DRC Linear Phase

[dfgoiuj]

Hi all,

I've been using DRC Designer to simplify the creation of room correction FIR filters using Denis Sbragion's DRC utility. I generally pick the BK3 curve and the 'normal' preset at 48kHz which loads the config from the normal48000.drc file. I understand that this is a minimum phase filter and I'd like to try linear phase to see if I prefer it. I find the documentation overwhelming. Has anyone got a .drc file I could try that does linear phase? Thanks in advance!

 

[palm]

I never used DRC designer and I haven’t experimented with DRC since a long time, but as far as I remember the normal preset does include phase correction.
When running DRC to compute the filters, I think a minimum phase version was generated systematically, do you see these?
If you look at the plots of the step response before and after correction it should also be quite visible.

 

[dfgoiuj]

I never used DRC designer and I haven’t experimented with DRC since a long time, but as far as I remember the normal preset does include phase correction.
When running DRC to compute the filters, I think a minimum phase version was generated systematically, do you see these?
If you look at the plots of the step response before and after correction it should also be quite visible.

Thanks for the reply and sorry for the late response. I've been away on holiday and forgot about this post. There are multiple config files, with names such as minimum48000.drc, normal48000.drc, strong48000.drc etc., but digging into them it seems that they are all minimum phase as they have a little note to say as much which refers to the postfiltering stage. This appears to make my question seem even more naive as there are multiple stages to each filter, and most have phase settings.

When you were using DRC, did you stick with the example config files, or did you create/modify a config to suit your system? What do you use instead these days?

Something else I found interesting in the config file notes was a comment about how peak limiting is set to 6dB. Again, this may demonstrate my naivety, but I find myself having to set the gain in CamillaDSP to -12dB to avoid clipping, and I wonder why this is. That's a lot of headroom to lose.

 

[palm]

No, most of the tuning was in the target response. That’s the thing: audibility of the magnitude is very high compared to the phase. This is why if you rapidly cycle through no correction/minimum phase filter/full correction filter, you won’t notice much difference between the last two. On the other hand, if you dislike the sound after correction you mainly have play with the target response. In the documentation, the author advised to use insane or strong correction to become familiar with the artifacts you need to pay attention to. You should be able to experiment by just playing with those presets without modifying them and put most of your efforts in the target curve instead.

DRC is where I started this journey. Later, I used Audyssey in my receiver, and then purchased IK ARC for my stereo setup. This is a minimal phase correction, but with multiple measurement points and it was easy to adjust the target response. I was tweaking it follow broadly the natural response of my system. It also had a double gain adjustment as far as I remember: one to avoid clipping because of the filter, and another one to bring the “gain” without correction as close as possible. It allows to compare the sound with and without the filter enabled without being too much biased by a difference of volume.

Of course it left me wondering if the lack of phase correction was big loss. This is where I tweaked DRC configuration files to let it correct only the phase and not the amplitude. Then I used rePhase to correct for the crossovers. You see that the step response is already much closer to ideal with just this theoretical correction. This is what an ideal room correction software should do: start from what is related to your speakers and then gently correct for the additional room effects. The conclusion of all these trials was that changes in the phase response weren’t that audible, if audible at all in my far from perfect room. In the end I was using ARC + rePhase and most importantly a loudness filter to adjust the volume.

A few years ago I moved to a new place and I still have to put the system together! I am currently hesitating between a 2ch streamer, DAC or soundcard, and a 4ch one that would allow to me to fiddle with the bass management (in all my previous experiments I was just using what’s provided by my K+H O800)

 

[jlo]

Hi all,

I've been using DRC Designer to simplify the creation of room correction FIR filters using Denis Sbragion's DRC utility. I generally pick the BK3 curve and the 'normal' preset at 48kHz which loads the config from the normal48000.drc file. I understand that this is a minimum phase filter and I'd like to try linear phase to see if I prefer it. I find the documentation overwhelming. Has anyone got a .drc file I could try that does linear phase? Thanks in advance!

Hello,
on loudspeakers.audio, you get min phase and lin phase corrections so you can easily listen and compare. Moreover, you'll get a target calculated and optimised for your speakers and room.

 

[dfgoiuj]

No, most of the tuning was in the target response. That’s the thing: audibility of the magnitude is very high compared to the phase. This is why if you rapidly cycle through no correction/minimum phase filter/full correction filter, you won’t notice much difference between the last two. On the other hand, if you dislike the sound after correction you mainly have play with the target response. In the documentation, the author advised to use insane or strong correction to become familiar with the artifacts you need to pay attention to. You should be able to experiment by just playing with those presets without modifying them and put most of your efforts in the target curve instead.

DRC is where I started this journey. Later, I used Audyssey in my receiver, and then purchased IK ARC for my stereo setup. This is a minimal phase correction, but with multiple measurement points and it was easy to adjust the target response. I was tweaking it follow broadly the natural response of my system. It also had a double gain adjustment as far as I remember: one to avoid clipping because of the filter, and another one to bring the “gain” without correction as close as possible. It allows to compare the sound with and without the filter enabled without being too much biased by a difference of volume.

Of course it left me wondering if the lack of phase correction was big loss. This is where I tweaked DRC configuration files to let it correct only the phase and not the amplitude. Then I used rePhase to correct for the crossovers. You see that the step response is already much closer to ideal with just this theoretical correction. This is what an ideal room correction software should do: start from what is related to your speakers and then gently correct for the additional room effects. The conclusion of all these trials was that changes in the phase response weren’t that audible, if audible at all in my far from perfect room. In the end I was using ARC + rePhase and most importantly a loudness filter to adjust the volume.

A few years ago I moved to a new place and I still have to put the system together! I am currently hesitating between a 2ch streamer, DAC or soundcard, and a 4ch one that would allow to me to fiddle with the bass management (in all my previous experiments I was just using what’s provided by my K+H O800)

Very interesting, thank you for sharing your DSP journey. So at one point you tried using IK ARC for amplitude correction along with DRC and rePhase together for phase correction, and concluded that you only needed the phase correction for the crossover? I guess that still introduces latency into the system?

I will have a play around with the presets that come with DRC. The documentation talks about being able to specify phase correction values per frequency in the points file if linear phase is enabled in the post filter stage, but also states "If not specified a value of 0 is assumed. Setting a phase different than 0, i.e. flat, is useless within normal HiFi systems in almost all circumstances". So from what I understand, leaving these values out will still correct the phase, but there is an option to adjust the phase at different points for whatever reason. I could do with a course on DSP!

 

[dfgoiuj]

Hello,
on loudspeakers.audio, you get min phase and lin phase corrections so you can easily listen and compare. Moreover, you'll get a target calculated and optimised for your speakers and room.

Registering! Thanks

 

[palm]

Very interesting, thank you for sharing your DSP journey. So at one point you tried using IK ARC for amplitude correction along with DRC and rePhase together for phase correction, and concluded that you only needed the phase correction for the crossover? I guess that still introduces latency into the system?

I will have a play around with the presets that come with DRC. The documentation talks about being able to specify phase correction values per frequency in the points file if linear phase is enabled in the post filter stage, but also states "If not specified a value of 0 is assumed. Setting a phase different than 0, i.e. flat, is useless within normal HiFi systems in almost all circumstances". So from what I understand, leaving these values out will still correct the phase, but there is an option to adjust the phase at different points for whatever reason. I could do with a course on DSP!

DRC in “phase correction only” was just a trial. In an untreated room it can be hard to read the phase plot, you need so much smoothing that in the end I felt like if the only correction that really made sense was the crossovers in the system. In fact this is what you get nowadays with the new Neumann monitors or subwoofer with a DSP. You get a nicer step response for sure, but I wouldn’t say I heard a significant improvement. I also didn’t get dramatic soundstage improvement from the “stereo” (per channel) correction with DRC. I suspect in some conditions (large asymmetry of speakers positions vs. walls) this might be more beneficial.

This was a relatively large, 7x9m or so, open space living room, with new standard low ceiling and one side was entirely glass, and the speakers were used for home theater as well with little possibility to change the position. A flawed setup and maybe in better conditions the phase correction would be more audible. But it didn’t harm, so I was running ARC + rePhase theoretical correction (same for both channels).

Latency yes of course, not a problem when you run from a computer, though. It depends on how low you want to correct. I wasn’t correcting for the bass reflex alignment for instance, only the 90Hz of the O800.

 

[dfgoiuj]

DRC in “phase correction only” was just a trial. In an untreated room it can be hard to read the phase plot, you need so much smoothing that in the end I felt like if the only correction that really made sense was the crossovers in the system. In fact this is what you get nowadays with the new Neumann monitors or subwoofer with a DSP. You get a nicer step response for sure, but I wouldn’t say I heard a significant improvement. I also didn’t get dramatic soundstage improvement from the “stereo” (per channel) correction with DRC. I suspect in some conditions (large asymmetry of speakers positions vs. walls) this might be more beneficial.

This was a relatively large, 7x9m or so, open space living room, with new standard low ceiling and one side was entirely glass, and the speakers were used for home theater as well with little possibility to change the position. A flawed setup and maybe in better conditions the phase correction would be more audible. But it didn’t harm, so I was running ARC + rePhase theoretical correction (same for both channels).

Latency yes of course, not a problem when you run from a computer, though. It depends on how low you want to correct. I wasn’t correcting for the bass reflex alignment for instance, only the 90Hz of the O800.

Interesting what you say about having no dramatic soundstage improvement with DRC. I wouldn't say I have a dramatic improvement, but it's definitely noticeable in both situations where I use it - one in the office with a pair of stand-mounted Focal Alpha 65, both positioned close to the wall, one close to a corner, and not that it matters but I'm running the filter on Equalizer APO on a Windows laptop -> famous Apple USB-C dongle -> cheap volume controller. This is a small room with a dual-height ceiling, and I think it's only around 2.5m x 2.5m. As you can probably imagine, this is a near-field setup. And the other is in my living room, which is a pair of wall-mounted ELAC DBR-62, positioned in each corner due to where the door is, and toed in to cross in front of the listening position. There's a 93L down-firing ported subwoofer in the corner under one of them with cutoff set to 100Hz and volume boosted in an effort to reduce the load on the ELACs once corrected. Again, not that it matters, but this one runs the filter on CamillaDSP on a Raspberry Pi -> Topping E30 DAC -> Fosi V3 -> sub amp -> speakers. It's bigger than the office but modestly sized at around 4m x 4m.

Anyway, you're probably on to something regarding speaker positions considering my setups. Neither have ideal positioning, and both appear to benefit from DRC. I hesitate to describe it, but if I had to, I'd say... the veil is lifted? Just kidding, but tracks do somehow sound less muddied / more precise, and the soundstage seems to move closer to me. I've not done any blind tests, but because I can switch between having it on and off pretty much instantly (I try to volume match between profiles too), it makes it easy for me to hear the difference, and I much prefer it on. I don't know if it's the better balance between channels and/or the smoothing of the frequency response, or something else.

 

[palm]

While the setup on which I made these tests was far from optimal, one nice thing was that both speakers were far from lateral walls and listening position far from the back wall with a bookshelf behind. But I’m pretty sure there are situations in which the alignment of both channels is really beneficial. In fact my car audio has an option to optimize for the driver instead of all occupants and this is probably done with a FIR. It takes a couple of seconds to adapt to the different sound and it’s mostly phase related I think (never measured but the perception is that overall response is unchanged)