Jump to content
  • The Computer Audiophile

    Understanding the State of the Art of Digital Room Correction

     

     

    Editor's Note: Mitch Barnett of Accurate Sound just published his video masterclass in digital room correction and I'm thrilled to help get the word out to audiophiles. I've used room correction in my main system for two years and I absolutely love the results in my main system and headphone system. This technology has come incredibly far over the years. I believe everyone should at least give it a shot and keep it as a tool in their audiophile toolbox. 

     

    Here's a small intro from @mitchco, followed by his nearly two hour, incredibly in-depth, video. 

     

     

    From Mitch:

     

    A deep dive presentation on the fundamentals of "proper" Digital Room Correction (DRC). Includes hands-on DSP FIR Filter Designer demos using Acourate and Audiolense.

     

    Having participated in many audio forum discussions, having watched online videos on Digital Room Correction (or DRC), and having reviewed over a dozen DRC products over the past 11 years, I have come to two conclusions. One is that there is considerable misunderstanding about DRC, how it works and even what problems DRC is trying to solve. And, just as important, understanding what is possible using the SOTA of DRC. I hope you find the content educational and practical. 

     

    This presentation assumes some basic understanding of loudspeaker measurements, room acoustics, and psychoacoustics, the latter being the science of how we perceive sound.

     

     

     

     

     

     




    User Feedback

    Recommended Comments



    Nice work @mitchco. Thanks so much for taking the time to create this and educate people. While the video is nearly two hours, I'm willing to bet you have much more time into creating this, and you're giving it away for free. Nice work!

    Share this comment


    Link to comment
    Share on other sites

    40 minutes ago, pga said:

    Below you can see the before and after results using Audiolense convolution filters designed by Mitch for TAD R1s in my room.  
     

    One chart is for stock TAD R1s with their passive crossover.  The other chart is for biamped TADs using a Pass XVR1 active crossover.  Big thanks to Andrew Jones, who designed the R1s, for his invaluable help setting up the XVR1 so that it would sum perfectly with the crossover that serves as the high pass for the coax in the TADs. 
     

    The biamped TADs were positioned close to the front wall and corners.  This placement resulted in significant bass boost that without EQ would have been very objectionable.  However, the  bass bump was easily dialed down with EQ, and this brought an added benefit of significantly increased headroom in the low end, in my case deep bass was attenuated 10-20db, which then became headroom.  Also, the bass dips with the TADs in the corners were much narrower, which makes the dips much less audible.  
     

    You can see both the stock and the biamped TADs benefited significantly from the EQ.  I tried using professional studio parametric EQs to neutralize room effects, but unless you have a very large, well behaved room, analogue EQ is just not powerful enough. The REW designed convolution filters are OK, but they fall fall short of what Mitch can do using Audiolense. 

    One surprise I had was that although the system was optimized using measurements taken at a single position exactly between the speakers at the listening position, the speakers sound great everywhere in the room.  The EQ delivered the much needed improvement in timbre. But what surprised me was how much better soundstage, depth and imaging was with EQ, making it very much irrelevant that the speakers were close to the front wall and corners. However this may also be a result of the significant RPG treatments on the front and side

     

    I only listen to music with Roon, so we have a set of Audiolense filters running on a i7 Roon NUC just for music. These filters sound great but they do have some latency (filter is “listening” to the music before you are, so it can start correcting the response before the actual note is heard, pretty cool). 
     

    I also use this system to watch TV.  Mitch designed filters that don’t have latency specifically for the TV.  Latency will cause lip sync issues with video.  These low latency filters run on a small Windows PC that has a miniDSP streamer with Toslink inputs attached.  These filters are not as capable as the ones running on Roon, but they still are head and shoulders better than what could be achieved with an AV receiver or processor.

     

     

    TAD R1 passive xover.jpg

    Paul FS4 Harman target.jpg

    Paul corrected step response.jpg


    So cool @pga

    Share this comment


    Link to comment
    Share on other sites

    50 minutes ago, pga said:

    Below you can see the before and after results using Audiolense convolution filters designed by Mitch for TAD R1s in my room.  
     

    One chart is for stock TAD R1s with their passive crossover.  The other chart is for biamped TADs using a Pass XVR1 active crossover.  Big thanks to Andrew Jones, who designed the R1s, for his invaluable help setting up the XVR1 so that it would sum perfectly with the crossover that serves as the high pass for the coax in the TADs. 
     

    The biamped TADs were positioned close to the front wall and corners.  This placement resulted in significant bass boost that without EQ would have been very objectionable.  However, the  bass bump was easily dialed down with EQ, and this brought an added benefit of significantly increased headroom in the low end, in my case deep bass was attenuated 10-20db, which then became headroom.  Also, the bass dips with the TADs in the corners were much narrower, which makes the dips much less audible.  
     

    You can see both the stock and the biamped TADs benefited significantly from the EQ.  I tried using professional studio parametric EQs to neutralize room effects, but unless you have a very large, well behaved room, analogue EQ is just not powerful enough. The REW designed convolution filters are OK, but they fall fall short of what Mitch can do using Audiolense. 

    One surprise I had was that although the system was optimized using measurements taken at a single position exactly between the speakers at the listening position, the speakers sound great everywhere in the room.  The EQ delivered the much needed improvement in timbre. But what surprised me was how much better soundstage, depth and imaging was with EQ, making it very much irrelevant that the speakers were close to the front wall and corners. However this may also be a result of the significant RPG treatments on the front and side

     

    I only listen to music with Roon, so we have a set of Audiolense filters running on a i7 Roon NUC just for music. These filters sound great but they do have some latency (filter is “listening” to the music before you are, so it can start correcting the response before the actual note is heard, pretty cool). 
     

    I also use this system to watch TV.  Mitch designed filters that don’t have latency specifically for the TV.  Latency will cause lip sync issues with video.  These low latency filters run on a small Windows PC that has a miniDSP streamer with Toslink inputs attached.  These filters are not as capable as the ones running on Roon, but they still are head and shoulders better than what could be achieved with an AV receiver or processor.

     

     

    TAD R1 passive xover.jpg

    Paul FS4 Harman target.jpg

    Paul corrected step response.jpg

    which RPG treatments did you use ?

    Share this comment


    Link to comment
    Share on other sites

    4 minutes ago, pga said:

    It is really a shame that high end audio dealers have not focused more on room EQ as an invaluable service.   While there are important differences in electronics, thankfully with all the progress made in the last 20 years understanding digital and amplifier design, the differences in electronics are subtle.  By far the two biggest variables impacting sound are your room and your speakers.  And most likely the third biggest is the quality of the recordings.  This is not to say that all DAC and all amplifiers sound the same.  Far from that. But you first have to get the room and speakers working together correctly, long before you start messing with the electronics or cables.  
     

    It would be unthinkable for an AV dealer to sell someone a high end projector and screen without calibration.  But almost no one in high end audio thinks about calibrating speakers.  You should think that the correct analogy is that your speakers are the equivalent of the projector, and the room is the movie screen.  But add to that that your room unfortunately is not reflective of all frequencies in the same way.  This would be as if the projector was shining light on a screen that varies is color or reflectivity.  Imagine that instead of a very nice perfectly flat Stewart screen you were projecting your 50,000 laser projector on your wall with all the artwork and shelves on it.  How would that picture look?  This is sort of what is happening when your speakers project sound into the room.  The speakers and the room become a system and this system hugely benefits from DSP to neutralize the effects of the room reflections on the sound.

     

    +100

     

    Given that room correction can be enabled and disabled with a couple taps of a finger in Roon, it only makes sense for consumers to have this in their toolboxes. I started out by using it a little bit because I wasn't used to it. Now, I only disable it a little bit because it's so good. 

     

    I wish every dealer either offered this service, or worked with Mitch to do it remotely. Compared to the price of our hardware and price of what we've spent on music over the years, calibrating an audio system costs peanuts. 

    Share this comment


    Link to comment
    Share on other sites

    4 minutes ago, Ozan Bolat said:

    which RPG treatments did you use ?

    They designed a package for the room, and this was about 20 years ago.  The front wall corners have Modex bass plate bass traps.  Also absorbers on the front wall.  The ceiling and one side was has absorbers.  The back wall has diffusers and abfusers.  

    Share this comment


    Link to comment
    Share on other sites

    2 minutes ago, pga said:

    Here are some photos showing the room 

    AFE8483F-130F-4153-97C6-4B325CCB8A32.jpeg

    7B192D53-8DA8-40E0-8B2C-BB590304A354.jpeg

    C93488C6-F1BE-4A2C-B09A-72E5AB8AE7C5.jpeg

    8AF96B90-BE1E-4D64-83FD-0C480448B8FA.jpeg

    7B3A61BB-5B51-48D9-AD4A-D76FCBA5372B.jpeg

     

    Great looking room. It's amazing you can get away with that speaker placement and still get amazing sound. Hats off to science :~)

    Share this comment


    Link to comment
    Share on other sites

    Chris, think about how many studios have soffit mounted speakers built into the wall … they work fine, but these are all professionally installed and calibrated for that location and boundary reinforcement. 

     

    The corners essentially are increasing the efficiency of the woofers.  The main problem with putting R1s closer to the corners is that the stock R1s overall frequency response is not tuned to have the extra level of bass the comes from the speakers being closer to a corner or front wall.  But Mitch can easily compensate for this, and the bonus is that the compensation then results in added bass headroom because the woofers are working in a more efficient location.  
     

    The other problem with the corner placement is side wall reflections.  That’s where the RPG treatment comes in.  These are 4” to 5” absorbers on the sides and front wall, so there is not much midrange reflection there.  And also the TADs are toed in so they point away from the side walls to a location slightly in front of the listening position.  Toe in is often used to dial in or attenuate the treble, but with the EQ dialing in the optimal level of energy at all frequencies, you can instead use the toe in to minimize reflections from the side walls. And then you rely on the EQ to set the energy levels of the midrange and treble.

    Share this comment


    Link to comment
    Share on other sites

    1 minute ago, pga said:

    Chris, think about how many studios have soffit mounted speakers built into the wall … they work fine, but these are all professionally installed and calibrated for that location and boundary reinforcement. 

     

    The corners essentially are increasing the efficiency of the woofers.  The main problem with putting R1s closer to the corners is that the stock R1s overall frequency response is not tuned to have the extra level of bass the comes from the speakers being closer to a corner or front wall.  But Mitch can easily compensate for this, and the bonus is that the compensation then results in added bass headroom because the woofers are working in a more efficient location.  
     

    The other problem with the corner placement is side wall reflections.  That’s where the RPG treatment comes in.  These are 4” to 5” absorbers on the sides and front wall, so there is not much midrange reflection there.  And also the TADs are toed in so they point away from the side walls to a location slightly in front of the listening position.  Toe in is often used to dial in or attenuate the treble, but with the EQ dialing in the optimal level of energy at all frequencies, you can instead use the toe in to minimize reflections from the side walls. And then you rely on the EQ to set the energy levels of the midrange and treble.

     

    Great point. Almost every studio I've seen has speakers placed in less than ideal positions. 

    Share this comment


    Link to comment
    Share on other sites

    1 hour ago, pga said:

    Here are some photos showing the room 

     

    Congratulations on this beautiful setup and homely room. 👍

     

    Do you cover the TV while listening to music? My TV is behind an acoustic curtain. I recently had a sound engineer with me. We pulled the curtain up once and quickly pulled it back. There are more reflections from the television than you might think.

     

    42329754xw.jpg

     

    My acoustic weak point is the table with a stool, which I have to tolerate because of the lack of space.

    Share this comment


    Link to comment
    Share on other sites

    The TV surely is not ideal, but at this point in my journey, convenience is also a big consideration.  My system all runs off an MX-450 universal remote with macros.  When the TV turns on, the UBS 5v output triggers a Parasound SCAMP signal detector / trigger that turns on my amplifiers.  It's 100% plug and play so anyone can operate the system with TV, Roon, Xfinity, AppleTV ...  Not much need for surround, center channels or woofers with a properly set up two channel system.

    Share this comment


    Link to comment
    Share on other sites

    Very interesting 2-hour video and extremely educational. Getting into the weeds here. It is interesting that for the Acourate example, Mitch uses the "Alternative to VBA" pre-filter that was discussed in the Acourate forum where the major peaks below 200Hz are assumed to be primarily minimum phase room resonances and pre-filtered first, leaving a few dB for phase correction.  Almost like using IIR filters to correct the frequency response below 200Hz first before running a convolution filter for the full correction. I've tried this many months back and found that I prefer a VBA (virtual bass array) pre-filter for the longitudinal room resonances only and then correcting the rest with the standard full frequency & phase correction over Mitch's current method. However, back then, I also really didn't know what I was doing so I must try what Mitch does now and see if I can optimize my filter more using his method.

     

    As for @pga's comment, I have owned Acourate for over a year and I continue to find myself further optimizing my convolution filter for my system. And this is in a passive crossover system, not an active crossover system. I also end up re-reading many things that Mitch wrote to understand what I'm hearing more. I think building truly optimized custom convolution filters are quite complicated and this is why Audyssey/Dirac/RoomPerfect components (at least to me) don't perform as well as a truly tailored custom convolution filter. Hence, I actually think it would be difficult for high-end dealers to provide this kind of service. Of course, I think high-end dealers should just refer their clients to Mitch Barnett.

    Share this comment


    Link to comment
    Share on other sites

    Great job Mitch. When I got back involved in the audiophile world in 2016, I looked around and found it obvious that the future of high performance audio was software. 

    Share this comment


    Link to comment
    Share on other sites

    24 minutes ago, ecwl said:

    It is interesting that for the Acourate example, Mitch uses the "Alternative to VBA" pre-filter that was discussed in the Acourate forum where the major peaks below 200Hz are assumed to be primarily minimum phase room resonances and pre-filtered first, leaving a few dB for phase correction. 

     

    I tried the VBA pre-filter today and I didn't like it. I lost 12dB. I probably did something wrong, even though the instructions are very good.

     

    The full correction with the Allpass filter, on the other hand, is excellent. Mitch showed the new Macro 6 in the Acourate pre-release. My correction is shown with a thick line.

     

    42330173tl.png

    Share this comment


    Link to comment
    Share on other sites

    5 minutes ago, StreamFidelity said:

     

    I tried the VBA pre-filter today and I didn't like it. I lost 12dB. I probably did something wrong, even though the instructions are very good.

     

    With VBA pre-filter, you shouldn't lose 12dB because you get to choose what gain you want the echo pulse to be at and you need to convolute the pre-filter with the measured pulse response to ensure it's correctly programmed. So if a gain of 0.8 causes a 12dB at the longitudinal room frequency, you need to set the gain lower, say 0.7 or 0.6 or even 0.5. Moreover, I found Uli's instructions recommends way too high a crossover for the VBA as you only need a high enough frequency to cover the longitudinal room frequency you're targeting. In my case, I have a 52Hz peak and I find that setting the upper frequency to 80Hz (instead of 150Hz) is more than enough. With either pre-filter, you always need to run a full convolution afterwards. Anyway, probably getting a little off topic.

    Share this comment


    Link to comment
    Share on other sites

    On 10/27/2021 at 4:18 PM, pga said:

    They designed a package for the room, and this was about 20 years ago.  The front wall corners have Modex bass plate bass traps.  Also absorbers on the front wall.  The ceiling and one side was has absorbers.  The back wall has diffusers and abfusers.  

    thank you ; I understand you wished something more resembling to B&K but I'm really puzzled by the response before eQ : did you ask RPG to design a flat in room response up to 10 K Hz? is a flat response a trade off for the pretty remarkable evenness you had with RPG even prior to eQ? is it due to your speakers ?

    Share this comment


    Link to comment
    Share on other sites

    4 hours ago, Ozan Bolat said:

    thank you ; I understand you wished something more resembling to B&K but I'm really puzzled by the response before eQ : did you ask RPG to design a flat in room response up to 10 K Hz? is a flat response a trade off for the pretty remarkable evenness you had with RPG even prior to eQ? is it due to your speakers ?

    RPG did this work quite a long time ago, so I don’t know how they would do this today.  That said, it’s not about having them design a room that’s flat, it’s more about treating reflection points and room modes.  RPG will identify the primary room nodes, and these are treated with bass traps.  Some may be too low in frequency for a practical approach to exist.  For example, this is the case in my room for the node caused by the 18’ dimension.  That one you just live with and treat as best you can with DSP.  Regarding the reflections, again, some are worth treating and some are not.  I didn’t want to block my window, for example.  I had removable panels made for this, but it was more work to put them up and down than it was worth.  The ceiling is an important reflection point.  Same for the front wall, but I like my TV … Back wall has diffusion and abfusion.   You don’t want parallel reflective surfaces.  But if you treat the ceiling, then the floor is less important to treat.  Same for my side walls.  The window is untreated, but the other side wall is, so you will not have flutter echo between these.  RPG will shoot for a certain reverberation time, that is relatively constant across frequencies.  Too much reverberation is sounds sloppy, too little sounds dry.  My room is closer to dry than to sloppy.  
     

    The test Mitch runs reveals much more than the frequency response, it also shows him your reverberation time, which is best addressed by room treatments.  Mitch then designs a filter with a target curve, this is what defines your frequency response.  
     

    I determined that there was just too much I didn’t know about acoustics to try so this without Mitch.  One thing I learned is our ears will fill in narrow gaps in the frequency response, even if they are deep.  But the6 are very sensitive to wider gaps even if they are shallower.  When my TADs were out into the room, they had wider gap in the bass response, from 50-80.  But in the corners, they had deeper but narrower gaps.  And in the corner that had too much deep base centered at 30 hz.  Without DSP you would not want them in the corners, but with DSP you can dial down the bass at 30hz.  So now I have flat response easily to below 20hz, with very significant headroom since the speakers are attenuated sharply below 30hz.  
     

    Also, as good as the TADs are, they are a little bright.  You can adjust this with toe in, but it’s better to point them to minimize reflections and just dial in the treble with DSP.  
     

    The side to side imagining of the TADs after Mitch EQd them is quite amazing.  And the center image is also very strong.  Most people would think I have a center channel.  It’s that good.  Also the timbre does not change as things move from left to center to right.  
     

     

    Share this comment


    Link to comment
    Share on other sites

    8 minutes ago, pga said:

    RPG did this work quite a long time ago, so I don’t know how they would do this today.  That said, it’s not about having them design a room that’s flat, it’s more about treating reflection points and room modes.  RPG will identify the primary room nodes, and these are treated with bass traps.  Some may be too low in frequency for a practical approach to exist.  For example, this is the case in my room for the node caused by the 18’ dimension.  That one you just live with and treat as best you can with DSP.  Regarding the reflections, again, some are worth treating and some are not.  I didn’t want to block my window, for example.  I had removable panels made for this, but it was more work to put them up and down than it was worth.  The ceiling is an important reflection point.  Same for the front wall, but I like my TV … Back wall has diffusion and abfusion.   You don’t want parallel reflective surfaces.  But if you treat the ceiling, then the floor is less important to treat.  Same for my side walls.  The window is untreated, but the other side wall is, so you will not have flutter echo between these.  RPG will shoot for a certain reverberation time, that is relatively constant across frequencies.  Too much reverberation is sounds sloppy, too little sounds dry.  My room is closer to dry than to sloppy.  
     

    The test Mitch runs reveals much more than the frequency response, it also shows him your reverberation time, which is best addressed by room treatments.  Mitch then designs a filter with a target curve, this is what defines your frequency response.  
     

    I determined that there was just too much I didn’t know about acoustics to try so this without Mitch.  One thing I learned is our ears will fill in narrow gaps in the frequency response, even if they are deep.  But the6 are very sensitive to wider gaps even if they are shallower.  When my TADs were out into the room, they had wider gap in the bass response, from 50-80.  But in the corners, they had deeper but narrower gaps.  And in the corner that had too much deep base centered at 30 hz.  Without DSP you would not want them in the corners, but with DSP you can dial down the bass at 30hz.  So now I have flat response easily to below 20hz, with very significant headroom since the speakers are attenuated sharply below 30hz.  
     

    Also, as good as the TADs are, they are a little bright.  You can adjust this with toe in, but it’s better to point them to minimize reflections and just dial in the treble with DSP.  
     

    The side to side imagining of the TADs after Mitch EQd them is quite amazing.  And the center image is also very strong.  Most people would think I have a center channel.  It’s that good.  Also the timbre does not change as things move from left to center to right.  
     

     

    It’s a bit hard to see, but if you compare the untreated stock TADs with the biamp version, you van see that they stock TAD is very flat overall up to 10k when it’s on axis.  The biamp version does have a small tilt down.  The XVR1 active high pass is used in conjunction with the high pass in the speakers.  Andrew Jones set this up so they would sum correctly with the woofers that are only using the low pass in the XVR1 with the passive low pass that’s in the TAD totally bi passed.

     

    It is unfortunate that high end audio has not moved in the direction of active crossovers and DSP.  It is very difficult to build a low pass at 250hz unless you use ferrite core inductors.  The ferrite core inductor in the TAD low pass is one of its weak points.  With that totally out of the circuit, my woofers are not directly connected to their amplifiers.  This made a very big difference in bass definition and dynamics.  Also now the amplifiers that drive the coax are not taxed with powering the woofers.  

    Share this comment


    Link to comment
    Share on other sites

    5 hours ago, pga said:

    It’s a bit hard to see, but if you compare the untreated stock TADs with the biamp version, you van see that they stock TAD is very flat overall up to 10k when it’s on axis.  The biamp version does have a small tilt down.  The XVR1 active high pass is used in conjunction with the high pass in the speakers.  Andrew Jones set this up so they would sum correctly with the woofers that are only using the low pass in the XVR1 with the passive low pass that’s in the TAD totally bi passed.

     

    It is unfortunate that high end audio has not moved in the direction of active crossovers and DSP.  It is very difficult to build a low pass at 250hz unless you use ferrite core inductors.  The ferrite core inductor in the TAD low pass is one of its weak points.  With that totally out of the circuit, my woofers are not directly connected to their amplifiers.  This made a very big difference in bass definition and dynamics.  Also now the amplifiers that drive the coax are not taxed with powering the woofers.  

    thank you I wondered which one was the biamp. I like the tilt down with biamp much better but strangely enough it's less clean in the 90-180 region (maybe you moved the speakers in addition?). Have you tried to simply shave off the modal peaks between 90 and 180 Hz down to #-12 dB, boost the few peaks you have in the overall  depressed region (ie the peak @ 60) a few dB and to round the sub bass bump to something like 30 Hz @ 0 dB 35 @ -3 approx, no messing with the time domain or, rather respecting modal induced modifications in the time domain*? That would be about only 5 or 6 eQ points per channel interfering, and only modal, with your natural in your treated room biamped speaker response ; just curious... On one hand I wish you would claim the full range DSP winner for it's much more affordable, OTOH, you've gone so far in the right direction with treatments and biamping that in your situation I would be tempted to do as little as possible with eQ, and only in modal region where it belongs. I understand SOTA DSP pretends to break that boundary with windowing etc but... and there's also a whole dimension of our brain/ears measuring and adapting to the room or concert hall we're in that commands limiting interfering. My 2 cents.
    *cf Bob Katz : https://www.digido.com/portfolio-item/digital-room-correction/ 

     

    Share this comment


    Link to comment
    Share on other sites




    Create an account or sign in to comment

    You need to be a member in order to leave a comment

    Create an account

    Sign up for a new account in our community. It's easy!

    Register a new account

    Sign in

    Already have an account? Sign in here.

    Sign In Now



×
×
  • Create New...