Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Parametric EQ Suggestions / DSP in Roon?
#1
Hey folks, I ran a few searches of the forum and didn't seem to find posts regarding a couple of questions I had about the DSP functions in Roon and how they cooperate with the sound of our systems.  I've often felt that the Devialet was meant to run as-is, and to let the unit and attached hardware provide the voice. 

A few nights ago I was reading something, and a suggestion was made to use a parametric EQ if available to make low-volume listening more pleasant.  I had decided initially to just let my hardware "be"—no treble or bass adjustment, room correction, etc., so a really low-load for my ROCK NUC—but the low volume issue is something that I've been dealing with.  So, I jumped in and entered the suggested settings in the Roon DSP Parametric EQ (bump of 7db at 100hz and 10000hz) and definitely noticed that lower-volume listening with my smallish standmounts (B&W 707s2) became more engaging, although it definitely introduced a coloration that I'm not sure I'm comfortable with.  

I'm running the settings now, along with an addition—a high-pass filter set at 30hz at 6db, the depth to which SAM can allegedly drive my speakers.  I believe the Devialet has a 20hz high-pass filter set by default, but I may be wrong.  My rationale for the high-pass filter caters to a realization I came to when listening to the opening track of "You Want It Darker"—which I have fondly named "You Want New Speakers" as the pulsing and undulating bassline is essentially inaudible with my current setup when set to defaults.  I figured that removing some of that low-end, barely-audible load from the main drivers might improve the overall presentation of things.  It seems to have worked.  That said, I did some googling, and apparently the crossovers inside most speakers include a high-pass filter designed to match the abilities of the speakers—but if that is the case, how can SAM work its magic by dropping the bass floor in so many models?

I also set it to upsample to 24/192 (well, the system max showing as 24/192).  

So who here is running the Parametric EQ, which settings and why, and do you upsample and why or why not?  Also, does anyone know what the story is re: the high-pass filter in speakers, and is my thinking off at all?  

I've been demoing new cables and speakers in search of the sound I want, and I certainly see the possibility of emulating some of the improvements you would realize with improved speaker cable, for example.  There are certain things a quality set of cables can do that the EQ can't emulate, but the balance between the value of upgrades vs. EQing your way to satisfaction seems like a rabbit hole that I don't quite have the knowledge to navigate.  Would appreciate any input or advice.
Reply
#2
(04-Mar-2019, 01:29)awkaplan Wrote: … That said, I did some googling, and apparently the crossovers inside most speakers include a high-pass filter designed to match the abilities of the speakers—but if that is the case, how can SAM work its magic by dropping the bass floor in so many models?…

It's hard to find any real detail about how SAM works but the little I've seen from Devialet mentions 2 things. One is control of the maximum woofer excursion which is really a speaker protection function. The other is a statement that SAM applies phase correction over the range in which it operates.

If you take a look at any speaker test reports like the ones that John Atkinson does in Stereophile, you will see that there are large phase swings in the bass region for most speakers and SAM is supposed to operate from 150 Hz on down. Check that area in the phase plots in Stereophile's speaker tests and you'll usually see some very wide ranging phase swings.

I was re-reading a bit of Floyd Toole's book "Sound Reproduction" a few months ago and I have a memory of seeing a comment to the effect that phase correction in the bass frequencies can effectively provide up to additional half an octave or so of bass extension for many speakers. Devialet claim that SAM provides more added extension than that for a lot of speakers but then I don't know what SAM is actually doing and whether it's doing more than just providing phase correction in order to deliver the extension it provides.

Whatever SAM does, it isn't changing the operation of anything in the speaker's own crossovers, it's changing the signal sent to the speaker. If the bass driver has a high pass filter operating and controlling its roll off at the low end of the speaker's range then that high pass filter is still going to be doing it's thing but it probably isn't a "brick wall" filter with a steep rolloff and it's probably operating with a 6 or 12 dB rolloff per octave. Let's say it's 12 dB/octave with a frequency of 20 Hz. That means that the level will be down 12 dB at 10 Hz from what it was at 20 Hz and that's roughly equivalent to a bit over a halving of perceived volume. While the speaker will be rolling off, the room could be providing considerable support at those frequencies depending on the position of the speaker in relation to adjacent walls and also on the location of the listening position and in-room response at low frequencies is usually quite different to the speaker's specified frequency response which is usually an anechoic measurement. I don't know how Devialet measure speakers in order to determine the frequency to which SAM provides extension but it is probably not the same way that the speaker manufacturer measured the speaker's response for the specification info they provide and that could make a big difference because some speakers with good setup in a room are capable of providing more than half an octave of bass in room over what their specification suggests. Add that to whatever extension SAM delivers and you could be seeing the sort of in room response claimed for a speaker using SAM ***PROVIDED*** you're in the room in which Devialet made their measurements and listening where Devialet had their microphone located. In another room with different speaker and listening position setup you might notice more bass with SAM but measurements might fall short of the range stated in the speaker's specifications. Low frequency in-room response of speakers is "highly variable" and often bears little similarity to what is claimed in the speaker's specifications, especially if the specifications are based on anechoic measurements.
Antipodes DS, Devilalet Expert 140 Pro CI, Focal Sopra 2, PS Audio P12, AQ Vodka ethernet, Kimber Select speaker cables, Shunyata Alpha NR and Delta EF power cables power cables, Grand Prix Audio Monaco rack, RealTRAPS acoustic treatment.

Brisbane, Qld, Australia
Reply
#3
(04-Mar-2019, 03:25)David A Wrote:
(04-Mar-2019, 01:29)awkaplan Wrote: … That said, I did some googling, and apparently the crossovers inside most speakers include a high-pass filter designed to match the abilities of the speakers—but if that is the case, how can SAM work its magic by dropping the bass floor in so many models?…

It's hard to find any real detail about how SAM works but the little I've seen from Devialet mentions 2 things. One is control of the maximum woofer excursion which is really a speaker protection function. The other is a statement that SAM applies phase correction over the range in which it operates.

If you take a look at any speaker test reports like the ones that John Atkinson does in Stereophile, you will see that there are large phase swings in the bass region for most speakers and SAM is supposed to operate from 150 Hz on down. Check that area in the phase plots in Stereophile's speaker tests and you'll usually see some very wide ranging phase swings.

I was re-reading a bit of Floyd Toole's book "Sound Reproduction" a few months ago and I have a memory of seeing a comment to the effect that phase correction in the bass frequencies can effectively provide up to additional half an octave or so of bass extension for many speakers. Devialet claim that SAM provides more added extension than that for a lot of speakers but then I don't know what SAM is actually doing and whether it's doing more than just providing phase correction in order to deliver the extension it provides.

Whatever SAM does, it isn't changing the operation of anything in the speaker's own crossovers, it's changing the signal sent to the speaker. If the bass driver has a high pass filter operating and controlling its roll off at the low end of the speaker's range then that high pass filter is still going to be doing it's thing but it probably isn't a "brick wall" filter with a steep rolloff and it's probably operating with a 6 or 12 dB rolloff per octave. Let's say it's 12 dB/octave with a frequency of 20 Hz. That means that the level will be down 12 dB at 10 Hz from what it was at 20 Hz and that's roughly equivalent to a bit over a halving of perceived volume. While the speaker will be rolling off, the room could be providing considerable support at those frequencies depending on the position of the speaker in relation to adjacent walls and also on the location of the listening position and in-room response at low frequencies is usually quite different to the speaker's specified frequency response which is usually an anechoic measurement. I don't know how Devialet measure speakers in order to determine the frequency to which SAM provides extension but it is probably not the same way that the speaker manufacturer measured the speaker's response for the specification info they provide and that could make a big difference because some speakers with good setup in a room are capable of providing more than half an octave of bass in room over what their specification suggests. Add that to whatever extension SAM delivers and you could be seeing the sort of in room response claimed for a speaker using SAM ***PROVIDED*** you're in the room in which Devialet made their measurements and listening where Devialet had their microphone located. In another room with different speaker and listening position setup you might notice more bass with SAM but measurements might fall short of the range stated in the speaker's specifications. Low frequency in-room response of speakers is "highly variable" and often bears little similarity to what is claimed in the speaker's specifications, especially if the specifications are based on anechoic measurements.

Interesting.  Yes, I understand that SAM is not changing the operation of the crossovers.  It would seem that the final signal sent, with SAM activated, might provide a certain push that challenges the high-pass filters in the crossovers, thus allegedly dropping the potential low-end response.

So I've shut off the Parametric EQ for now as it was becoming distracting, and I also noticed that the low-end boost produced undesirable results with certain tracks.  That said, is my hypothesis about reducing load on the main driver by using a high-pass filter in Roon at say, 30hz/6dB, reasonable?  Yes, it could be argued that the speaker is designed to accept a full-range signal, so why bother.  I don't understand the physics of drivers.  But it would seem that relieving it of certain duties that are at the threshold of its ability would perhaps help.  It seemed to do so earlier today, but the 100hz / +7dB / Q1 filter didn't play well with a few tracks, so I just shut the whole thing off. 

One thing that annoys me about my current speakers is the reproduction of some digital bass as sort of "pings" where the attack is overpowering yet the follow-through that I know is present on the track is weak and almost inaudible, or a very deep note rendered barely audible, or low bass intruding on the midrange (which could also be a factor involving the recording).  Yes, there are arguments regarding coherence and sticking with a simple 2-way, 2-way vs. 3-way or more and the cost/benefit, seemingly endless design tricks, tweaks, what have you.  I do wonder if the aforementioned issues would be eliminated with a 3-way (or more) speaker, or if it's a factor that even matters.  I'm assuming at least a very competent bass/midrange cone would circumvent the issues I mentioned above. 

The standmount/bookshelf is more appropriate for my listening space, and I'm hoping that the more capable options at higher price points won't have the same shortcomings/trade-offs that I'm working with now.  

The SAM claims can be a bit outlandish prima facie, especially looking at something like the Harbeth M30.1 and a claimed floor around 22hz dropped from the 40s.  And perhaps my thoughts on the high-pass filter could be beneficial.  From head of Harbeth: "the harder a speaker - any speaker -is driven, especially in the bass region, the inevitable higher the distortion, which is linked to cone movement. So to boost the bass output of a speaker, there surely will be a reduction in bass fidelity as the bass unit must be moving further and working harder - harder than the designer intended."
Reply
#4
Many questions in one post @awkaplan . I shall try to answer some of them.

Devialet does not impose a 20hz high-pass filter by default. Or at least not on the digital inputs. The RIAA 1976 phono stage equalisation curve does include a high pass filter, intended to remove the low-frequency rumble that you can get with many turntables.

As for SAM, it does provide a degree of bass boost by increasing the gain at lower frequencies. So if a speaker is rolling off by -4dB at 50Hz, SAM will add +4dB, but will ultimately restrict this to avoid exceeding the speakers thermal rating or maximum bass driver excursion. The SAM phase correction works 100% when SAM is enabled, but the SAM % adjustment can be used to adjust the amount of bass gain / boost that SAM applies. As @David A explains above, the room and speaker positioning can have a huge effect on the actual in-room bass response, so the SAM % adjustment can be used to fine tune the overall level of bass to suit your room and taste. One point to clarify, the SAM measurements are taken at close proximity to the speaker using a laser to measure driver movement, no microphones are involved, so the measurements themselves are not influenced by the room. The links to videos below explain this in detail.

https://www.youtube.com/watch?v=AcLYb4sBhwQ

https://darko.audio/2016/09/waiting-for-...ts-samlab/

Upsampling is a very complex topic. Upsampling can be used as a tool to eliminate some of the unwanted side effects of digital to analogue conversion. For example, upsampling to 24/192 is supposed to help remove the unwanted "aliasing" effect created by digital to analogue conversion. This is not a simple topic and relates to the various filters that can be used. As an example, Roon allows you to select four different filters, some software applications such as HQPlayer have many more. I think the filters make only a very subtle difference, so this one is largely to be determined by personal preference or fine-tuning a particular system. Indeed, I know of some that select particular filters depending on the type of music that they are listening to, so there is no absolute right answer here. It is worth mentioning that the Devialet upsamples all digital feeds to 24/192 anyway. One argument for performing 24/192 upsampling in the software is that the Devialet then has to do less work dealing with the incoming signal, thus reducing noise in the amp. This is a controversial area, and I would say the effects are subtle at best, but the is no reason not to experiment to find out what you prefer, if you are so inclined.

Then you mention parametric eq. This is something that can produce some very unsubtle effects! You change the parametric eq, the system will sound different. I have been experimenting with parametric eq, for me though this is to compensate for different recordings. With some recordings, my system sounds superb, just right. With other recordings, it can sound overly bright or thin, so I have some eq settings to produce a gradual drop off at higher frequencies. I am still experimenting with this, but it does seem to work on some of the old 16/44 rips that I have, where the original recording is not the best.

In terms of using parametric eq to make low-level listening sound better, this is something covered by the "Fletcher Munson Curve", which determines how the how human hearing perception of frequency response varies with volume. Or in simple terms, music at low volumes can sound better if you tweak the bass and treble up a touch. This topic is covered extensively on the 'net, so maybe worth some time researching if you are interested.

The Roon parametric eq function is extremely powerful, it is also fairly complex. One suggestion I would make is that you start by experimenting with something far more simple, that is simply using the Devialet's tone controls to tweak the bass and treble slightly. When you start to get some clarity with how this is influencing the enjoyment of your system, then you could move to using Roon parametric eq to fine tune this further to suit your taste and system.

I will sign off by saying that use of Roon EQ and indeed the Devialet's tone controls are for me an interesting area. I have very distant memories as a child of my parents "Radiogram", which featured a "Loudness Button", which invoked some kind of Fletcher Munson compensation, or maybe it just boosted the bass and treble a bit. Then later in life, we had an era when any audio product that featured a tone control was deemed to be inherently rubbish, and the good stuff did not include such nasty devices that would reduce the purity of the signal. Because of this I still feel that turning the treble down by 2dB or something is somehow cheating, but I think I need to get over this, because simple adjustments like this can work absolute wonders with some recordings that I have. For me, it would indeed be very interesting to get a summary from others using the Devialet tone controls, Roon EQ or similar, to fine-tune their systems, what do you use and why?
1000 Pro - KEF Blade - SOtM sMS-200Ultra Neo - tX-USBultra - Mutec REF10 - MC3+USB - Pro-Ject Signature 12
Reply
#5
I am assuming you use one of the Expert models since SAM is discussed. There is also a ICM toggle in the Devialet menu (after Balance, Bass, Treble...) This is 'Intelligent Cinema Mode'. I do not use this myself, but I think this may do some of what you want for low SPL listening. As Devialet says: "This mode optimizes the dynamics of the incoming audio content, allowing you to enjoy your favourite films without having to make any compromise on experience or power."
I notice that Devialet is not talking about ICM on their new web-site so maybe this is a feature scheduled for removal. It is still present in the current firmware though so worth a try.
*
microRendu 1.4::LPS-1::USPCB::Mutec MC-3+USB::250 Pro CI::Roon life, Core on Intel NUC with Debian Linux, music on USB3 disk::Audiolense 5.3::Monitor Audio MA201 (thoroughly rebuilt, now with a SAM profile)


Reply
#6
(04-Mar-2019, 13:21)Confused Wrote: Many questions in one post @awkaplan .  I shall try to answer some of them.

Devialet does not impose a 20hz high-pass filter by default.  Or at least not on the digital inputs.  The RIAA 1976 phono stage equalisation curve does include a high pass filter, intended to remove the low-frequency rumble that you can get with many turntables.

As for SAM, it does provide a degree of bass boost by increasing the gain at lower frequencies.  So if a speaker is rolling off by -4dB at 50Hz, SAM will add +4dB, but will ultimately restrict this to avoid exceeding the speakers thermal rating or maximum bass driver excursion.  The SAM phase correction works 100% when SAM is enabled, but the SAM % adjustment can be used to adjust the amount of bass gain / boost that SAM applies.  As @David A explains above, the room and speaker positioning can have a huge effect on the actual in-room bass response, so the SAM % adjustment can be used to fine tune the overall level of bass to suit your room and taste.  One point to clarify, the SAM measurements are taken at close proximity to the speaker using a laser to measure driver movement, no microphones are involved, so the measurements themselves are not influenced by the room.  The links to videos below explain this in detail.

https://www.youtube.com/watch?v=AcLYb4sBhwQ

https://darko.audio/2016/09/waiting-for-...ts-samlab/

Upsampling is a very complex topic.  Upsampling can be used as a tool to eliminate some of the unwanted side effects of digital to analogue conversion.  For example, upsampling to 24/192 is supposed to help remove the unwanted "aliasing" effect created by digital to analogue conversion.  This is not a simple topic and relates to the various filters that can be used.  As an example, Roon allows you to select four different filters, some software applications such as HQPlayer have many more.  I think the filters make only a very subtle difference, so this one is largely to be determined by personal preference or fine-tuning a particular system.  Indeed, I know of some that select particular filters depending on the type of music that they are listening to, so there is no absolute right answer here.  It is worth mentioning that the Devialet upsamples all digital feeds to 24/192 anyway.  One argument for performing 24/192 upsampling in the software is that the Devialet then has to do less work dealing with the incoming signal, thus reducing noise in the amp.  This is a controversial area, and I would say the effects are subtle at best, but the is no reason not to experiment to find out what you prefer, if you are so inclined.

Then you mention parametric eq.  This is something that can produce some very unsubtle effects! You change the parametric eq, the system will sound different.  I have been experimenting with parametric eq, for me though this is to compensate for different recordings.  With some recordings, my system sounds superb, just right.  With other recordings, it can sound overly bright or thin, so I have some eq settings to produce a gradual drop off at higher frequencies.  I am still experimenting with this, but it does seem to work on some of the old 16/44 rips that I have, where the original recording is not the best.

In terms of using parametric eq to make low-level listening sound better, this is something covered by the "Fletcher Munson Curve", which determines how the how human hearing perception of frequency response varies with volume.  Or in simple terms, music at low volumes can sound better if you tweak the bass and treble up a touch.  This topic is covered extensively on the 'net, so maybe worth some time researching if you are interested.

The Roon parametric eq function is extremely powerful, it is also fairly complex.  One suggestion I would make is that you start by experimenting with something far more simple, that is simply using the Devialet's tone controls to tweak the bass and treble slightly.  When you start to get some clarity with how this is influencing the enjoyment of your system, then you could move to using Roon parametric eq to fine tune this further to suit your taste and system.

I will sign off by saying that use of Roon EQ and indeed the Devialet's tone controls are for me an interesting area.  I have very distant memories as a child of my parents "Radiogram", which featured a "Loudness Button", which invoked some kind of Fletcher Munson compensation, or maybe it just boosted the bass and treble a bit.  Then later in life, we had an era when any audio product that featured a tone control was deemed to be inherently rubbish, and the good stuff did not include such nasty devices that would reduce the purity of the signal.  Because of this I still feel that turning the treble down by 2dB or something is somehow cheating, but I think I need to get over this, because simple adjustments like this can work absolute wonders with some recordings that I have.  For me, it would indeed be very interesting to get a summary from others using the Devialet tone controls, Roon EQ or similar, to fine-tune their systems, what do you use and why?

Yep, a lot to unpack.  Thanks for bearing with me.  Not sure where I saw the 20hz default in the configurator—could've been a misread.  I am familiar with the way SAM profiles are generally created, including the use of laser measurement.  Still a little confused about the speaker crossover situation and high-pass filters within the speakers as they interact with SAM.

It's funny, I'm also very reluctant to touch the tone controls within the unit itself.  Part of the reason I even started experimenting with the Parametric EQ is that I'm currently doing a few rounds of demos of speaker cable, and while I'm in limbo I have to go back to using my own cables, lower-end AQ, albeit factory-finished and biwired.  I was accidentally sent a pair of Shunyata Venoms in an unusual double-biwire configuration, and after a week listening through that much copper I feel like I'm in the penalty box and am looking to reproduce some of what I heard, temporarily.

I've left the upsampling on for the time being, and may experiment more when the next set of demos arrives.  Ideally, I'd like to let the hardware speak for itself.  My listening area doesn't present much of a challenge, but I may also experiment with convolution and return to the EQ down the road if I'm missing something with my new speakers and cables.
Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)