NOTE: this is a part 2 post. Read
http://www.whatsbestforum.com/showt...Bass-Array-to-Sound-Field-Management-Overview
first.
OK, now the meaty part: I want to compare the Double Bass Array to Sound Field Management on a point by point basis. Efficiency seems like a good place to start.
Efficiency
It might seem that the DBA approach would take an efficiency hit, since we are creating just a front to back plane wave (like in a free field), thus removing the room gain.
The intended effect of the DBA is to remove the room and create a free field plane wave response. We are removing the room modes. It is clear that the room modes increase the average sound pressure level in the room, so removing them must reduce the average SPL in the room. Of course it may not be easy to see this, since at some locations/frequencies, when we remove the modes we are removing cancellation dips.
Of course, subwoofer optimization can be said to rely on cancellation of room modes in one way or the other. For Sound Field Management, however, we are only trying to optimize the sound at a select number of seats, not the entire room. Remember that DBA effectively optimizes the entire room, there is no measurement at any particular seat. For SFM, it is entirely possible that efficiency can be increased at the selected seats even while the average SPL in the room is decreased. We are simply “moving the sound around”. With the “purest” version of SFM we ignore bass efficiency, and only try to optimize seat to seat consistency. However, if we change the selection criteria for SFM to choose among the best solutions, we can sacrifice a small amount of seat to seat consistency to get more bass efficiency and usually end up at least not losing any bass, often gaining. It’s a matter of changing one line of code.
Interestingly, at frequencies just below the first mode, where the room acts as a lumped element system, I will contend that in the case of DBA the delayed and inverted rear speakers will be in phase with the front. So, at these very low frequencies, DBA won’t lose any bass. I will leave this as an exercise for the reader!
Optimizing particular seats versus the entire room.
DBA in effect attempts to optimize the entire room, due to the plane wave produced. This would be a particular advantage if/when you don’t know where the listeners will be. So, if there are no fixed seats or designated seating area DBA would have an advantage.
Convenience
In the author’s opinion, DBA has a major disadvantage in this category. Implementation requires 2 or 4 subwoofers at very particular positions, duplicated on both front and rear walls, not just sitting on the floor either but in or on the walls. Very small rooms could use 2 subs each on front and rear wall, but otherwise 4 subs on front and rear walls, for a total of 8 subs, is required.
http://www.avsforum.com/avs-vb/showthread.php?t=837744
DBA requires 4 or 8 identical subwoofers. The SFM algorithm has no particular requirements that the subs be identical, though they should be within a few dB of each other in max output capability.
SFM normally uses subwoofers on the floor, measured at multiple locations, but locations which are preselected to meet WAF considerations. Normally 2 or 4 subwoofers are all that is required.
As for configuring the system, the SFM method is at a disadvantage, requiring multiple impulse response measurements (no 1/3 octave RTA’s here). Each possible combination of potential subwoofer location and seat must be measured. For DBA the delay and gain setting can be calculated (though measured values might give a slightly better result).
As for signal processing required, SFM does require a bit more. Level, delay and one biquad filter is required for each subwoofer channel. In DBA one global delay and gain would suffice.
Applicability to non rectangular rooms
SFM will work in any room, since it is based on actual room measurements and does not care about the shape of the room. DBA is based on the acoustics of a rectangular room and wont work in a non rectangular room. It also may not work as advertised in rooms which are acoustically non-rectangular. This could be the case for rooms with very different wall construction on different walls, for example.
Performance
Though no systematic comparison has been made between DBA and SFM, I will divulge my unbiased opinion on the matter and proclaim that SFM will give better performance. The reason I can say this is that if I allow longer delay values and polarity inversion in the SFM algorithm, the DBA solution will pop out if it indeed is the best solution. However, if there is a better solution, that alternate solution will pop out, and be an improvement over straight-up DBA. The author has done this and even in “rectangular” rooms (which are often not acoustically rectangular), DBA did not always give the best solution. DBA can thus be considered as a special case of the more general SFM approach. Still, to be fair, DBA does give a solution which is likely to be a very good one.
SUMMARY
I would summarize by saying that in rectangular rooms, if you don’t know where the seats will be, and you can get the subs located in/on the front and back walls, use DBA. Also if making impulse response measurements is too difficult, use DBA. In all other cases SFM is a better approach. For non-rectangular rooms, DBA is not an option.
Afterthoughts
Unfortunately, neither method discussed here is particularly easy to implement at present. For DBA, it’s because of the requirement for 4 or more likely 8 subs to placed in exact locations in/on the front and rear walls. For SFM it’s because the search algorithm is not widely commercially available yet. In the case of SFM, it is incorporated into the Harman Audio Test System, and available as part of a JBL Synthesis custom installation. An optimization algorithm which does what SFM does (albeit using a different method) is commercially available under the JBL Performance line. This is the BassQ 4-channel Automatic Room Mode Corrections Processor. Some other commercial algorithms have been around for a while (R.A.B.O.S for example), and more are coming on line all the time (Audyssey, Trinnov, Lyngdorf, etc., etc., etc., etc…..). Most of these do not address seat to seat consistency and/or do not use separate processing for different subs (or subwoofer groups).
OK, if you’re still with me, and don’t just accept everything I just said in this missive, good for you! Fire away….
http://www.whatsbestforum.com/showt...Bass-Array-to-Sound-Field-Management-Overview
first.
OK, now the meaty part: I want to compare the Double Bass Array to Sound Field Management on a point by point basis. Efficiency seems like a good place to start.
Efficiency
It might seem that the DBA approach would take an efficiency hit, since we are creating just a front to back plane wave (like in a free field), thus removing the room gain.
The intended effect of the DBA is to remove the room and create a free field plane wave response. We are removing the room modes. It is clear that the room modes increase the average sound pressure level in the room, so removing them must reduce the average SPL in the room. Of course it may not be easy to see this, since at some locations/frequencies, when we remove the modes we are removing cancellation dips.
Of course, subwoofer optimization can be said to rely on cancellation of room modes in one way or the other. For Sound Field Management, however, we are only trying to optimize the sound at a select number of seats, not the entire room. Remember that DBA effectively optimizes the entire room, there is no measurement at any particular seat. For SFM, it is entirely possible that efficiency can be increased at the selected seats even while the average SPL in the room is decreased. We are simply “moving the sound around”. With the “purest” version of SFM we ignore bass efficiency, and only try to optimize seat to seat consistency. However, if we change the selection criteria for SFM to choose among the best solutions, we can sacrifice a small amount of seat to seat consistency to get more bass efficiency and usually end up at least not losing any bass, often gaining. It’s a matter of changing one line of code.
Interestingly, at frequencies just below the first mode, where the room acts as a lumped element system, I will contend that in the case of DBA the delayed and inverted rear speakers will be in phase with the front. So, at these very low frequencies, DBA won’t lose any bass. I will leave this as an exercise for the reader!
Optimizing particular seats versus the entire room.
DBA in effect attempts to optimize the entire room, due to the plane wave produced. This would be a particular advantage if/when you don’t know where the listeners will be. So, if there are no fixed seats or designated seating area DBA would have an advantage.
Convenience
In the author’s opinion, DBA has a major disadvantage in this category. Implementation requires 2 or 4 subwoofers at very particular positions, duplicated on both front and rear walls, not just sitting on the floor either but in or on the walls. Very small rooms could use 2 subs each on front and rear wall, but otherwise 4 subs on front and rear walls, for a total of 8 subs, is required.
http://www.avsforum.com/avs-vb/showthread.php?t=837744
DBA requires 4 or 8 identical subwoofers. The SFM algorithm has no particular requirements that the subs be identical, though they should be within a few dB of each other in max output capability.
SFM normally uses subwoofers on the floor, measured at multiple locations, but locations which are preselected to meet WAF considerations. Normally 2 or 4 subwoofers are all that is required.
As for configuring the system, the SFM method is at a disadvantage, requiring multiple impulse response measurements (no 1/3 octave RTA’s here). Each possible combination of potential subwoofer location and seat must be measured. For DBA the delay and gain setting can be calculated (though measured values might give a slightly better result).
As for signal processing required, SFM does require a bit more. Level, delay and one biquad filter is required for each subwoofer channel. In DBA one global delay and gain would suffice.
Applicability to non rectangular rooms
SFM will work in any room, since it is based on actual room measurements and does not care about the shape of the room. DBA is based on the acoustics of a rectangular room and wont work in a non rectangular room. It also may not work as advertised in rooms which are acoustically non-rectangular. This could be the case for rooms with very different wall construction on different walls, for example.
Performance
Though no systematic comparison has been made between DBA and SFM, I will divulge my unbiased opinion on the matter and proclaim that SFM will give better performance. The reason I can say this is that if I allow longer delay values and polarity inversion in the SFM algorithm, the DBA solution will pop out if it indeed is the best solution. However, if there is a better solution, that alternate solution will pop out, and be an improvement over straight-up DBA. The author has done this and even in “rectangular” rooms (which are often not acoustically rectangular), DBA did not always give the best solution. DBA can thus be considered as a special case of the more general SFM approach. Still, to be fair, DBA does give a solution which is likely to be a very good one.
SUMMARY
I would summarize by saying that in rectangular rooms, if you don’t know where the seats will be, and you can get the subs located in/on the front and back walls, use DBA. Also if making impulse response measurements is too difficult, use DBA. In all other cases SFM is a better approach. For non-rectangular rooms, DBA is not an option.
Afterthoughts
Unfortunately, neither method discussed here is particularly easy to implement at present. For DBA, it’s because of the requirement for 4 or more likely 8 subs to placed in exact locations in/on the front and rear walls. For SFM it’s because the search algorithm is not widely commercially available yet. In the case of SFM, it is incorporated into the Harman Audio Test System, and available as part of a JBL Synthesis custom installation. An optimization algorithm which does what SFM does (albeit using a different method) is commercially available under the JBL Performance line. This is the BassQ 4-channel Automatic Room Mode Corrections Processor. Some other commercial algorithms have been around for a while (R.A.B.O.S for example), and more are coming on line all the time (Audyssey, Trinnov, Lyngdorf, etc., etc., etc., etc…..). Most of these do not address seat to seat consistency and/or do not use separate processing for different subs (or subwoofer groups).
OK, if you’re still with me, and don’t just accept everything I just said in this missive, good for you! Fire away….
Last edited:
