147
http://www.hometheatershack.com/forums/home-audio-subwoofers/1736-bass-effects-downmixing-dd-stereo-spectrum-labs-information.html

A properly set up receiver getting digital in should boost the channel 10 dB by itself. So the LFE-channel needs to be 10 dB lower.  Correcting it in the sub is not a good solution imho.

[Alex B]

Actually, I meant just that the subwoofer may need to be set 10 dB louder than the other speakers when all signal processing is bypassed and only the last "volume knobs" are in the use. It doesn't really matter where these "knobs" are. They can be inside MC, in other software after MC, or in external hardware. The important thing is that the redirected bass content may need to be adjusted by -10dB before it reaches the final speaker adjustments and before it is mixed with the separate LFE channel. (edited this a bit)

For instance, according to what I have read lately, the signal that is used in the automatic speaker calibration systems in HT receivers (like in the YPAO system in my Yamaha receiver) is actually 10 dB lower in the LFE channel. This will automatically adjust the final subwoofer output to be 10 dB louder.

When we are feeding the analog outputs more of less directly to the poweramps we still need to calibrate the speakers somehow. It might be advantageous to preset the volume knob on the subwoofer to a quite loud position.

[mojave]

Some Blu-rays have 7.1 audio, but most still have 5.1. I choose to upmix using JRSS all 5.1 to 7.1, but I still listen to 2.0 as 2.1. There may be some that don't choose to upmix and will listen in 7.1, 5.1, and 2.1. For these situations, you need to start your calculations with redirected bass from 7 channels and use the maximum headroom available. You will then need to further attenuate the mains/surrounds for 5.1/2.1 to match the levels from the 7 channel calculations. This is necessary in order for a calibrated system to be played back at the same levels regardless of the how many channels are being played.

In practical use, though, I think I would rather have each mode (7.1, 5.1, and 2.1) use the maximum headroom. Those of us using bass management probably aren't using a receiver. We don't have our volume control calibrated for reference levels (at least I don't) and we can just adjust the volume to our liking.

 

[flac.rules]

Actually, I meant just that the subwoofer may need to be set 10 dB louder than the other speakers when all signal processing is bypassed and only the last "volume knobs" are in the use. It doesn't really matter where these "knobs" are. They can be inside MC, in other software after MC, or in external hardware. The important thing is that the redirected bass content may need to be adjusted by -10dB before it reaches the final speaker adjustments and before it is mixed with the separate LFE channel. (edited this a bit)

For instance, according to what I have read lately, the signal that is used in the automatic speaker calibration systems in HT receivers (like in the YPAO system in my Yamaha receiver) is actually 10 dB lower in the LFE channel. This will automatically adjust the final subwoofer output to be 10 dB louder.

When we are feeding the analog outputs more of less directly to the poweramps we still need to calibrate the speakers somehow. It might be advantageous to preset the volume knob on the subwoofer to a quite loud position.

The signal strength used during calibration has no direct bearing to the final output, the calibration playes a sound at a given level, and calibrates the sub so the level at the microphone is right. At least in principle. You can calibrate to excatly the same lever using a 70 dB signal as a 80 dB signal, as long as the software actually knows what level the signal is played at. That being said playing the LFE-channel lower in calibration makes no sense to me, why would you do that? The whole point of the LFE-channel is that is louder, the max SPL-level is louder, because it gets boosted 10 dB by the receiver. (higher max spl traded against higher noise floor). Besides background noise is much higher at lower freqs, frankly it seems like a stupid design. If you have the pwer, its always best to calibrate at a the highest volume pracitcal, and its not like adding or subtracting 10 dB needs to be done in the calibration itself.

[Alex B]

I think we are speaking about the same thing in different words.

Regarding the automatic adjustment systems, I meant the built-in hard coded test signals, not something that is user adjustable.

Here is a quote from my "AVS-forum" link:

Introduction
Initially, I'm only going to consider LFE in a system without bass management.

The LFE channel was originally found on multi-channel film recording formats, such as Dolby Stereo 70 mm Six Track. The purpose was to make more room for loud explosions, rumbles etc. The normal sound channels were calibrated so they could produce up to about 105dB SPL of output. Pretty loud, but not quite enough for the sort of low-frequency effects that you feel as much as hear.

So, to increase the impact, they added an extra channel, dedicated to low frequency effects, to be sent to a dedicated subwoofer. And this channel was calibrated to play 10dB louder than the rest - it could output 115dB SPL of energy. And 10dB difference is quite a lot - it means a signal over 3 times the amplitude.

Calibration
This 10dB boost is achieved by calibration in the monitoring amplifiers at the studio; a full-scale LFE signal on the tape is set up to play 10dB louder than a full-scale signal on any other channel. The cost is increased noise - the channel has been turned up, so the general hiss and noise on the magnetic recording will also be 10dB louder. But this is not too much of a problem, as all sounds above 150Hz are filtered out anyway on playback. The channel is only handling low frequencies.

To reproduce the effect heard in the studio, cinema playback systems are also calibrated the same way - a 10dB boost is added to the LFE channel amplifier.

Calibration tests generally hide this 10dB difference; a basic pink noise test will contain a signal that should play at 75dB SPL for each channel. For the normal channels, that means a signal 30dB below full scale (105dB-30dB = 75dB), but for the LFE channel it is a signal 40dB below full scale (115dB-40dB = 75dB): 1/3 of the signal amplitude. If those test signals are used to calibrate the amplifier so they all output at 75dB SPL, then everything is set up okay. The LFE channel now has 10dB of extra headroom above the other channels and can produce the full 115dB SPL when required.

Interconnects
One important point is that the 10dB boost to the LFE is performed at the final amplification stage. The potential extra volume of the LFE channel could not pass through interconnects, analogue recordings or digital recordings without going over the specified limits by a factor of 3. This, indeed, is the whole raison d'ętre of the LFE channel. It's a channel where by convention everything is recorded 10dB low to make room for louder bangs.

So at all points in a system - analogue interconnects, PCM signals, encoded Dolby Digital - the LFE channel is "10dB too low". A 75dB SPL signal on the LFE channel will be 3 times, or 10dB, smaller than a 75dB SPL signal on the other channels. It will only sound correct when fed to an amplifier that amplifies LFE by 10dB more than it amplifies the other channels.

This 10dB difference is dictated by Dolby and DTS, and the international standard ITU-R BR.1384-1:
Quote:The LFE channel is recorded with a level offset of –10 dB. This offset is compensated for in the reproduction system, where the LFE loudspeaker has an acoustic output (within its low frequency passband) of +10 dB with respect to the other channels.

In the home
The LFE channel progressed from early analogue magnetic formats, to the digital channel in cinematic versions of Dolby Digital, and in turn to the home versions of Dolby Digital, DTS and MPEG Multichannel found on DVDs.

The rules remain the same: the encoded LFE signal needs a 10dB boost. In receivers with integrated decoders, this is handled automatically in the internals, and the user is generally unaware of it. It is only when we try to connect amplifiers to external decoders that things can go wrong. Many receivers do not offer the necessary boost for LFE coming in from an external decoder.

LFE level adjustment
Some receivers with integrated decoders do offer an adjustment for the LFE level - an option to NOT boost the LFE by 10dB. This is needed for some early music mixes using DTS; the studio mixers, being unaccustomed to 5.1 production, hadn't calibrated their LFE channel 10dB high. Thus the recorded LFE signal on those discs is at equal level with the other channels, and doesn't require boosting.

The unboosted LFE option on the receiver is often called "LFE -10dB", with the standard boosted option being "LFE 0dB". Alternatively the options might be called "LFE 0dB" and "LFE +10dB". Whatever, the higher option is correct for films and almost all DD and DTS multichannel music. Since those early errors, both Dolby and DTS have specifed that LFE in music releases should be mixed for a 10dB LFE boost to match film soundtracks. However, SACD still differs... more on this below.

The article continues by explaining bass redirection. It is a good read.
http://www.avsforum.com/avs-vb/showthread.php?t=748147

[mojave]

Hey, Alex, I already linked to that thread in post #22.   

The thread does have good information and points out why receivers boost the LFE by 10dB before combining with the redirected bass. A Guide to Bass Management over at the Blu-ray forum also has some good information and charts on how receivers handle bass management. All charts show the LFE channel being boosted by 10dB before being combined with redirected bass.

[Alex B]

Hey, Alex, I already linked to that thread in post #22.  

Oops...

... All charts show the LFE channel being boosted by 10dB before being combined with redirected bass.

Correct. (...or the redirected bass is attenuated by -10 dB and the combined signal is boosted - the result is the same.)

When MC handles all processing the following options are necessary:

      LFE     redirected

1.  +10          0        default

2.     0           0        the content is mixed without taking the +10 dB boost into account

3.     0        -10        the boost happens after MC in some other component

4.  -10        -10        the content is mixed without taking the +10 dB boost into account & the boost happens after MC in some other component


The #3 and #4 can be achieved by reducing the speaker volume by 10 dB so maybe they are not needed separately. In any case, if the +10 dB correction is applied outside MC in the audio HW then the clipping prevention adjustment can be less excessive.

[Matt]

We've added this to a coming build:
NEW: Added DSP Studio > Output Format subwoofer option "Calibrated to +10 dB (Dolby standard)". (used by JRSS and also Bass Management)

If you're redirecting bass, I would recommend making sure something (amp or subwoofer itself) is boosting by 10 dB so you can enable this option.  The extra headroom means we have to turn other channels down a lot less for clip protection.

[mojave]

The +10 dB correction cannot take place outside MC for the LFE track if it is already combined with the redirected bass. The only way this could happen is if you were routing the redirected bass to a different output than the LFE channel. Once it is combined, there is now way to only adjust the LFE track. Movie theaters and the studio can do it because only the LFE track is going to the subwoofer.

I'm curious how downloaded 5.1 music, like from iTrax, is mixed. I know SACD did things different, but I don't think any one listens to an SACD on a computer. We may need option 1 & 2 or 3 & 4, but I don't think we need all four options. If DVD-Audio, DVD, and Blu-ray all attenuate the LFE track and downloaded music does, too, then we don't even need an option.

[Alex B]

We've added this to a coming build:
NEW: Added DSP Studio > Output Format subwoofer option "Calibrated to +10 dB (Dolby standard)". (used by JRSS and also Bass Management)

If you're redirecting bass, I would recommend making sure something (amp or subwoofer itself) is boosting by 10 dB so you can enable this option.  The extra headroom means we have to turn other channels down a lot less for clip protection.

So it can handle the #1 and #3. Correct?

[Alex B]

I think the #2 and #4 are essential because:

From: http://www.hometheaterhifi.com/technical-articles/163-the-misunderstood-01-lfe-channel-in-51-digital-surround-sound.html

DTS's LFE channel in consumer applications, unlike its cinema counterpart, is discrete but still has a few 'special' considerations.  In the early days of mixing 5.1 for DTS CDs, the studios were not being calibrated correctly, with the end result of the LFE channel being set too low.  When the material is played back on a correctly calibrated system, the LFE channel is way too high.  THX was forced to introduce DTS Music and DTS Movie playback modes which distinguish between a correct LFE setting and a -10 dB setting to compensate for material assembled under mis-calibrated circumstances (though not all THX processors offer this convenience).

[flac.rules]

I think we are speaking about the same thing in different words.

Regarding the automatic adjustment systems, I meant the built-in hard coded test signals, not something that is user adjustable.

Yeah, but it is adjustable to the manufacturer, test tones should be a given level, (or a few levels), and they should be as loud as possible, having a low level on bass test tones is not wise.

[flac.rules]

I think #2 and #4 are essential because:

From: http://www.hometheaterhifi.com/technical-articles/163-the-misunderstood-01-lfe-channel-in-51-digital-surround-sound.html

Hmm, let me think, i miscalibrated 5.1 track would have the same level on the LFE track as the rest, MC adds the extra data to the LFE track from other channels, so the relative bass level is ok betwenn normal channels and LFE, however, the LFE is still 10 db to high. The difference should maybe be wheter 10 db is subtractet BEFORE or AFTER the channels are added in to LFE. (before being the default with well behaved source material)

[mojave]

I think #2 and #4 are essential because:

I just e-mailed AIX Records/iTrax to see if #2 & #4 are necessary for their 5.1 music mixes.

[flac.rules]

The +10 dB correction cannot take place outside MC for the LFE track if it is already combined with the redirected bass. The only way this could happen is if you were routing the redirected bass to a different output than the LFE channel. Once it is combined, there is now way to only adjust the LFE track. Movie theaters and the studio can do it because only the LFE track is going to the subwoofer.

I'm curious how downloaded 5.1 music, like from iTrax, is mixed. I know SACD did things different, but I don't think any one listens to an SACD on a computer. We may need option 1 & 2 or 3 & 4, but I don't think we need all four options. If DVD-Audio, DVD, and Blu-ray all attenuate the LFE track and downloaded music does, too, then we don't even need an option.

Thats no problem, as long as channel bass is attenuated 10 dB before its added to the LFE track.

[Matt]

So it can handle the #1 and #3. Correct?

The option attenuates redirected bass by 10 dB, which I think is #3 (which is functionally identical to #1).  The option is enabled by default.  With the option off, you get #2 and #4 (we don't know if that native LFE line is assuming a +10 dB gain).

From what I can gather, FFDShow and AC3Filter (two popular AC3 / DTS decoders) both assume there's an external +10 dB gain on the 0.1 they output.  It may turn out to be best for us to just recommend (force) calibration that matches this assumption, so that we're not trying to accommodate too many permutations.