Acoustic treatment finished - your thoughts on the results?

[Skol303]

Greetings John Sayers folk!

This is my first post here. I normally frequent Gearslutz and am familiar with some of you from that particular forum (notable hellos to DanDan and Jason Foi, amongst others).

I've just finished treating my room - see images and link to REW file included below.

...and as a final step I'd like to solicit some objective opinion on how "good" the end results really area (especially as I've been very close to this project for about a year and it's become difficult to judge it without my own bias). I'm aware that this community has a different membership to other forums, so I thought it would be a good place to glean some new perspectives and insights. Hopefully so

So here are the details. The general questions being:

• 1) Would you be happy with this room?
  2) Are there any issues that stand out as needing further attention?

All comments very gratefully received.

----------

Room location: UK
Room dimensions: 5.35 x 2.17 x 2.35m (L/W/H)
Monitoring set to K14 at ~76dB
Acoustic treatment: extensive fibre-based traps with some membrane traps targeting low frequency modes (30-35Hz and 65-70Hz). Room surface area is approximately 50% treated.
Digital Signal Processing: Dirac Live.

The Room

Front wall:

Room_front.jpg

Rear wall:

Room_back.jpg

Plan view (ceiling panels 'hidden'):

Room_plan.jpg

Room Measurements

Frequency response at listening position:

1_Freq response.jpg

Frequency response averaged from 18 measurements within 1m square of listening position (L/R speaker averaged at 9 spots):

2_Freq response_average.jpg

Waterfall:

3_Waterfall.jpg

Impulse Response:

4_Impulse response.jpg

Group Delay:

5_Group Delay.jpg

RT60 (Topt):

6_Topt.jpg

A copy of the REW file can be downloaded here:

https://www.dropbox.com/s/11v3f2pucahh9 ... .mdat?dl=0

[Soundman2020]

Hi there "Skol303". Please read the forum rules for posting (click here). You seem to be missing a couple of things!

This is my first post here. I normally frequent Gearslutz and am familiar with some of you from that particular forum (notable hellos to DanDan and Jason Foi, amongst others).

Yup! I've seen you over there. Nice to have you over here too! And congrats on completing your room!

...and as a final step I'd like to solicit some objective opinion on how "good" the end results really area (especially as I've been very close to this project for about a year and it's become difficult to judge it without my own bias).

Smart move! It's easy to get too close to the room, and loose objectivity.

I'm aware that this community has a different membership to other forums, so I thought it would be a good place to glean some new perspectives and insights

Very true! To start with, there are no vested interested here: No adds, no hidden agendas, no product promotions, nobody trying to sell you anything, or promote their own pet solution that has no basis in anything but pixie dust (or money...) ... John strictly forbids all types of advertising here, the forum is supported by donations alone, so the only time you'll see any products or services recommended here, is because they really do work, and have been proven to do so. Not because the forum or anybody associated with it gets a kickback or royalties or other "benefits" going under the table. Folks here help out because they WANT to help out. Most are giving back to the community, after getting lots of help in designing and/or building their own places.

Plus, everything you see here is based on only two things: 1) Solid, sound, proven acoustic science (no fantasies, moonbeam dreams, or wild speculation). 2) A proven track record in real studio builds, real people, who are real happy, going back decades. John has been designing and building studios longer than some folks on GS have been alive! So I kinda think he knows what he's doing.... As demonstrated by the numerous fantastic studios you see here, both the ones the John designed for customers (click the link at the top right corner of this page), and also the ones he has offered his free advice on, for numerous forum members. There's nobody better in the business, in my opinion, and John is a pretty humble guy, actually (even though he has totally earned the right to be arrogant, proud, boastful, and stuck-up, if he wanted to be: but he doesn't).

OK, so that's where the plug for the forum ends! Let's get on to your place...:

Room dimensions: 5.35 x 2.17 x 2.35m (L/W/H)

That's a long, thin room! An acoustic challenge, for sure. As you already discovered!

So here are the details. The general questions being:

1) Would you be happy with this room?
2) Are there any issues that stand out as needing further attention?

Well, it seems like you are asking for "brutal honesty", and I'm sort of known for that around here ( ), so please don't take offense at anything I say... it's meant to be constructive, not destructive. You asked for an honest comments, so I'll do my best to "tell it like it is", calling it the way I see it.

First, the room is too dry. The overall decay time (often incorrectly called "RT-60"), is around 100 ms across most of the spectrum. For that size room, I would suggest something more like 200 ms. There's too much absorption, not enough reflection/diffusion. That is clear from the RT60 plot:

Skol303--REW--RT--40-12k--R.png

That's taken directly from your MDAT file, for the right speaker. The left is similar, but combined as LR they look at lot better... fictitiously....

What does that mean, subjectively? It probably sounds nice in there initially, but I would think it is a bit fatiguing to be in there for long sessions, since it is unnaturally dead. Similar to the problems with the original LEDE designs of yore, and some more contemporary designs to a lesser extent (Eg, NER). They sound good at first impression, but are mentally tiring to work in for long periods. That would be my best guess.

What can you do about it? Open up some of your pure absorption devices, and put reflective surfaces inside, to return some of the energy to the room again, instead of just sucking it up and converting it into heat. Eg. thick plastic strips across the front faces. But do it selectively, so that you address the right frequencies. Another option might be non-numeric based diffusers, such as untuned broad slats, or poly-cylindrical diffusers, placed in front of your absorbers. There are options. You could recover some of that "liveness" that is lacking at present.

Second: your low end is looking good: you have bass trapping under control nicely. Well done! But it is probably the very bass trapping that is also sucking out the mids and highs, and at least some of it would need the above solutions.

Third: you are obviously using some type of digital tuning (often incorrectly called "room correction"), either in hardware or software, but you are chasing the wrong goals.

Yes, you have almost ruler flat Frequency Response (FR), which looks impressive to the untrained eye, but you have made the mistake of assuming that flat FR is the most important aspect of acoustic response: it isn't. That's a very common mistake, though, so don't feel too bad. Lots of people want their FR to be flat, thinking that it shows the room is performing perfectly. It doesn't. FR is important, but not the biggest issue. Time-domain response is the biggest issue. And you have gone after the less-important at he expense of the more-important:

Skol303--REW--WF--20-1k--R.png

Skol303--REW--IR--180ms--R.png

The top one is your waterfall plot for the right speaker, and there is clearly some ringing going on there, at 104, 148, 166, 205 and a few other places. I'm pretty sure those are filters where you are boosting some frequencies too much, into instability. Technically, you have either a pole or a zero outside the unit circle. Non-technically: it sounds like crap! That needs fixing. This is one of the reasons why I don't recommend digital tuning until people fully understand all the issues.

The second graph above is your Impulse Response graph, and shows the other issue you should be shooting for: smooth decay. As you can see, the Schroeder integral (black line) is not smooth: it wobbles up and down, indicating that the sound is not decaying evenly over time: it decays faster, then slower, then faster, then slower.... The goal is to have it decay smoothly, such the Schroeder integral is a straight line (ideally), or at least a smooth curve.

So based on that, I'd say that you have the impression that the room is not right in the bass: it seems a bit "off". Not as tight as you expected, but something subtle that you can't really put your finger on... it just seems slightly strange...

Here's some graphs from a room I'm working on right now, for one of my clients in California. We are in the final stages of tuning that room, nearly finished, but here's where we are right now. That room is a little larger than yours, but not by much. They are both small rooms, far below the recommended minimum size. Your room floor area is 11m2 floor area, this room is 13.3m2. Your room volume is 26m3, this one is 39m3. Even worse, this room is square! At least yours isn't.

Below, I'm showing the same three graphs as I did above, also using only the right speaker, to get the comparison as fair as possible:

Decay times (full spectrum):

FRANK--REW--RT--40-12k--R.png

Waterfall plot (low end):

FRANK--REW--WF--12-500--R.png

Impulse response:

FRANK--REW--IR--180ms--R.png

As you can see, the frequency response is slightly less smooth than for your place, but still pretty darn flat, and I am very happy with it, because of the other two graphs: you can see that there is no ringing going on, and the decay is very smooth, the overall RT-60 time is around 200 ms, and the Schroeder integral is smooth, and very close to being a straight line. The room is smooth, even, flat, solid, and tight. All the way down to about 14 Hz. (Yes, really: 14Hz. Check the scale on this graphs).

So, hopefully that is helpful!

It would be good if you were to post another set of REW tests, with the "room correction" turned off, so we can see just how the room itself is doing.


- Stuart -

[Skol303]

Wow, thanks Stuart! Really appreciate you taking the time to write such a detailed critique. Very useful and exactly the sort of feedback I was looking for

Hi there "Skol303". Please read the forum rules for posting (click here). You seem to be missing a couple of things!

Oops! I’ve now added my location to my profile (UK). I think all other sins were absolved, but let me know if not.

...there are no vested interested here... Most are giving back to the community, after getting lots of help in designing and/or building their own places.

I’ve been having a browse and I think that community spirit really shows here. Solid advice and without some of the egos found elsewhere. And some familiar names here too, which is nice.

Well, it seems like you are asking for "brutal honesty", and I'm sort of known for that around here ( ), so please don't take offense at anything I say...

Not at all! I’m not interested in sugar-coating, so “brutal honesty” is very welcome.

First, the room is too dry... Open up some of your pure absorption devices, and put reflective surfaces inside, to return some of the energy to the room again, instead of just sucking it up and converting it into heat.

Nice idea. I had wondered about the dryness of the room and previously considered adding some slats (or other partially-reflective surface) to some of my absorbers. I might re-visit that thinking as it’s something I could probably DIY relatively easily.

Second: your low end is looking good: you have bass trapping under control nicely.

Thanks. Yeah… my goal has been to very much pummel the low end into submission with absorbers and some tuned membrane traps on the back wall (including some fairly hefty ones targeting 30Hz, which I wheel into position when mixing - as my back wall also features a door!).

Third: you are obviously using some type of digital tuning (often incorrectly called "room correction"), either in hardware or software, but you are chasing the wrong goals.

Digital EQ is achieved using Dirac Live software. And don’t worry, I’m not in the business of creating a flat frequency response for the sake of vanity

Dirac actually does a good job of tackling some modal issues as well as just the frequency response (I’ve also tried Sonarworks, but Dirac produces far better results in my experience). With the DSP on, the frequency response is within +/- 5dB or better around a 1 metre square area of my listening position (compared to +/- 8dB without it). It's a small area, but workable. Although of course as you say, there are definitely compromises involved in using DSP… and it certainly does boost some frequencies that would perhaps be better left alone (I’ve found these easiest to spot in the distortion plot).

...there is clearly some ringing going on there, at 104, 148, 166, 205 and a few other places. I'm pretty sure those are filters where you are boosting some frequencies too much, into instability.

Those frequencies you note also show ringing when the DSP is switched off, so whilst the EQ might be contributing to the ringing, I don’t think it’s the root cause - and suspect it might be room mode interaction - ?

Room Modes List.png

The second graph above is your Impulse Response graph, and shows the other issue you should be shooting for: smooth decay... based on that, I'd say that you have the impression that the room is not right in the bass: it seems a bit "off". Not as tight as you expected, but something subtle that you can't really put your finger on

Totally agree that the IR curve is far from ideal. The bass actually sounds very good - in fact, the room sounds almost identical to my HD600 headphones (if that’s a good thing!?). But the IR graph is… ugh. And it's something I've always struggled to improve. There’s a lot of 'awkward' furniture in the room, including a large desk and an upright piano, which throws the symmetry off-kilter. So I’ve always assumed that the IR irregularities were caused by dodgy reflections. Happy to take further advice on that.

Here's some graphs from a room I'm working on right now, for one of my clients ("the corner control room").

the corner control room looks fantastic! (both the measurements and the room itself). Always very heartening to see what can be achieved with the application of solid know-how. Obviously my room is a kind of 'retro-fit' (a converted garage) rather than purpose-built from scratch, but plenty of lessons to learn from what you've achieved there. Job very well done.

PS: I’m currently traveling and don’t have all of my REW files on laptop, but I’ve found a measurement from last year that shows L&R speakers (plus sub) with my DSP switched off. Here’s a Dropbox link if you’re interested:

https://www.dropbox.com/s/w4wpn2n8dmi32 ... .mdat?dl=0

DSP off.jpg

Thanks again. Very generous info gratefully received

[Skol303]

PS: don't know if it make much odds, but for the sake of comparison I notice that the waterfall plot from the corner control room is showing 10dB less 'noise floor' than the plot from my own room (down to ~35dB rather than ~25dB). Still looks like a very controlled room

[JasonFoi]

[Soundman2020]

PS: don't know if it make much odds, but for the sake of comparison I notice that the waterfall plot from the corner control room room is showing 10dB less 'noise floor' than the plot from my own room (down to ~35dB rather than ~25dB). Still looks like a very controlled room

Fair point, but there's a reason for that: in the corner control room I am calibrating at about 80 dBC per channel, you are calibrating yours at about 75 dBC. There's no direct comparison (even ignoring the fact that the corner control room goes down an entire octave lower than yours). Also, you have a house curve applied, which raises the low end and cuts the mids and highs. We have not yet applied my house curve to the corner control room, as that would disguise some of the remaining minor issues that we still need to fix. Therefore it's hard to compare levels directly. But across that region of the spectrum, up to 500 Hz (the most important region by far), your level averages about 7 dB lower than the corner control room. Therefore, I had adjusted the scale of the original graphs to give a more fair visual comparison.

But for the sake of "full disclosure", or whatever they call it these days, here you have both waterfalls at the exact same scale:

The corner control room:

FRANK--REW--WF--12-500--R--v3.png

Your room:

Skol303--REW--WF--12-500--R--v3.png

That's both waterfalls shown with a vertical scale of 25 dB to 110 dB, and horizontal from 12 Hz to 500 HZ. You can see that even though your levels are 7 dB lower, your peaks are significantly higher, and longer (time domain).


Here's the two frequency response graphs on the same page, so you can see the level difference that I referred to above:

Skol303-vs-Frank--REW--FR--12-500--7dB.png

So, what I did was to boost your levels by 7 dB to match the corner control room levels, and I arbitrarily chose the 200 Hz point to match your curve against his curve.

Boosting your level is not really valid, acoustically, since in reality if there actually was a higher level of energy in your room it would likely trigger your modes more deeply, causing them to ring longer and louder than is actually apparent here, and perhaps trigger other modes that aren't apparent yet, as well as increasing other issues.... but the discrepancy favors your graph, not the corner control room, so I think it is reasonable to do this. (If I would have done the reverse, and lowered the corner control room level to match yours, that would have been very unfair to the corner control room, since if we really did drop the level in that room by 7 dB, fewer modes would have been triggered, at lower levels, and for shorter periods, so the response would have been even better. So "faking" it by just pulling down the curve, would not show that improvement. Boosting your level is more fair to both of you).

Here's what the two FR curves look like after the boost:

Skol303-vs-Frank--REW--FR--12-500-+7dB.png

Here's your new waterfall, boosted to the same level as the corner control room:

Skol303--REW--WF--12-500--R--v3--+7db.png

And here's a direct overlay of your results on his results. Your data is the yellowish color, the corner control room is the red:

Skol303-vs-Frank--REW--WF--12-500--overlay.png

That's probably about as fair as the comparison could be.

I think the differences are fairly clear: You have several "things" that are ringing rather loud and rather long, significantly higher than the corner control room: as much as 10 to 15 dB higher in places. I'm surprised that you aren't noticing those in your listening tests: I would have expected them to be fairly apparent. Subtle, for sure, but still audible. You have some peaks that are ringing several hundred milliseconds, at levels only about 30 dB or so below the peak level, so that should be audible. For example, at 148 Hz, you have a mode/instability that is down from 81 dB to 49 dB (just 32 dB lower), and that is 300 ms after the impulse. At 400 ms it is 45 dB, and 500 ms it is still 40 dB. It decays very slowly, for around 900ms. I would expect that to be very audible: A ringing bass tone that stretches out for nearly one second, when the overall room RT is 1/10th of a second, should be noticeable!

By comparison, in the corner control room, pretty much everything is -60 dB or lower at 500 ms, and even at 300 ms there is nothing higher than about -50 dB until you get down below 40 Hz or so. Whereas in your room, you have plenty going on at - 30 dB, even three octaves higher than that, and way out to almost one second.

I'm not saying that your room is terrible! Please don't get me wrong! Not at all.... in fact, I'm saying that it's pretty good, better than many home studios, for sure... but it could be better if you could deal with whatever it is that's causing those modal/resonance issues, and also brighten it up a bit, to get closer to the ITU BS.1116-3 spec.

You DID ask for a honest opinion, no holds barred!

- Stuart -

[Skol303]

Hi Jason! Good to see you. I’ve noticed DanDan here too. Nice to be amongst some familiar company

[Skol303]

You DID ask for a honest opinion, no holds barred!

Haha, don’t worry Stuart…! No risk of causing upset whatsoever. This is precisely the kind of constructive feedback I’m looking for.

Fair point, but there's a reason for that: in the corner control room I am calibrating at about 80 dBC per channel, you are calibrating yours at about 75 dBC... Therefore, I had adjusted the scale of the original graphs to give a more fair visual comparison.

Makes perfect sense. And thanks for the comparative frequency response and waterfall plots, very illuminating and nice to see that my room isn’t too far off a purpose-built space like the corner control room, particularly <50Hz where things typically get very gnarly in small rooms (PS: I’m slowly reading through the corner control room build thread at the moment… great job).

You have several "things" that are ringing rather loud and rather long, significantly higher than the corner control room: as much as 10 to 15 dB higher in places. I'm surprised that you aren't noticing those in your listening tests: I would have expected them to be fairly apparent. Subtle, for sure, but still audible.

Perhaps too subtle for my ears! I honestly haven’t noticed any significant low end ringing/smearing during listening tests, but you’re certainly right that there is noticeable ringing and especially between ~100-200Hz. Could it be that these ringing frequencies are just disappearing into the noise floor of the room, which is why I’m not finding them to be audible? Or perhaps I’m long overdue a hearing test?!

Here’s a bunch of layered RTA plots of the room with speakers off, for reference:

RTA.jpg

NB: there is a road a couple of streets over, which probably accounts for some of the low end rumble seen in the plots.

I’ve never been able to make good improvements at those spiky/ringing frequencies, even when I’ve seen gains at other lower frequencies due to treatment. I’ve often wondered whether they might be caused by vibrations in the room itself… for example: I have metal radiator on the wall beneath my speakers and right next to the sub (reminds me I need to remove it as it’s switched off); my desk is made from a kind of hollow board with wooden laminate finish; and there’s an upright piano against one of the walls (themselves made from drywall)… all of which I assume could be possible causes of resonance. But whether such vibrations could be causing the ringing seen in the waterfall plot is something I have no idea about (?).

I'm not saying that your room is terrible! Please don't get me wrong! Not at all.... in fact, I'm saying that it's pretty good, better than many home studios, for sure... but it could be better if you could deal with whatever it is that's causing those modal/resonance issues, and also brighten it up a bit, to get closer to the ITU BS.1116-3 spec.

Thanks and for sure, I’m very happy with the room and think it performs way beyond my initial expectations, given its limitations as being a “brickwork shoebox”. I don’t think I can do much else to tackle those ringing frequencies and may just have to live with them, which is fine as my cloth ears don't notice them! But adding some brightness back into the room by adding slats/ surfaces to some of my existing absorbers is a great idea and something I’m going to experiment with.

Thanks again for your time on this. Solid advice

[Skol303]

Sorry Stuart… pestering you for further advice here; and I've only been on there forum for a day or so!

You mention that my room has:

…some peaks that are ringing several hundred milliseconds, at levels only about 30 dB or so below the peak level, so that should be audible. For example, at 148 Hz, you have a mode/instability that is down from 81 dB to 49 dB (just 32 dB lower), and that is 300 ms after the impulse. At 400 ms it is 45 dB, and 500 ms it is still 40 dB. It decays very slowly, for around 900ms.

I’ve been looking at my waterfall plots and must admit that I tend to only view the top 45dB down from the peak level. My rationale has always been that below 45dB down I’m getting close to the noise floor, which in a small domestic room like mine is perhaps relatively high (in comparison to a better isolated pro-studio).

So for instance, when viewed as per the plot below, the ringing you mention appears fairly inconsequential (mostly disappeared within 300ms)...

Waterfall1.jpg

...but when using a larger Y-axis on the dB and ‘looking further’ towards the noise floor, the ringing obviously appears more pronounced as per this example:

Waterdfall2.jpg

However, does that matter when we're viewing down to 30-40dB - i.e. close to the ambient background level of the room? Or to put it another way, are we viewing 'meaningful' measurements at that level in a domestic room such as mine? I ask, because it seems rare for a domestic/prosumer (i.e. non-purpose built) room to have a workable noise floor much lower than this at low frequencies.

Note that my aim here isn't to sanitise my measurements with favourable viewing settings! I'm not interested in showboating I'm just curious as to whether I should be using a deeper dB axis when assessing my measurements, as I tend to err on the shallower side of waterfalls.

That's it. I'll leave you in peace now!

[Soundman2020]

I’ve been looking at my waterfall plots and must admit that I tend to only view the top 45dB down from the peak level.

Calibration level is 86 dB, and that's a level that is not uncommon for engineers to mix at. If you only consider -45dB from there, that puts you at about 41 dBC. Go into your room, and play contemporary music then turn down the level until you get 41 dBC on your hand-held sound level meter. Can you hear it? I think that might answer your concern...

(Actually, for quiet levels you should play that music at 41 dBA, rather than 41 dBC, since the "A" weighting scale is a closer match to human hearing sensitivity for quite sounds).

My rationale has always been that below 45dB down I’m getting close to the noise floor,

If your noise floor in your room is NC-40, then you have a SERIOUS problem with the isolation of your studio! Most pro studios are designed for NC-15 or so. Home studios should be around NC-20, maybe NC-25 worst case. If your place is NC-40, you must have the window and door open, and a party going on outside... OK, maybe not quite that bad, but you get my point.

This is your actual noise floor, according to REW:

Skol303--REW--Noise-Floor.png

Looks like maybe NC25-ish, maybe. Perhaps less. (It should be measured dBA, not dBC, but that's the way REW shows it.) Your noise floor is at least 50 to 55 dB below your calibration level, which is already 6 B too low, so your actual range is around 60 dB. There's a reason why RT-60 is measure over a range of 60 dB....

In the mid range, you have levels of around 80 dB, and the noise floor is around 25 dB. So there's a range of at least 55 dB there.

Thus, if you only look at the top 45 dB below peak, then you are missing all that stuff happening below that, down near the REAL noise floor. At least 15 dB in your case.

So for instance, when viewed as per the plot below, the ringing you mention appears fairly inconsequential (mostly disappeared within 300ms)...

Well, yes, it is indeed possible to adjust REW to show you anything that you want to see... or that you DON'T want to see! That's why I always keep the ranges adjusted to reasonable limits when I show results from rooms I am testing, and I show them at various resolutions as well, where relevant, so I can't be accused of "doctoring" the graphs to hide stuff. I generally set the vertical dB scale to show 30 to 110, since that's the hearing range that you will typically use in a control room. If you are hitting levels above 110 dBC, you won't be in the mix engineering business very long as your ears will be totally shot, and there's not much interesting below 30 dB, except in high precision rooms. I also usually set the frequency range to 15Hz to 500 Hz, since that is by far the most important part of the spectrum, and the hardest to get right. Pretty much all of your big issues will be under 200 Hz, and some people set their upper limit to 200 Hz for that reason, but in my experience there's often stuff in the range 200-400 that can mess you up, so I like to see that as well. I can always zoom in if I need more detail, but I don't like having important detail off-screen, where I might miss it.

However, does that matter when we're viewing down to 30-40dB - i.e. close to the ambient background level of the room?

Do the test I outlined above: Can you hear music when you play it at 41 dB in your room?

When you are mixing, there will be loud stuff and quiet stuff. If your room is hiding the quiet stuff in some type of flutter echo, SBIR dip, phase cancelled reflection, modal ringing, EQ ringing, comb-filtering, etc. then you won't hear it. And if you can't hear it, then you cant mix it!

This is why pro studios are generally designed so that the LR reverb tails are a lot longer than the CR reverb tails: to ensure that the engineer can actually hear the subtleties of instruments tracked in the LR completely, in order to mix them properly. If the tails of the Live Room are masked by the tails of the CR, then he won't hear them... but the guy listening on his iPhone earbuds certainly will hear them! The RT-60 time of your ear canal is basically null (not quite, but pretty much).

Or to put it another way, are we viewing 'meaningful' measurements at that level in a domestic room such as mine?

The issue is not the type of room: the issue is: can you hear it? If you can't hear what you are supposed to mix, then there's a problem: If you can't hear it, then you can't mix it!!! You might not be able to hear it, but others will, when they listen in different situations. The interior of a car, for example, has very low decay times, so you could potentially hear stuff when listening in a car, that was NOT heard by the engineer when he mixed the song, if the room was masking low-level sounds.

I ask, because it seems rare for a domestic/prosumer (i.e. non-purpose built) room to have a workable noise floor much lower than this at low frequencies.

On the contrary, most home studios have a noise floor that is way below what you think it is! And certainly below NC40. NC40 is the recommended level for factories, shops, restaurants, computer rooms, office halls, corridors, and lobbies, and similar places.

And what about dynamic range? Without getting into the loudness wars, just think about what you are saying here: If you do not want to hear anything that is 45 dB below peak, then you are limiting the dynamic range of your mixes to 45 dB! Nothing can be quieter than 45 dB below your peak, .. so you would be unable to mix a symphony orchestra, which covers 50 dB at least.... Sure, you will most likely compress the hell out of that if you are participating in the loudness wars, and maybe leave 15 or 20 dB in the final mix, but when you are TRACKING it, you WILL need the full dynamic range.... And your own ears have a theoretical dynamic range of over 120 dB... ) In practice, in the real world, more like about 70 or 80... but that's still much higher than you are allowing for.

I'm assuming that your DAW is 24-48 or 24-96, or something like that. Or maybe an older 16-44.1 In other words, 16 bit sample depth, or 24 bit sample depth. If you are running an older 16 bit DAC, then your dynamic range is 93 dB. If you are running 24 bit DAC, then it is more like 115 dB. Why would you need all that head room if you only ever use a tiny fraction of it?

Then there's the rather large issue that the data you are looking at in REW, is steady-state room data, from sine sweeps all played at constant power per octave. Music is nothing at all like that, with percussive transients, swells and dips, harmonics, large differences between frequency bands. There might well be peaks that hit much higher than your calibration level, and the AVERAGE level might well be much lower... The calibration is not the music. Calibration is just to make sure that the room presents a "blank canvas" onto which you can "paint" the music, with confidence that the room is not "coloring", and 86 dBC is equivalent to checking that the canvass looks perfectly white when you have a bright white light shining on it. The canvas will look a whole lot different, once you start painting...

I'm just curious as to whether I should be using a deeper dB axis when assessing my measurements, as I tend to err on the shallower side of waterfalls.

Above, I mentioned the levels that I use initially for looking at the results of REW tests. I use those for tests done with a calibration level of 80 dB for each speaker, thus 86 dB for both speakers, which is "standard" industry practice: that's how cinemas and most studios are calibrated: 86 dB with all speakers running. Of course, if there's a sub involved, then the individual levels would be lower. Ditto for 5.1 71, etc. But the point is to get 86 dB at the listening position for the calibration. You are running yours about 6 dB too low. Set your graph axis limits to reasonable values that represent what you can actually hear in the room, for the calibration test signals.

If you are using levels 6 dB lower to start with, then you should probably set your vertical axis scale to something like 25-105 dB

Yeah, I know that there will be some people that don't agree with these ranges, and will try to tell you that you should only look at certain part of the data, that there's "nothing to see" in other parts... but my clients tell me different. If they can hear it in their room, then I want to be able to see it in the REW data. so I can fix it. So set your limits realistically, to see EVERYTHING that the room is doing, then zoom in if you see something of interest, or zoom out to get the "big picture". And also beware of using the "1/xx octave smoothing" incorrectly: It's very easy to miss important detail in the low end if you set the smoothing to high, and just as easy to chase your tail around a problem in the high end that doesn't even exist in reality, if you set the smoothing too LOW.

One day I might write a tutorial on using realistic REW settings...


- Stuart -

[Skol303]

Thanks Stuart! Another very detailed and useful reply

Absolutely agree that anything at 40dB (or less) is audible I suppose what I was getting at is: should we be concerned about decay below 30-40dB? My questioning was partly based on comments I’ve read from John Mulcahy on the topic, such as:

“The main use for the waterfall is to look at the relative rates of decay across the frequency span. That decay starts right at the top of the curve, an area where decay is slow will stand out increasingly in the later slices of the waterfall. Features that only emerge very late in the decay are more usually external noise than relevant parts of the response.”

Regardless, from your explanation the answer to “should we be concerned about decay below 30-40dB?” is clearly yes, we should! Noted.

On that basis and just for reference, here’s an updated copy of my waterfall plot spanning 25-105dB as you suggest (with DSP engaged)…

With DSP.jpg

…and just for comparison, here’s another plot with the DSP switched off. I’ve noted some additional ringing between ~100-200Hz (probably caused by DSP boosts as you suggested), but also some very narrow ringing that appears on both plots. My suspicion has always been either strong modes at those frequencies, or resonance from something in the room. But I'm just guessing.

Without DSP.jpg

Calibration level is 86 dB, and that's a level that is not uncommon for engineers to mix at.

My monitors are calibrated to 76dB at K14, which I understand is the recommended level for a room of this size. And out of habit I always take REW measurements at that same level, which by coincidence is roughly the same level recommended by John Mulcahy for REW measurement sweeps (he suggests 75dB). Interesting discussion on that here if you're interested - especially the concluding post by DanDan (of this parish!):

https://www.gearslutz.com/board/studio- ... t-rew.html

Thanks again.

[Soundman2020]

Interesting discussion on that here if you're interested

Interesting, yes, but not relevant to what I'm talking about here. Apples and bananas.

That discussion is about signal to noise ratios, and how it is possible for REW to pick out the signal very well, even with a fairly high noise floor. Absolute true, but that has nothing at all to do with the issue here, which is ROOM response!

This is very simple: In the REW data, you want to see how the room will respond under the conditions where you will normally be listening. And since most engineers listen in the 80 - 90 dBC range, then they boost to maybe 10 or (gasp!) 20 dB more occasionally to "check the low end", then that's the level you want in your room! Simple and logical. If you measure at a significantly lower level (such as 70 or 75 dBC), then you cannot be sure that all of the modes have been triggered, or that the ones that ARE triggers have been triggered to the same level as they will be under normal conditions. 70 dB is one hundred times less energy than 90 dB, so one hundred times less likely to trigger all the modes fully. 75 dB is ten times less energy than 85 dB. You might well be triggering all the modes fully at 70 dB (or 75 dB), but you'll never know unless you also check at a much higher level.

Simple case: Imagine that you are mixing a song with a bass guitar solo in it. The bass is obviously going to be the ONLY thing putting out sound at that point. If the level you are listening at is 85 dBC (quite common) then pretty much all of what you hear is pure bass tones at 85 dB. Bass guitar is not far off from simple sine wave: exactly what REW produces to analyze the room! You want to be sure that the room is NOT messing with that bass, by perhaps adding it's own coloration: In other words, no modal stuff going on, no phase cancellations at the mix position, etc. And the only way you can do that, reliably and realistically, is to run REW at 85 dBC, to ensure that REW is seeing the room the same way that you will "see" it when that bass solo comes around. If you run REW at 70 dB, then you are only putting out less than 1/10th the energy level that would be there at 86 dBC, so you are not testing the same thing at all.

In fact, if you think there might be something strange going on in the bass while you are mixing, and you want to check that, it would be very common to temporarily boost the level up to eleven (!), so you can more clearly hear it. In that case, you also want to be certain that what you are hearing is the bass guitar itself, not what the room is adding to the bass, or taking away from the bass. And that has nothing at all to do with S/N ratios. IT is purely about the sound power in the room, and what the room is doing in response to that sound power. If you whisper in the room, you will not trigger any modes at all, not even slightly. If you scream, you might trigger some in the mid range. But if you blast out a bass-heavy song at 86 dBC, you will most definitely excite every single mode that can possible be excited. That's what you want REW to show you.

which I understand is the recommended level for a room of this size.

The size of the room is also irrelevant: What matters is how loud the speakers will be in a typical intense mixing session. You might well assume that you'll be mixing at 70 dBC, but that won't last for long: sooner or later you'll need to push that up, so you can hear what is going on more clearly.

“The main use for the waterfall is to look at the relative rates of decay across the frequency span.

True. But if the test signal was not loud enough to even trigger the mode, how will you be able to look at it's peak and decay rate?

On that basis and just for reference, here’s an updated copy of my waterfall plot spanning 25-105dB as you suggest (with DSP engaged)…

And from that, you can clearly see how the filters are causing some pretty impressive ringing at around 50 Hz, 65 Hz, 80 Hz, and 100 Hz, plus making the existing ringing WORSE at several points in the range 150 Hz to 200 Hz. If that were my room, I would deal with those issues by fixing those filters to REDUCE the ringing (certainly not allow them to make it worse), even if it means messing up the frequency response a little. Ruler-flat FR is nice for bragging rights, yes, but it's not representative of how your ears actually perceive sound: time domain response is more important, especially in the low end. Take a look at some of Floyd Toole's papers on that.

- Stuart -

[Skol303]

Thanks Stuart, very useful food for thought.

I'm fairly strict with the whole K-System approach and leave my monitor controller set to 76dB at K14 (measured C-weighted/slow). In fact I don't ever touch my monitor gain control when mixing; I just keep an eye on my desk's analog meters and push the faders until everything sounds "loud enough". So in my case I guess taking REW measurements at around 76dB per channel is appropriate, because that's my maximum listening level (rightly or wrongly... Fletcher-Munson curves, etc, etc).

Really good point about the DSP filters and something I'll look into tweaking further. I've always lived with the trade-off of some additional ringing vs flat frequency response, as the extra ringing never really looked - or sounded - problematic to my eyes/ears. But now you mention it I'll re-investigate and perhaps slacken off some of the DSP filters at certain frequencies (Dirac Live thankfully allows for very detailed adjustment of the target curves).

In the past I have use much more gentle DSP settings, which led to a more uneven low end frequency response that was certainly audible - i.e. I could clearly hear the dips when running a stepped sine test between 20-200Hz. As it is now, the low end response sounds very even in spite of the additional ringing - and I considered that to be the better compromise in terms of mixing (as you say, "you can't mix what you can't hear"). So for sure, I need to make any further adjustments using my ears as much as my eyes But I'll do some experimenting and see if I can find a better balance between freq response / decay.

Thanks again for taking the time to respond with detail. You've helped to shine a light on some issues that will allow me to improve my room further - precisely the purpose of this thread. Cheers!

[Skol303]

Greetings all. A quick update on this thread…

I’ve been experimenting to try and reduce some of the ringing in my room around 100-200Hz - i.e. the narrow, ‘high Q’ decay seen in my waterfall plots. After some messing around with my DSP software, I surmised that the likely cause wasn’t in fact the DSP but resonance in the room itself (one likely culprit being a metal radiator at the back of my desk, adjacent to my sub/mains). So I installed some additional fibre-based traps in the only space left untreated…! Which involved stuffing some cotton laundry bags with fluffy insulation and placing them around the sides and back of my desk.

Here’s a waterfall plot before the extra treatment with DSP:

BEFORE.jpg

Here’s a comparative plot from after (again with DSP):

AFTER.jpg

And here's a comparative before/after frequency plot for reference:

Freq Response.jpg

Observations:

The narrow ringing between 100-200Hz has been reduced, but there appears to be increased ringing at 30Hz (where I have a room mode). Curious; I'd have expected this low-end ringing to have decreased with the addition of more fluffy fibre.

The notch in the frequency response at 100Hz has deepened (by ~1dB to -3dB, so still fairly insignificant); but the previous dip at 200Hz has disappeared (an improvement of around +2.5dB here).

Questions:

1) Any idea what might be causing the increase in low-end ringing at 30Hz (with the addition of more fluffy fibre)?

2) Which result do you think is preferable - before or after? My own opinion is that the 'after' result is better; that the increase in low end decay (30Hz) and notch at 100Hz (down by -1dB) is worth the compromise for a slightly ‘cleaner’ result between 100-200Hz in terms of both decay and frequency response. But I’m perhaps biased by wanting to see an improvement in the results! So as always, objective options very welcome.

Thanks guys.

PS: as an aside, I'm starting to think that the 50Hz ringing is mains hum (or a ground loop). I'm in the UK so 50Hz would be the telltale frequency for mains; and it's also very prominent in the results from my little Mixcube speakers which roll off steeply below around 200Hz. Example here:

Mixcubes.jpg

[Gregwor]

If you run another test today, is that 30 Hz issue still showing up? What I'm getting at is maybe a truck drove by or something that caused it during this measurement.

Greg