TOPIC:

^
You still don't seem to understand what the potential issues are. Without first understanding that you won't glean anything useful from measurements.

markus wrote:

^
You still don't seem to understand what the potential issues are. Without first understanding that you won't glean anything useful from measurements.

I might not completely understand the potential issue but I do understand your point of using two speakers (speaker + sub woofer) where only one speaker should have been used for the measurement when the signal is sent out for measurement. What I have noticed is that this theoretical discussion can go on and on. But if there is a way to confirm this Potential issue, then i'd help everybody.

Today, I downloaded Dirac free version, installed it and calibrated the system.
Markus: WIth PC version, I couldn't take REW reading. I tried all outputs (OASIS/HDMI OUT) but either showed me any differnece in reading with Dirac On/Off. So that was a no go.

Before I did any calibration, I turned Audyssey off in my avr and set all level trims to 0.

Difference b/w Audyssey Multi-XT vs Dirac After Calibration: Multi-XT had louder (not necessarily better) bass. Dirac had less mostly because bass was set low by Dirac. I tried to increase the filters for Bass, Optimized and saved filter. Loaded the filter in Dirac Controller but it didn't make any difference. Not sure how to increase the bass in Dirac. When I turned bass trim in avr, it did increase bass but I don't think that's the right way to go.

In terms of Mid-Bass, I think Dirac has better result. In terms of over-all sound quality, Dirac was much cleaner as compared to Audyssey XT.

I have spent several days testing various kinds of BM such as in the sound card, media player, AVR etc. before and after DL. I have concluded that BM after DL where DL measures the combined response of speaker + subwoofer is the best.

Player > Dirac Live > Bass Management > Amplifier

The reason for my conclusion is that most BM implementations are black boxes that severely impact the audio at the crossover point (unless you carefully crafted the crossover with a nanoAVR or a custom made filter). In some cases the level of the crossed over signals is also incorrect! When you increase the Subwoofer level in an AVR, you can be sure they're just doing it at the amp stage including the crossed over signals and not just the LFE channel! With BM before correction, the subwoofer's frequency response will be corrected, but the level of the crossed over signals will be incorrect!

However with BM after DL, the test tones will be played through the crossover and the combined response of the crossover is measured and corrected. The level of the crossed over signals and the LFE channel will also be corrected and the final subwoofer output will be correct. Dirac will cut or boost the lower frequencies in each channel so that the crossed over level at the subwoofer is correct and it will cut or boost the higher frequencies so that the speaker levels are correct. The result is a perfectly corrected target without any damage due to the crossover.

There is a concern about DL treating the speaker + subwoofer as one. Unlike a multi-way speaker where all drivers are in the same cabinet, with BM after DL the bass will play from across the room! However I don't think this matters in practice. Dirac still reports delays accurate up to the millisecond and sets the channel gain appropriately. Here the channel gain seems to correspond to the speakers playing the higher frequencies, but the crossed over lower frequencies' gain is further corrected within the filters so the net levels are still correct.

The disadvantage of this is you cannot touch the crossover after calibration. It seems to me that most Audyssey implementations in AVRs perform BM before Audyssey. The reason they can do this is the crossover implementation is not a black box to the manufacturer. They provide a fixed set of crossovers (40, 60, 80 Hz etc.) and they know the net response of each. Thus Audyssey can correct just the speakers, and they can apply the crossover before Audyssey and you can vary it after calibration. Even then I doubt they've done it perfectly.

The +10 dB LFE is a codec specific thing and should only be applied for Dolby sources. When bit streaming to an AVR that uses Audyssey, the AVR can -10 dB just the crossed over signals from the other channels and then boost the whole subwoofer by +10 dB if it detects a Dolby source. When using the source player as the decoder in the case of DL the behaviour is undefined. The player will most likely have to -10 dB all the other channels and then crossover.

Ok so let's say we have the following setup
source --> nonoAVR --> AVR
Before calibration, we reset all speakers level in AVR. This sets bass trim to 0 as well. Now we run calibration and afterwords find out that I need more bass. What are my options now? Should I change LFE curve and give it a boost? Or I can just change bass trim in AVR?

Currently I'm using PC version and waiting for nanoAVR to arrive. In PC version, changing LFE curve didn't make any difference. Changing bass trim did but I'm not sure its the right way to go?

Also since LFE curve didn't change any sound effect for me, what's its purpose then?

sheraz wrote:

Ok so let's say we have the following setup
source --> nonoAVR --> AVR
Before calibration, we reset all speakers level in AVR. This sets bass trim to 0 as well. Now we run calibration and afterwords find out that I need more bass. What are my options now? Should I change LFE curve and give it a boost? Or I can just change bass trim in AVR?

Currently I'm using PC version and waiting for nanoAVR to arrive. In PC version, changing LFE curve didn't make any difference. Changing bass trim did but I'm not sure its the right way to go?

Also since LFE curve didn't change any sound effect for me, what's its purpose then?

This is normal. Since BM is after Dirac, the Dirac target affects the only source channel, not the speaker. You only increased the LFE track. You have to edit all the channel targets. However I recommend that you edit the Dirac target only for the purpose of "room correction" and not for personal taste. Boosting the Dirac target beyond the Auto target will almost always result in clipping (even with the -8.0 dB DSP gain!). For personal preferences such as bass boost it's better to just increase the subwoofer volume (either on the subwoofer itself or on the AVR) so that it uniformly increases the bass of all the channels.

As far as the AVR is concerned, I recommend running the AVR's calibration first. Then switch off only the EQ for the Dirac input source but leave the speaker levels and crossovers as they are. You must not use Direct mode or the crossover won't work. This will give Dirac far more headroom to make corrections. The calibration must be done before you increase the subwoofer level further.

Boosting the Dirac target beyond the Auto target will almost always result in clipping (even with the -8.0 dB DSP gain!).

Well then if I wanted to boost Mid-Bass, how would I do that. I was thinking of raising 3-4 from 300 to cross over point.

When I calibrated avr, it sets subtrim to - 4.5. I was resetting all speakers including subs to zero before running Dirac. According to you, If I leave speakers levels to what Audyssey set, I'm starting subs with -4.5. Before I was starting with 0 and now with -4.5. Wouldn't that make bass weaker than before?

You must not use Direct mode or the crossover won't work

What do you mean by Direct mode?

As far as the AVR is concerned, I recommend running the AVR's calibration first. Then switch off only the EQ for the Dirac input source but leave the speaker levels and crossovers as they are. You must not use Direct mode or the crossover won't work. This will give Dirac far more headroom to make corrections. The calibration must be done before you increase the subwoofer level further.

sheraz wrote:

Well then if I wanted to boost Mid-Bass, how would I do that. I was thinking of raising 3-4 from 300 to cross over point.

I suggest you learn to like the Auto target. It is somewhat close to an equal loudness curve for the bass at 85 dB(SPL). If you must boost a frequency, you can achieve that by lowering everything else except the desired frequency. It's thus boosted in effect. You'll have to turn up the amplifier volume to compensate.

sheraz wrote:

When I calibrated avr, it sets subtrim to - 4.5. I was resetting all speakers including subs to zero before running Dirac. According to you, If I leave speakers levels to what Audyssey set, I'm starting subs with -4.5. Before I was starting with 0 and now with -4.5. Wouldn't that make bass weaker than before?

If you set it to 0 dB, Dirac will apply a -4.5 dB gain anyway when you export the filter. You can see the gain values if you expand the filter in Dirac Audio Processor. By allowing the AVR to set the initial levels, the Dirac gain correction will be smaller. This makes it easier to compare Filter On and Off in DAP. Otherwise when you switch it off the volume will increase by 4.5 dB.

sheraz wrote:

What do you mean by Direct mode?

It's one of the modes like Music, Movie, Game, Auto. Direct mode disables all DSP including Audyssey and crossover. You need the crossover. Thus you have to use another mode such as Auto and switch off only Audyssey but leave the crossover enabled.

Thanks for all the inputs. I do like the part of reducing the values instead of increasing. Makes sense as long as I don't run out of volumne

If you set it to 0 dB, Dirac will apply a -4.5 dB gain anyway when you export the filter. You can see the gain values if you expand the filter in Dirac Audio Processor. By allowing the AVR to set the initial levels, the Dirac gain correction will be smaller. This makes it easier to compare Filter On and Off in DAP. Otherwise when you switch it off the volume will increase by 4.5 dB.

My question was more towards how to run subs hotter. Your point is more towards what to do before running calibration. Say I run Dirac calibration based on the levels Audyssey sets. Then after the calibration, I feel like I need more bass. In that case what should be done?
Another way to ask the same question is that generally in avr with Audyssey, I turn sub trim higher after calibration if I want more bass. What is equivalent to turning sub in Dirac to do the same?

kathampy wrote:

I have spent several days testing various kinds of BM such as in the sound card, media player, AVR etc. before and after DL. I have concluded that BM after DL where DL measures the combined response of speaker + subwoofer is the best.

Player > Dirac Live > Bass Management > Amplifier

The reason for my conclusion is that most BM implementations are black boxes that severely impact the audio at the crossover point (unless you carefully crafted the crossover with a nanoAVR or a custom made filter). In some cases the level of the crossed over signals is also incorrect! When you increase the Subwoofer level in an AVR, you can be sure they're just doing it at the amp stage including the crossed over signals and not just the LFE channel! With BM before correction, the subwoofer's frequency response will be corrected, but the level of the crossed over signals will be incorrect!

However with BM after DL, the test tones will be played through the crossover and the combined response of the crossover is measured and corrected. The level of the crossed over signals and the LFE channel will also be corrected and the final subwoofer output will be correct. Dirac will cut or boost the lower frequencies in each channel so that the crossed over level at the subwoofer is correct and it will cut or boost the higher frequencies so that the speaker levels are correct. The result is a perfectly corrected target without any damage due to the crossover.

There is a concern about DL treating the speaker + subwoofer as one. Unlike a multi-way speaker where all drivers are in the same cabinet, with BM after DL the bass will play from across the room! However I don't think this matters in practice. Dirac still reports delays accurate up to the millisecond and sets the channel gain appropriately. Here the channel gain seems to correspond to the speakers playing the higher frequencies, but the crossed over lower frequencies' gain is further corrected within the filters so the net levels are still correct.

The disadvantage of this is you cannot touch the crossover after calibration. It seems to me that most Audyssey implementations in AVRs perform BM before Audyssey. The reason they can do this is the crossover implementation is not a black box to the manufacturer. They provide a fixed set of crossovers (40, 60, 80 Hz etc.) and they know the net response of each. Thus Audyssey can correct just the speakers, and they can apply the crossover before Audyssey and you can vary it after calibration. Even then I doubt they've done it perfectly.

The +10 dB LFE is a codec specific thing and should only be applied for Dolby sources. When bit streaming to an AVR that uses Audyssey, the AVR can -10 dB just the crossed over signals from the other channels and then boost the whole subwoofer by +10 dB if it detects a Dolby source. When using the source player as the decoder in the case of DL the behaviour is undefined. The player will most likely have to -10 dB all the other channels and then crossover.

@ kathampy,

Thanks so much for the update and we're glad to hear a bit more feedback on testing both setup. Maybe some graphs would be helping a bit more if you ever get some spare time..

This whole Before vs After Bass Management talk is really interesting and we've been a bit too busy in the past few days (it's almost Chinese new Year) to spend more time to provide some information but our general comments were indeed similar. It works both way, it just depend what you're trying to do and how much you're willing to tune.

DevTeam

sheraz wrote:

Thanks for all the inputs. I do like the part of reducing the values instead of increasing. Makes sense as long as I don't run out of volumne

On a PC if you have an analog sound card you can adjust the bass there without clipping. On my Sound Blaster Zx I am able to add a large bass boost in the equaliser without clipping. On an AVR you can disable Audyssey and set it to Graphic Equaliser mode and boost the bass there. This is preferable to modifying the Dirac target since it has limited headroom for boosting (maximum of +8dB for both correction & bass boost).

My question was more towards how to run subs hotter. Your point is more towards what to do before running calibration. Say I run Dirac calibration based on the levels Audyssey sets. Then after the calibration, I feel like I need more bass. In that case what should be done?
Another way to ask the same question is that generally in avr with Audyssey, I turn sub trim higher after calibration if I want more bass. What is equivalent to turning sub in Dirac to do the same?

Increasing the sub level in the AVR is fine. However it will only boost the frequencies below the crossover point for music. This does not sound as nice as a bass boost slope from say +4 dB at 30 Hz to +1 dB at 500 Hz.

devteam wrote:

kathampy wrote:

I have spent several days testing various kinds of BM such as in the sound card, media player, AVR etc. before and after DL. I have concluded that BM after DL where DL measures the combined response of speaker + subwoofer is the best.

Player > Dirac Live > Bass Management > Amplifier

The reason for my conclusion is that most BM implementations are black boxes that severely impact the audio at the crossover point (unless you carefully crafted the crossover with a nanoAVR or a custom made filter). In some cases the level of the crossed over signals is also incorrect! When you increase the Subwoofer level in an AVR, you can be sure they're just doing it at the amp stage including the crossed over signals and not just the LFE channel! With BM before correction, the subwoofer's frequency response will be corrected, but the level of the crossed over signals will be incorrect!

However with BM after DL, the test tones will be played through the crossover and the combined response of the crossover is measured and corrected. The level of the crossed over signals and the LFE channel will also be corrected and the final subwoofer output will be correct. Dirac will cut or boost the lower frequencies in each channel so that the crossed over level at the subwoofer is correct and it will cut or boost the higher frequencies so that the speaker levels are correct. The result is a perfectly corrected target without any damage due to the crossover.

There is a concern about DL treating the speaker + subwoofer as one. Unlike a multi-way speaker where all drivers are in the same cabinet, with BM after DL the bass will play from across the room! However I don't think this matters in practice. Dirac still reports delays accurate up to the millisecond and sets the channel gain appropriately. Here the channel gain seems to correspond to the speakers playing the higher frequencies, but the crossed over lower frequencies' gain is further corrected within the filters so the net levels are still correct.

The disadvantage of this is you cannot touch the crossover after calibration. It seems to me that most Audyssey implementations in AVRs perform BM before Audyssey. The reason they can do this is the crossover implementation is not a black box to the manufacturer. They provide a fixed set of crossovers (40, 60, 80 Hz etc.) and they know the net response of each. Thus Audyssey can correct just the speakers, and they can apply the crossover before Audyssey and you can vary it after calibration. Even then I doubt they've done it perfectly.

The +10 dB LFE is a codec specific thing and should only be applied for Dolby sources. When bit streaming to an AVR that uses Audyssey, the AVR can -10 dB just the crossed over signals from the other channels and then boost the whole subwoofer by +10 dB if it detects a Dolby source. When using the source player as the decoder in the case of DL the behaviour is undefined. The player will most likely have to -10 dB all the other channels and then crossover.

@ kathampy,

Thanks so much for the update and we're glad to hear a bit more feedback on testing both setup. Maybe some graphs would be helping a bit more if you ever get some spare time..

This whole Before vs After Bass Management talk is really interesting and we've been a bit too busy in the past few days (it's almost Chinese new Year) to spend more time to provide some information but our general comments were indeed similar. It works both way, it just depend what you're trying to do and how much you're willing to tune.

DevTeam

Not sure what "It works both way" really means within this context. There will be sound coming from the speakers in both cases, that's for sure but there is only one way to do it technically right. That would be "Decoder > Bass Management > EQ/Delay/Gain".

markus wrote:

Not sure what "It works both way" really means within this context. There will be sound coming from the speakers in both cases, that's for sure but there is only one way to do it technically right. That would be "Decoder > Bass Management > EQ/Delay/Gain".

If you do this, the volume of the crossed over frequencies played through the subwoofer will be wrong. There is no easy way to fix this. The crossed over frequencies have already been mixed into the subwoofer input and thus when Dirac corrects what it thinks is just the LFE channel, it is affecting the volume of the crossed over frequencies incorrectly.

So when you play a sweep through the Front channels, the crossed over frequencies will be too loud or too soft, but the frequencies played from the front speakers will be correct. To fix this after calibration, you'll need to manually measure the output of just the crossed over frequencies from the subwoofer and then tweak the internals of the crossover to attenuate or boost the signals being mixed into the subwoofer. You also have the issue of incorrect volume at the crossover point itself, since it is already mixed before going to Dirac.

kathampy wrote:

markus wrote:

Not sure what "It works both way" really means within this context. There will be sound coming from the speakers in both cases, that's for sure but there is only one way to do it technically right. That would be "Decoder > Bass Management > EQ/Delay/Gain".

If you do this, the volume of the crossed over frequencies played through the subwoofer will be wrong. There is no easy way to fix this. The crossed over frequencies have already been mixed into the subwoofer input and thus when Dirac correct what it thinks is just the LFE channel, it is affecting the volume of the crossed over frequencies incorrectly.

So when you play a sweep through the Front channels, the crossed over frequencies will be too loud or too soft, but the frequencies played from the front speakers will be correct. To fix this after calibration, you'll need to manually measure the output of just the crossed over frequencies from the subwoofer and then tweak the internals of the crossover to attenuate or boost the signals being mixed into the subwoofer.

No, volume would not be wrong. Dirac Live corrects each speaker individually and adjusts the level so each speaker will sound as loud as the other one. This is also true for the subwoofer channel.

This is how bass management works:



Please have a look at www.avsforum.com/forum/90-receivers-amps...-dirac-live-box.html
In that thread you'll find measurements showing that everything is correctly handled when you're using a custom speaker layout. If you're using the predefined layouts a 10dB boost is applied to the subwoofer channel. The latter can lead to levels being off.

markus wrote:

No, volume would not be wrong. Dirac Live corrects each speaker individually and adjusts the level so each speaker will sound as loud as the other one. This is also true for the subwoofer channel.

Dirac will correct the subwoofer, but only with respect to the pre-mixed LFE channel being fed to it. The crossed over component and LFE component cannot be varied independently as it should be. It will make the total output of the subwoofer sound as loud as the other channels, when actually only the LFE component should be as loud as the rest. The total output should actually be louder since the crossed over frequencies are also added.