Print

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

[TheLion]

Just to be clarify, those sliders only adjust the relative volume of channels to one another.  In other words, if you do +20 dB to all the channels, it's the same as doing nothing.  Room Correction always targets a 0 dB volume change so that it doesn't cause clipping or clip protection in cases where there is not headroom.

You can add +20 dB (or any other amount) with the parametric equalizer.

Regarding what you say above:

My setup uses the following DSP steps: parametric EQ: Subwoofer (=LFE channel) +10db (this is standard configuration to bring the LFE channel up to the correct level) -> Convolution (Normalized per channel) -> Room Correction to set the relative volume of my 7.1 channels.

My question is: The LFE channel for multichannel movie content needs to be upped by 10 db per specification (LAV AUDIO decoder doesn't do that and I am not using output to an receiver which applies such a boost before final output). I need to add this +10 db before convolution because convolution does my bass management (when i would add 10db to the sub after convolution this would also increase the low-passed speaker channels, which would be very wrong ;-))

Is it possible that this intended and correct +10db boost of the LFE channel results in a digital level >0dbFS? In that case to avoid clipping all channels must be decreased in digital level. But this wouldn't be necessary because Room Correction is applying -19.2db to the subwoofer channel (relative to the center channel with 0db) as final DSP step. Therefor there would be more than enough headroom available in the end. Is your audio engine smart enough and always leaves enough headroom for +10db LFE after LAV Audio does the decoding and/or does your 64bit audio engine consider the "absolute headroom" of a channel (when attenuation is done as final step in Room Correction and other DSP filters before that would cause "clipping")?

[Matt]

My setup uses the following DSP steps: parametric EQ: Subwoofer (=LFE channel) +10db (this is standard configuration to bring the LFE channel up to the correct level) -> Convolution (Normalized per channel) -> Room Correction to set the relative volume of my 7.1 channels.

You should add the +10 dB to the subwoofer in Room Correction.  That way, it knows about the additional headroom (from the other changes) and can utilize it.

Said another way, remove extra volume stages elsewhere in the chain.  Use the 'Level' checkboxes in Room Correction and get all the pink noise playing at the same level by adjusting the 'Level' sliders there.

[TheLion]

You should add the +10 dB to the subwoofer in Room Correction.  That way, it knows about the additional headroom (from the other changes) and can utilize it.

Said another way, remove extra volume stages elsewhere in the chain.  Use the 'Level' checkboxes in Room Correction and get all the pink noise playing at the same level by adjusting the 'Level' sliders there.

Matt,

I cannot do that. As described above when I would add the +10db in Room Correction (=after convolution and therefor after bass management (I high pass my speakers at 80Hz using convolution)) I would increase the (relative) volume of the subwoofer output channel. Per multichannel spec ONLY the LFE channel needs to be increased by 10db. Therefor I need  to apply this before bass management (= only to the LFE channel instead of the LFE+low-passed speakers mix that is the subwoofer output channel).

But that is exactly my question: is it possible to change internal volume so that the additional headroom (due to level attenuation in Room Correction) of a channel is "known" throughout the audio chain, and therefor even to all filters before Room Correction? 

[mojave]

TheLion, I still don't understand why convolution creating the crossovers needs any internal gain on the LFE channel. I have setup about 15 different sub systems using JRiver in the past few months and I was able to add the necessary gain at the subwoofer with either an internal or external amp's gain control. There was only one case in which I didn't have enough gain at the amp and needed more in JRiver. If I was to use Audiolense and measure my system, it would all be the correct level because I increased the gain at the subwoofer amp. Audiolense or Acourate doesn't care where the gain comes from. It is using the measured signal, not the digital signal. Adding 10 dB within JRiver to the LFE channel defeats the whole purpose the LFE is -10dB in the first place. You lose 10 dB of headroom on all your other channels when you do this.

[TheLion]

TheLion, I still don't understand why convolution creating the crossovers needs any internal gain on the LFE channel. I have setup about 15 different sub systems using JRiver in the past few months and I was able to add the necessary gain at the subwoofer with either an internal or external amp's gain control. There was only one case in which I didn't have enough gain at the amp and needed more in JRiver. If I was to use Audiolense and measure my system, it would all be the correct level because I increased the gain at the subwoofer amp. Audiolense or Acourate doesn't care where the gain comes from. It is using the measured signal, not the digital signal. Adding 10 dB within JRiver to the LFE channel defeats the whole purpose the LFE is -10dB in the first place. You lose 10 dB of headroom on all your other channels when you do this.

Mojave,

we know that the LFE channel is mastered at -10db to provide 10db additional headroom relative to all other channels. So when you play eg. a pink noise signal at -20dbFS digital level for all channels it should read 85db at listening position for all speaker channels BUT 95db for the subwoofer/LFE. And now it gets complicated. The low-passed speakers routed to the subwoofer play at 85db level on the subwoofer (measured at listening position). The LFE channel routed to the subwoofer output channel plays at 95db level on the subwoofer. To make this work the digital level of the LFE channel needs to be increased by 10db somewhere in the playback chain. And no - when the gain of your subwoofer/amp is high enough you don't loose any headroom on the other channels by adding +10 db to the LFE channel.

Fact is that LAV decoder doesn't add the +10db to the LFE channel. Media Center doesn't add the +10db to the LFE channel. When you calibrate the subwoofer output channel +10db than the low-passed speaker mix is much too high.

I use AIX calibration disc as well as Disneys WOW to confirm the right level balance. Without adding 10db per parametric EQ my LFE channel is 10db down relative to the other channels but the low-passed speaker mix is spot on after Audiolense/Acourate does its job.

[Matt]

TheLion, this topic is complicated for you because you're doing bass management with convolution filters.  I think the convolution filters should understand that when they add something to the subwoofer channel, they need to do -10 dB on what they're adding.

In other words, I'd make the signal still expect a +10 dB boost to the subwoofer when coming out of convolution.  I'd do this +10 dB boost (replace +10 dB with the actual decibel measurement) with Room Correction.

mojave is suggesting you turn up the gain on the amplifier for the subwoofer.  This is good when it's possible, but it's not always possible.  For example, I use the same amplifier for the subwoofer as the other speakers, so I have like a 30 dB swing (with Room Correction) between the center and subwoofer channels.

[TheLion]

TheLion, this topic is complicated for you because you're doing bass management with convolution filters.  I think the convolution filters should understand that when they add something to the subwoofer channel, they need to do -10 dB on what they're adding.

In other words, I'd make the signal still expect a +10 dB boost to the subwoofer when coming out of convolution.  I'd do this +10 dB boost (replace +10 dB with the actual decibel measurement) with Room Correction.

mojave is suggesting you turn up the gain on the amplifier for the subwoofer.  This is good when it's possible, but it's not always possible.  For example, I use the same amplifier for the subwoofer as the other speakers, so I have like a 30 dB swing (with Room Correction) between the center and subwoofer channels.

Matt,

I understand your point. As you know I am using Acourate to do my filters for convolution. All my filters are "normalized on a per channel basis" (each channel filter has the frequency with the least required attenuation at 0dbFS digital level) - so I need to use Room Correction for setting relative channel volume instead of doing this with convolution. Your suggestion with making the convolution add the low-passed speaker mix at -10dbFS to the subwoofer output channel is completely correct (= the same as adding +10db to the LFE channel). But how can I do that? Only way to do it is by adapting the configuration file. Mine is:

48000 8 8 0
0 0 0 0 0 0 0 0
0.25 0.25 0.00 2.45 2.79 2.52 3.85 3.79
C:\EQ\Referenz-3db\Left-3_48.wav
0
0.0
0.0
C:\EQ\Referenz-3db\Right-3_48.wav
0
1.0
1.0
C:\EQ\Referenz-3db\Center-3_48.wav
0
2.0
2.0
C:\EQ\Referenz-3db\LFE_Flat-3_48.wav
0
3.0
3.0
C:\EQ\Referenz-3db\Left_Back-3_48.wav
0
4.0
4.0
C:\EQ\Referenz-3db\Right_Back-3_48.wav
0
5.0
5.0
C:\EQ\Referenz-3db\Left_Surround-3_48.wav
0
6.0
6.0
C:\EQ\Referenz-3db\Right_Surround-3_48.wav
0
7.0
7.0
C:\EQ\Referenz-3db\Subwoofer_Flat-3_48.wav
0
0.0 1.0 2.0 4.0 5.0 6.0 7.0
3.0

The last four lines are for the low-passed speaker channel mix. Which parameter do I need to set there in order to make your convolution engine apply -10db to all summed low-passed signals?

Thanks for your support! (although the question turns out to be a little off-topic ;-)

[mojave]

we know that the LFE channel is mastered at -10db to provide 10db additional headroom relative to all other channels. So when you play eg. a pink noise signal at -20dbFS digital level for all channels it should read 85db at listening position for all speaker channels BUT 95db for the subwoofer/LFE.

The pink noise should be at 85 dB for all channels including LFE. The pink noise for LFE in the AIX Blu-ray calibration disc is -10 dB per Mark Waldrep so that when 10 dB is added at the subwoofer it plays at 85 dB. Most other calibration discs (AVIA, 5.1 Audio Toolkit from Seaton Sound, AIX Calibration Disc) and methods I have seen also have the speakers and LFE all calibrated at the same level. The Digital Video Essentials disc is incorrect and has the LFE test tone at the wrong level.

And now it gets complicated. The low-passed speakers routed to the subwoofer play at 85db level on the subwoofer (measured at listening position). The LFE channel routed to the subwoofer output channel plays at 95db level on the subwoofer.

The LFE channel actually plays at 75 dB. All other channels are supposed to be attenuated 10 dB when mixed and then the final LFE level is adjusted depending on how many channels were mixed in with the LFE channel. If you boost the LFE by 10 dB first, then the levels will be off because sometimes just two channels are mixed in and other times 7 channels are mixed in to the LFE channel.

[Matt]

C:\EQ\Referenz-3db\Subwoofer_Flat-3_48.wav
0
0.0 1.0 2.0 4.0 5.0 6.0 7.0
3.0

This means add channels 0, 1, 2, 4, 5, 6, 7 to channel 3 (the subwoofer).

I think you would want this instead:

C:\EQ\Referenz-3db\Subwoofer_Flat-3_48.wav
0
0.31622776601683794 1.31622776601683794 2.31622776601683794 4.31622776601683794 5.31622776601683794 6.31622776601683794 7.31622776601683794
3.0

You're just saying to scale each input by 0.31622776601683794 (-10 dB).

[Matt]

For level calibration, I believe the best approach is to use the 'Level' checkboxes in Room Correction and adjust so that all the channels, subwoofer included, play at the same decibel level.

You don't want to factor in the bass redirection when calibrating levels, so this approach is better than a test file since a test file will have bass management applied.

[TheLion]

This means add channels 0, 1, 2, 4, 5, 6, 7 to channel 3 (the subwoofer).

I think you would want this instead:
You're just saying to scale each input by 0.31622776601683794 (-10 dB).

Thank you very much, Sir. I will try it at once. How to you calculate the necessary scaling factor for a given attenuation in db?

[TheLion]

For level calibration, I believe the best approach is to use the 'Level' checkboxes in Room Correction and adjust so that all the channels, subwoofer included, play at the same decibel level.

You don't want to factor in the bass redirection when calibrating levels, so this approach is better than a test file since a test file will have bass management applied.

But the level checkbox in Room Correction doesn't take convolution into account (when used after convolution in the filter chain), does it?

I use a much more elaborate method to level match my channels: REQ RTA while playing pink noise and routing it (per parametric EQ) to each channel individually. First step is to mute the subwoofer output channel and therefor channel match the high-passed 7 speaker channels. Then I match the subwoofer level (speaker low-pass) with any of the high-passed channels. Therefor all channels are perfectly matched with all filters applied (convolution).

[Matt]

But the level checkbox in Room Correction doesn't take convolution into account (when used after convolution in the filter chain), does it?

No.  If convolution is changing the volume per speaker, you'll need to use test tones that play through convolution.

I use a much more elaborate method to level match my channels: REQ RTA while playing pink noise and routing it (per parametric EQ) to each channel individually. First step is to mute the subwoofer output channel and therefor channel match the high-passed 7 speaker channels. Then I match the subwoofer level (speaker low-pass) with any of the high-passed channels. Therefor all channels are perfectly matched with all filters applied (convolution).

As long as you play a tone that's _only_ a subwoofer for the second part, that makes sense.  If you played a tone on all speakers for the second part, I think the redirected bass would cause trouble.

[Matt]

How to you calculate the necessary scaling factor for a given attenuation in db?

multiplier = 10.0^(decibels / 20.0)

[TheLion]

As long as you play a tone that's _only_ a subwoofer for the second part, that makes sense.  If you played a tone on all speakers for the second part, I think the redirected bass would cause trouble.

Yes indeed. I route a mono pink noise to each channel individually, not multiple channels at once. Given a flat frequency target the low-pass on the subwoofer channel must match the high-pass on the speaker channel in level. Upcoming revisions of Acourate will make multichannel convolution much easier and therefor level matching will be "automatic" (right now Acourate is doing each channel filter individually). Audiolense does automatic multichannel level matching.

[TheLion]

multiplier = 10.0^(decibels / 20.0)

Thank you! Perhaps when you sort out your own convolution configuration this can be made more "intuitive" ;-)

[mojave]

I ordered an iSEMcon EMM-7101-CHTB measurement microphone with custom calibration down to 3 or 4 Hz and an SMA to XLR phantom power adapter. It should be here next week. I hope to get Audiolense or Acourate soon and start down the convolution path.

[TheLion]

I ordered an iSEMcon EMM-7101-CHTB measurement microphone with custom calibration down to 3 or 4 Hz and an SMA to XLR phantom power adapter. It should be here next week. I hope to get Audiolense or Acourate soon and start down the convolution path.

Welcome to the club, mojave! As you know I have owned and used both programs for over a year now. My recommendation clearly is Acourate from a "sound quality" point of view (depending on your room and setup your experience may vary). When usability is a priority for you Audiolense is the better option.

[Matt]

Yes indeed. I route a mono pink noise to each channel individually, not multiple channels at once. Given a flat frequency target the low-pass on the subwoofer channel must match the high-pass on the speaker channel in level.

The hard part to doing it this way is that turning a level slider down in Room Correction could theoretically turn other speakers up, since the volumes are all relative.

I'm not sure how to handle this in an easier way.

[mojave]

Welcome to the club, mojave! As you know I have owned and used both programs for over a year now. My recommendation clearly is Acourate from a "sound quality" point of view (depending on your room and setup your experience may vary). When usability is a priority for you Audiolense is the better option.

Audiolense 4.3 is supposed to be released 3.4 (March 4  ) with some major changes. Here is what Bernt said on his forum:

Audiolense 4.3 will have a complete rework of everything related to time domain correction. The new methodology creates far less pre-ringing than the conventional approach. The actual method will remain a trade secret at least for now. Worst case simulations says 95% reduction and upwards. Truncation will still have a place, but the results should be much better and more reliable than before. And less truncation should be needed to get rid of any audible artifacts if there were any to begin with.

I've got it working with digital crossovers now. It produces very good results compared to 4.2 and earlier, but not as good as  I believe this new technology is capable of. As I work on this I am learning about implementation details that I didn't even knew existed. I will get out a release the week after next week. It will not be the end of it, but hopefully be a big step in the right direction on most systems. It will probably be a beta release with a few different blind options. I've spent a lot of time running some code, looking at results, tweaking the code, looking at the results and so forth. The stuff that I am looking into here is so far away from the common knowledge about DSP based sound correction that I don't even know if I am able to explain them. I will simply label them option 1...4 or something similar. I will need help from you guys to sort out what works best because I have a feeling that we're getting into a landscape where there may be a discrepancy between what simulates best and what sounds best. I also plan to start a thread at the forum with a tutorial about this release and hopefully we can have a discussion going. Expect a beta release by Sunday march 4th, before 12 am. Or at least before Monday march 5.th 4am.

[TheLion]

Audiolense 4.3 is supposed to be released 3.4 (March 4  ) with some major changes. Here is what Bernt said on his forum:
 

I discussed version 4.3 extensively with Bernt and I am looking forward to it. Then I will benchmark it against Acourate again. But the biggest difference is not only in pre-ringing artifacts (with my setup severe with Audiolense and none with Acourate) but also in the psychoacoustic model behind these two solutions. And here I have a clear preference.

[TheLion]

Matt,

with Internal Volume and convolution there are basically 3 options to level match the speakers:

1) the necessary gain/attenuation is "baked into" the DRC filter and therefor levels are matched during convolution. (This is how it is done when using eg. Audiolense)

2) if the individual DRC filters are normalized to 0dbFS (therefor are not correcting level diffeences between speakers) the level must be set after convolution:

a) by using scaling in the convolution configuration file (just as you advised me to do for bass management.

eg. when I need -10db to match my subwoofer channel with the other channels my example from before would read:

C:\EQ\Referenz-3db\Subwoofer_Flat-3_48.wav
0
0.31622776601683794 1.31622776601683794 2.31622776601683794 4.31622776601683794 5.31622776601683794 6.31622776601683794 7.31622776601683794
3.31622776601683794

b) by using the level setting in Room Correction after convolution. (This is how I do it atm)


Which of these options is recommended for best gain structure/headroom/"sound quality" when used with the "new" internal volume? When you have the choice between 2) a) or b) which one is to be preferred?

Thank you!

[Trumpetguy]

1) the necessary gain/attenuation is "baked into" the DRC filter and therefor levels are matched during convolution. (This is how it is done when using eg. Audiolense)

@TheLion,
I find this topic somewhat advanced and have some difficulties to understand if I have my setup correct or not. My understanding of Audiolense has always been what you stated, that the level (and timing) differences between the channels are levelled in the filter. And to me, both music and movie audio tracks sound well balanced. I use the DSP Output format plugin, with JRSS upmixing to 7.1. All frequencies are sent to the sub(s), since the AL filter takes care of the cross-over. I have ticked the "My source has correctly mixed LFE signal +10dB" (or something like that). I use ROHQ, i.e. LAV splitter and decoders. The audio is then routed (ASIO) to DAC and power amplifiers. No receiver, processor or receiver in between.

The question is then - is my sub level correct? If not, how to correct it?

[TheLion]

@TheLion,
I find this topic somewhat advanced and have some difficulties to understand if I have my setup correct or not. My understanding of Audiolense has always been what you stated, that the level (and timing) differences between the channels are levelled in the filter. And to me, both music and movie audio tracks sound well balanced. I use the DSP Output format plugin, with JRSS upmixing to 7.1. All frequencies are sent to the sub(s), since the AL filter takes care of the cross-over. I have ticked the "My source has correctly mixed LFE signal +10dB" (or something like that). I use ROHQ, i.e. LAV splitter and decoders. The audio is then routed (ASIO) to DAC and power amplifiers. No receiver, processor or receiver in between.

The question is then - is my sub level correct? If not, how to correct it?

Trumpetguy,

I haven't used Audiolense for quite some time now after my switch to Acourate. I did discuss this matter with Bernt. As far as I remember he isn't doing the +10db boost on the LFE channel. So you would need to manually add this before convolution per parametric EQ.

Please double check with a calibration disc - AIX or Disney WOW is recommended.

[Trumpetguy]

Trumpetguy,

I haven't used Audiolense for quite some time now after my switch to Acourate. I did discuss this matter with Bernt. As far as I remember he isn't doing the +10db boost on the LFE channel. So you would need to manually add this before convolution per parametric EQ.

Please double check with a calibration disc - AIX or Disney WOW is recommended.

Let's assume the +10dB boost is not done by Audiolense. It makes sense, since the channels are levelled, and you would not want a 10dB boost on multichannel music (I think....?). I have all my 7 high-pass channels (XO 80Hz on fronts and 120Hz on rear/sides) routed to the subs in addition to the LFE track (channel 3). The filter is 15 paths. The subs are thus receiving path 4 (ch.3, the LFE track) + the last 7 paths from the filter (the bass content from the speakers). Should only channel 3 (path 4) be boosted +10dB, or should this be done on all the paths routed to the subs?

I think this was discussed earlier in this thread, but I could not figure it out.

@Matt:
What is actually done in Output Format when activating or deactivating the "Source is correctly mastered for +10dB calibration" option?

[TheLion]

Only the LFE channel (path 4 in your case) needs the 10db boost, if the subwoofer output channel is level matched with the speaker channels.

[TheLion]

Only the LFE channel (path 4 in your case) needs the 10db boost, if the subwoofer output channel is level matched with the speaker channels (given the same digital level as input signal) .

[Trumpetguy]

Only the LFE channel (path 4 in your case) needs the 10db boost, if the subwoofer output channel is level matched with the speaker channels (given the same digital level as input signal) .

Thanks!

I still wonder if I can assume the same digital level from all channels when activating the "Source is correctly mastered for +10dB calibration" in DSP Output Format. Waiting for a reply from Matt.

[TheLion]

Thanks!

I still wonder if I can assume the same digital level from all channels when activating the "Source is correctly mastered for +10dB calibration" in DSP Output Format. Waiting for a reply from Matt.

No, it is the other way round. When activating the "Source is correctly ..." option the LFE channel will end up -10db digital level from all other channels after decoding (eg. LAV) (if disabled it will end up -20db). Audiolense level matches all channels so that the same digital input level results in the same volume at listening position for all channels. Now your LFE channel needs a 10db boost to have the same digital level as the other channels and therefor results in the same volume after convolution.

[Trumpetguy]

No, it is the other way round. When activating the "Source is correctly ..." option the LFE channel will end up -10db digital level from all other channels after decoding (eg. LAV) (if disabled it will end up -20db). Audiolense level matches all channels so that the same digital input level results in the same volume at listening position for all channels. Now your LFE channel needs a 10db boost to have the same digital level as the other channels and therefor results in the same volume after convolution.

If you are right, and I have no doubt that you are, I have during the past two years watched a billion movies with 10dB less fun in the bottom octave (at least in the wow! effects department) 

What about multichannel music? Is the .1 track mastered the same way?

[Buckster]

OP how are you getting on with clipping ?

I too am using 7.1 out - digital (via USB) to my amp - I use the crossover/EQ inside of my amp (all in the dgitial domain) so only have to do the +10dB on the LFE (as my amp is too old to apply it) - the 10dB I add by adding 10dB to the LFE channel in Room Correction

BUT I constantly have problems with clipping ....

I only stop getting clipping at almost the lowerst "pre-amp" setting available in the EQ filters - otherwise all action films constantly clip ...

problem is when you get to point of clipping being removed, the output is so low - that it doesn't sound right - yes my amp is quite beefy and I can get volume up to loud volumes still - but it seems to loose all dynamics

[Buckster]

btw - in latest versions I've found if you go to audio-path - info - it tells you how much "negative" gain is being applied to try and normalise to remove clipping

often on mine this can be anywhere down to -9dB - not good at all

[hulkss]

I too am using 7.1 out - digital (via USB) to my amp - I use the crossover/EQ inside of my amp (all in the dgitial domain) so only have to do the +10dB on the LFE (as my amp is too old to apply it) - the 10dB I add by adding 10dB to the LFE channel in Room Correction, BUT I constantly have problems with clipping ....I only stop getting clipping at almost the lowerst "pre-amp" setting available in the EQ filters - otherwise all action films constantly clip ...

Cinema audio content in the LFE channel is commonly recorded all the way to 0 dBFS which will produce a 115 dB SPL with reference level acoustic gain. You can not add any more digital level with out clipping as you have found out. Instead, you should boost the analog gain of your LFE channel by 10 dB. If you can not do this, turn down all the channel digital levels by 10 dB except for the LFE. No more clipping.

[hulkss]

btw - in latest versions I've found if you go to audio-path - info - it tells you how much "negative" gain is being applied to try and normalise to remove clipping often on mine this can be anywhere down to -9dB - not good at all

If you want more level, just apply Parametric EQ with +XdB on all channels.

[hulkss]

No, it is the other way round. When activating the "Source is correctly ..." option the LFE channel will end up -10db digital level from all other channels after decoding (eg. LAV) (if disabled it will end up -20db). Audiolense level matches all channels so that the same digital input level results in the same volume at listening position for all channels. Now your LFE channel needs a 10db boost to have the same digital level as the other channels and therefor results in the same volume after convolution.

It is lame if Audiolense works this way, especially not clearly pointing this out to users. Cinema audio content in the LFE channel is commonly recorded all the way to 0 dBFS which will produce a 115 dB SPL with reference level acoustic gain. You can not add any more digital level with out clipping low frequency peaks. We will need to push the LFE channel analog gain 10 dB and then attenuate any bass rerouted to the LFE channel by 10 dB.

I have started a thread at Audiolense to get answers to this issue. http://groups.google.com/group/audiolense/browse_thread/thread/5ce7cb90415c4e5e

[Matt]

I still wonder if I can assume the same digital level from all channels when activating the "Source is correctly mastered for +10dB calibration" in DSP Output Format. Waiting for a reply from Matt.

A properly mastered source assumes something later in the chain will make the subwoofer +10dB.

Said another way, it assumes 1 volt on the subwoofer line will be 10dB louder than 1 volt on any other line.

If the source is not properly mastered to account for this +10dB shift, it would assume 1 volt is the same volume on all lines.  It would be rare to uncheck 'Source is correctly mastered for +10dB calibration'.  You would only do it for rare, particular files.  Do not use this option for level matching -- Room Correction should be used for this.

When you uncheck 'Source is correctly mastered for +10dB calibration', it turns the subwoofer channel down -10dB to make the source correctly mastered.

I sort of wonder if we should just remove this option?

[Matt]

with Internal Volume and convolution there are basically 3 options to level match the speakers:

1) the necessary gain/attenuation is "baked into" the DRC filter and therefor levels are matched during convolution. (This is how it is done when using eg. Audiolense)

2) if the individual DRC filters are normalized to 0dbFS (therefor are not correcting level diffeences between speakers) the level must be set after convolution:

a) by using scaling in the convolution configuration file (just as you advised me to do for bass management.

eg. when I need -10db to match my subwoofer channel with the other channels my example from before would read:

C:\EQ\Referenz-3db\Subwoofer_Flat-3_48.wav
0
0.31622776601683794 1.31622776601683794 2.31622776601683794 4.31622776601683794 5.31622776601683794 6.31622776601683794 7.31622776601683794
3.31622776601683794

b) by using the level setting in Room Correction after convolution. (This is how I do it atm)


Which of these options is recommended for best gain structure/headroom/"sound quality" when used with the "new" internal volume? When you have the choice between 2) a) or b) which one is to be preferred?

Any approach will provide a perfect level match, so the only difference is how the headroom can best be utilized by Internal Volume.

It seems reasonable to me to feed a signal that uses the full range (0-1 / up to 0 dB) into Room Correction on all channels.  Then use Room Correction for level matching.

Some earlier steps in the chain that change volume (like JRSS) keep track of the headroom available in each channel.  Room Correction can then use this information to safely squeeze a little more volume out in certain cases.  If you do level matching in Convolution, it won't know about this headroom.

If you have enough volume, the discussion about headroom isn't important.  It only matters when you want to safely squeeze the most volume possible out of the system.

[hulkss]

A properly mastered source assumes something later in the chain will make the subwoofer +10dB. When you uncheck 'Source is correctly mastered for +10dB calibration', it turns the subwoofer channel down -10dB to make the source correctly mastered. I sort of wonder if we should just remove this option?

This is a valuable option to play some improperly mixed LFE content. Maybe say "uncheck to correct improperly mixed LFE recordings where LFE bass is too loud". I assume this only effects a change to LFE from the source, not rerouted bass.

[Matt]

Discussion about removing the option about correctly mastered sources here:
http://yabb.jriver.com/interact/index.php?topic=70303.0

[Trumpetguy]

Cinema audio content in the LFE channel is commonly recorded all the way to 0 dBFS which will produce a 115 dB SPL with reference level acoustic gain. You can not add any more digital level with out clipping as you have found out. Instead, you should boost the analog gain of your LFE channel by 10 dB. If you can not do this, turn down all the channel digital levels by 10 dB except for the LFE. No more clipping.

When using convolution and Audiolense filters that are level matched, doing a +10dB adjustment of the analog gain on my subs would be a pain (potmeters). Adjusting all other channels -10dB in the digital domain would then result in a digital attenuation of something like -20 to -25dB in total because of the attenuation created by the filters. Is this what it takes to get a correct sub level? What about dynamics?

If you want more level, just apply Parametric EQ with +XdB on all channels.

@hulkss: Isn't this contradictory to what you posted in the previous post?

I have started a thread at Audiolense to get answers to this issue. http://groups.google.com/group/audiolense/browse_thread/thread/5ce7cb90415c4e5e

Nice

[Buckster]

Cinema audio content in the LFE channel is commonly recorded all the way to 0 dBFS which will produce a 115 dB SPL with reference level acoustic gain. You can not add any more digital level with out clipping as you have found out. Instead, you should boost the analog gain of your LFE channel by 10 dB. If you can not do this, turn down all the channel digital levels by 10 dB except for the LFE. No more clipping.

thanks for that but even with LFE channel at 0 and all other channels at -10dB I'm getting a substantial amount of clipping

[Trumpetguy]

thanks for that but even with LFE channel at 0 and all other channels at -10dB I'm getting a substantial amount of clipping

This makes no sense to me. If clip protection reduces your gain by -9dB (your earlier post), a fixed gain reduction of -10dB should mean your are in the clear (given you are testing on the same source material). Unless some other process is increasing your digital level somewhere in the signal chain.

Do you have normalize volume activated? If so, wouldn't this try to apply a gain in low volume passages which can lead to clipping in transient passages (i.e. normalize volume and clip protection fighting against each other)?

[hulkss]

thanks for that but even with LFE channel at 0 and all other channels at -10dB I'm getting a substantial amount of clipping

Well.....I would guess you are clipping on the LFE channel. Crank up your subwoofer amplifier gain some and remeasure your system. The resulting filters won't have to push the LFE level so much.

[hulkss]

I'll start a new thread on this issue. http://yabb.jriver.com/interact/index.php?topic=70334.0

When using convolution and Audiolense filters that are level matched, doing a +10dB adjustment of the analog gain on my subs would be a pain (potmeters). Adjusting all other channels -10dB in the digital domain would then result in a digital attenuation of something like -20 to -25dB in total because of the attenuation created by the filters. Is this what it takes to get a correct sub level? What about dynamics?

You simply must crank up the LFE channel analog acoustic gain +10 dB.

[hulkss]

I said: If you want more level, just apply Parametric EQ with +XdB on all channels.

hulkss: Isn't this contradictory to what you posted in the previous post?

From the point that it could cause clipping of the signal yes, it's really not a good suggestion.

[Buckster]

clipping seems to be an issue all round

I just tried adding +10dB to the LFE - using suggested method of setting LFE channel in Room Correction to 0dB ...

And setting all other channels to -10dB

BUT in Audio-path its still saying its adjusting volume -4dB to account for clipping - therefore is limiting dynamic range etc

[Buckster]

for one Bluray I've tried this evening

played back with levels for channels set to all -10dB apart from Sub @ 0dB

Audio path stated it had needed to apply -3.5dB gain to stop clipping

so set channel levels lower - so all -14dB apart from Sub now @-4dB

Audio path still stating -3.5dB gain required !!

just can't seem to avoid clipping

[hulkss]

Are you developing convolver filters in Audiolense?
Are you allowing more than 6dB of gain in the filters?
Try using the default settings in Audiolense if you haven't.
Were your audio files ever "normalized" to 100%?
Are you applying positive gain anywhere in the digital processing?

[Buckster]

hulkss - I've re-read thread and actually my comments may be more relevant in the other thread as I'mnot suing convolver (yet - as I can't get rid of clipping to start with)

I'm decoding from original files (DTS MA etc) - and I'm not applying any positive gain anywhere in the digital processing

[Mikkel]

The solution by editing the convolution-file seems logical. But I struggle to understand the logic behind the Room Correction-solution:
- How does lowering all channels except the LFE/sub by -10db prevent low-passed signals sent to the sub from being 10db too loud?
If I set the LFE to 0db in Room Correction and measure 75dbC then all other channels should measure 65 dbC when calibrated. But since all other channels pass on the bass to the LFE-channel the low-pass signal will be 10db to high while the LFE-signal will be correct.

Am I way off?

EDIT:
Is it a solution if JRiver automatically cuts 10db from the low-passed signal? Then we can calibrate all channels to the correct level provided the convolution files are normalized to the same level.