John Sayers' Recording Studio Design Forum Archive

Hey there!

I posted my first post a while back asking for some tips on my potential new control room, well I am glad to say its now done! Just a couple of diffusion panels for the back of the room but I cant imagine them making too much more of a difference as I'm usually mixing at appropriate levels. Even so, I have little diffusion in here and will be installing them soon! (They are pallets haha)

Along with some decoration thats left to go I am all good. I'll post some pics whens its 100%! SO out of curiosity, I have been getting to grips with REW software and its pretty mind boggling at first. Actually it still is pretty mind boggling. My FR charts look to be decent enough, but as i read in other posts, its the least useful chart! Sadly i dont know what the hell is going on or what i am supposed to be looking for with the waterfall and spectrogram stuff so im hoping someone will be able to tell me if my room looks to have reasonable acoustics. I spent a lot of time researching and designing the room so hopefully its not too bad!


I have no idea what at least half of this stuff is haha

I will also attach the rew file if anyone wants to look further. I did the soundcard calibration at first but for some reason it was picking up a massive boost at around 10k, which it then compensated for but it really didnt look right to me so i removed the calibration file.

I'm using a Roland Studio Capture and used a Superlux ECM 999 mic to do these tests. Im hoping I dont have to do any more acoustic treatment

There are the images there of LEFT/RIGHT SPL and PHASE, these are without smoothing....the ones with 1/48 smoothing are also there

So, how does my room sound? Did i miss something? Do i need to do something differently in the REW?

EDIT: I tried to upload the REW file in a zipped folder but its too large to attach here. Images look a bit fuzzy as i resized them as per forum rules

Cheers in advance for any feedback !

I would need the actual MDAT file. The graphs you posted are pretty useless, since they are all zoomed out to the maximum, with no detail at all visible. Please upload the MDAT file to a file sharing service, such as Dropbox, and post the link here.

My FR charts look to be decent enough,

Actually, it doesn't! It only looks good because you are zoomed all the way out. If you look from the moon, then the Cantabrian Mountains and the Pyrenees look dead flat and perfectly smooth too, but when you stand right next to them, you see that isn't actually the case at all...

In the FR window, click on the "Limits" button, and set the limits like this: Left=18, Right = 500, Top = 100 and 720, Bottom = 35 and -720. Then click "Apply Settings". That's what your frequency response really looks like. Update the views on all your other graphs to something similar, and you'll see that things are not nearly as good as you think they are.

From what little I can see on your graphs, you clearly have a modal issue at around 40 Hz, and the usual SBIR null that is typical of many small rooms at around 85 Hz, plus some other serious issues between around 100 Hz and 200 Hz. There's a lack of bass, as though you maybe tried to compensate for the modal issues by rolling it off a bit: it seems to dip about 5 dB below 120 Hz, all the way down to the huge modal beast at 40 Hz. There also seems to be some type of surge in levels between about 4 and 10 kHz, which is hard to explain. Or maybe it's a dip around 1 to 3 kHz.... hard to say, with so little to go on. The overall decay times seem rather low as well. Or maybe it's just a very small room that is over-treated.

Now, take all of that with a pinch of salt, because I'm trying to interpolate the zoomed-out results from experience alone, without any real detail to work from.

Sadly i dont know what the hell is going on or what i am supposed to be looking for with the waterfall and spectrogram stuff...

On the page at the link below, there's a set of graphs that shows what you can expect from a good room:

http://www.johnlsayers.com/phpBB2/viewt ... =2&t=20471

Scroll down a bit, and you'll see the full set of graphs from that room, posted by the owner, with explanations. There are both waterfall and spectrogram graphs there, showing what your room should be like. The graphs on that page are fairly good already, but in fact, after further treating and tuning, the final graphs were quite a bit better than what you see there. But your graphs should resemble those.

I spent a lot of time researching and designing the room so hopefully its not too bad!

Please post photos of the actual room interior, and details of the treatment that you installed.

Just a couple of diffusion panels for the back of the room but I cant imagine them making too much more of a difference as I'm usually mixing at appropriate levels.

I'm not sure I understand that comment at all. Diffusers have nothing at all to do with monitoring levels. There's not even the slightest connection. A diffuser will work well across the complete range of typical monitoring levels, from very quiet to very loud. Also, most types of diffusion are not an option in small rooms: For diffusers based on numeric sequences, the room must be large enough that there are no listening locations within about 10 feet (3m) or any diffuser. How big is your room?

Even so, I have little diffusion in here and will be installing them soon! (They are pallets

They then aren't proper diffusers! Diffusers are tuned to specific ranges and placed at specific locations to deal with specific issues. Bits of a pallet hung on the wall does not constitute a tuned device, and won't deal with the issues,

I did the soundcard calibration at first but for some reason it was picking up a massive boost at around 10k, which it then compensated for but it really didnt look right to me so i removed the calibration file

So you discarded the most important step? And now you have no idea if the results are even valid? That's the entire purpose of the calibration process: to compensate for issues with the signal chain. If you did not calibrate REW correctly, then the results are suspect. Did you notice that the frequency response curve for the ECM-999 does indeed have a peak at around 10 kHz, with a deep roll-off above that? Why did you think that your "eyeball guesstimate" is more accurate than the carefully controlled precision digital calibration process?

There's a number of questions that we'll need answers to, in order to interpret the data correctly:

1) How did you have the mic set up? (location and angle)
2) What pre-amp (if any) did you use, and how did you have the controls set on that?
3) What console (if any) did you use, and how did you have the controls set on that?
4) What interface (sound card) did you use, and how did you have the controls set on that?
5) What speakers did you use, and how did you have the controls set on those?
6) If they are passive speakers, what amp did you use, and how did you have the controls set on that?
7) What are the exact dimensions of the room? From what I can see on the graphs, it's a small room: Maybe 5m x 3m x 2.5m, at a very rough guess?

So please post the MDAT file some place where I can download it, and also provide the answers to those questions, as well as photos of the room as it is right now. Also post the calibration file that you discarded.

- Stuart -

Stuart! Thanks so much, I will answer those questions and post pictures of the results with the proper zooming, the mic placement and the room when i get a chance to do that, hopefully tomorrow. For now I have uploaded the mdat and cal files to dropbox, this is the link:

https://www.dropbox.com/sh/o0ip9pdrnaoq ... LuVLa?dl=0

Not a bad guess on the room size, its pretty much that. Technically the space im working in is approx 4.9m x 2.75m, height 2.8m but I will explain more with the pics tomorrow. As for what i said about having researched and designed the room, i did mean to say that i did it on a tight budget haha....what im hoping for is to have a room that sounds better than it should....i have virtually no budget left for the time being, so I just need to make sure its somewhat reasonable...... please do let me know what you think after taking a look at the REW files, i will be back on tomorrow to provide all the other information

Great! I downloaded the MDAT file and also the calibration file. So here's a brief analysis.

First, there's no data from both speakers at once, so I can't say at all how the room reacts to that. There are 3 measurements in the file you posted: "Left speaker only", "Right speaker only", and "average" which is no use at all. The average of the left and right tells you nothing at all about how the room behaves to having both speakers running. Think of it this way: If you put your head in the freezer and your feet in the oven then your average body temperature is just fine! But that tells you nothing at all about the state of health of your body... Averages aren't a whole lot of use when you are talking about speakers in a room.

Next, you did the tests at a level that is about 10 dB too low! You did the tests are around 70 dB, but we need them done ten time louder, at exactly 80 dB for each individual speaker. If you do that correctly, then the test with both speakers on should automatically come out at exactly 86 dB, which is the "standard" calibration level for studios and cinemas.

Third, your calibration file does not look at all like a calibration file! Please re-do that, carefully following the instructions in REW. Are you SURE you had the loop-back connection set up correctly? It almost looks like you used the mic and speaker, instead of the loop-back...

So please do another set of tests, this time with the sound card correctly calibrated, REW carefully calibrated to 86 dB, and each speaker carefully calibrated at 80 dB. Do a "L" test, and an "R" test, and also do a third test with both speakers on, WITHOUT ADJUSTING ANYTHING! After you calibrate, don't touch any controls, and don't touch the mic. Just do one "L" test, one "R" test, and one "LR" test. For each of those three tests, you must leave the room before it starts running: Use the "Start delay" option to give you enough time to get out of the room, plus another 5 seconds or so to allow for things to settle down in the room after you are gone.

But based on these initial tests that you sent:

There's a big difference between the left and right channels. The room is not symmetrical, or the speakers are wildly different from each other, or they are not set up symmetrically, or the treatment is not symmetrical, or some combination of the above.
That's a graph showing the frequency response of the left (blue) and right (red) speakers, for the low end (below 50 Hz). There are differences of 5 to 10 dB between the two channels. It¿s not just the low end, either. There are significant differences all the way up to about 5 kHz. That needs to be fixed.

Next, the low end is rolling off too steeply:
Below about 120 Hz there's a drop of well over 10 dB down to 45 Hz, before the massive modal problem at 40 Hz hides it. That should be flat (see the graphs at the link I posted), and some acousticians even prefer to have a slight rise at around 80 dB.

What speakers are you using, and how are the controls set on those? Where are the speakers set up in relation to the front and side walls, and to the mix position? Where is the mix position in relation to the front and rear walls?

There's also a massive dip at around 418 Hz on both channels: Not sure that that it is, but I suspect a large reflection of some type.

Next, overall decay:
The room is slightly too dry in the top end: the overall decay time is around 170 ms, and it should be around 200 ms for that room, but that's acceptable. It's also fairly smooth...down to 500 Hz. But below that, it's all over the place!

The reason for that is clear from the waterfall plot for the low end:

There's clear modal stuff at around 40 Hz, which is surprising as it does not match the dimensions you mentioned. There's also modal stuff at 80 Hz and 120 Hz, all of which are related to your 0.0.1 axis (front- back direction), but wold only make sense in a room that is 60 cm shorter than yours: that would fit a room about 4.3 m long, but not one 4.9 m long. Strange!

There's also signs of other modal issues in there, but not too severe.

Next up: Sectrogram:
That is VERY revealing! The low end is not smooth at all. There's a huge hole at 85 Hz, the modal peaks at 40 Hz, 120 Hz, and 142 Hz are very evident, as is the general lack of energy down there.

Finally, the actual impulse response: There's a major reflection coming in at about 34 ms: Actually, it's several related reflections coming in close proximity. That works out to a round-trip distance of roughly 10m, so my guess is those are all coming off the back wall. And that also implies that the back wall has little to no absorption on it.

There's also a couple of even larger reflection coming in at around 1ms and 3.4ms, which works out to about 30cm and 105cm difference in path length, respectively, so I would suspect a bounce off the desk or console for the first one, and a bounce off the floor or ceiling for the second one. There's other stuff going on in there too, but those are the big obvious ones.

Overall conclusion: There's nowhere near enough treatment on the rear wall, especially for low frequencies: The mids and highs seem reasonably good, but there's an overall lack of flatness and smoothness in the frequency response. It might be poor quality speakers, or it might be EQ not set correctly in the signal path, or a faulty mic, or bad speaker placement, or bad room treatment. Or a combination.

But anyway, please carefully re-calibrate REW and your system, then repeat the tests with the set of three, and make sure you are not in the room.


- Stuart -

Thanks for the feedback. I will get on getting those more accurate results. I did struggle to get the calibration file to look right to be honest, i spent pretty much a whole day tinkering trying to get it to work properly but was unable to. I'll give it another crack I'm using a Roland Studio Capture, have turned down the direct monitoring feature etc....using the headphone out into insert 3....would i need to be using a line out into say a hi-z input? Using Tannoy Reveal 501 Active monitors, Volume on them is at 50% and the high frequency cut/boost switch is at 0db.

Forgive me for such bad results....thats two days of my life i wont be getting back haha...

I positioned the mic pointing upwards where my head would usually be, at ear height......is that right? or should i be pointing them at the speakers directly? or to the centre point between them?

Looks like i managed to get the calibration to work!

I did in through the Line in and also again through a mic in. The REW website says to use the input that i will be using for the mic, so i assume the mic in is the file that applies here, even though they specify using a line in... anyway here they are:

https://www.dropbox.com/sh/34vfxkjsl9mj ... tsdla?dl=0

Ive got some bands rehearsing here at the moment so ill start setting up and all that....get some testing going on this afternoon

Also yes i realised the mic itself isnt entirely flat, i was unable to find a cal file for the mic though... how can i compensate for its FR? Also should a be using the foam cover for the mic?

As in....i used it like this originally
Should i used the windshield?
Also....im trying to get the levels to 80db each as you said.....the rew page says 75db and im not sure if im doing this right but is this what im supposed to be adjusting here>

I did struggle to get the calibration file to look right to be honest, i spent pretty much a whole day tinkering trying to get it to work properly but was unable to.

It's not that hard to do! All you need is to connect a cable from the output connector of your sound card (where your speaker is normally connected) directly back to the input connector (where your mic is normally connected). That's it! You will need to reduce the output level (or input gain) to prevent clipping, but it's not more complicated than that. There's nothing you need to to do "make the curve look right". It's just a simple loopback connection.

The REW website says to use the input that i will be using for the mic, so i assume the mic in is the file that applies here, even though they specify using a line in

I'm not sure what you mean by that. You use the exact same connectors as you would be using for your mic and speaker. You just attenuate the levels. That's all. The "mic in" file you sent shows valid data, roughly what is expected for a typical sound card, so it appears you did it right, but please confirm that's how you had it cabled

Also should a be using the foam cover for the mic?

Yes.

im trying to get the levels to 80db each as you said.....the rew page says 75db

REW was originally designed for home theaters and audiopihile setup. I have no idea why they chose 75 dB as their calibration level, but the industry standard for studios and cinemas is 86 dB. That's the level taht you want with all the speakers turned on and producing full-spectrum pink noise. So if you have two speakers, then each of them must be set up by itself to produce 80 dB, all on its own. If that is done correctly, then when you test bother together, you will automatically be getting 86 dB, since two speakers produce exactly twice the power of one single speaker, which is, of course, exactly 6 dB louder, SPL.

im not sure if im doing this right but is this what im supposed to be adjusting here

You are in the correct place, yes, but your meter is showing "OVER", which means your input gain is too high, and your sound-card is clipping. So you will have to set your gain structure correctly on the input signal chain.

Make absolutely certain that you have the gain structure set correctly, starting at the mic and following the signal chain all the way through. If you have your gain structure set correctly, and you are seeing 75 dBC on your hand held meter, then there is no way that you can be getting "Overs" on the input. That's impossible. Even at 80 dBC for each individual speaker, that's still extremely unlikely. Your system should be able to handle over 100 dBC SPL, so there's no way it should be clipping at 75.

Of course, I'm assuming that when you saw the reading that you show in the photo on the REW meter, you were also seeing 75 dB on your hand-held sound level meter, at the same time? Is that correct? And you have your hand-held sound level meter set to "C" weighting and "Slow" response? And you are holding it close to where the measurement mic is, in the room? In other words, the mic on your hand-held meter and the measurement mic are very close together, almost in the same place, but not actually touching? What brand and model of sound level meter are you using?

Also, did you set REW to use the correct signal when you did that setup? There are three possible signals: you must select the option that says "Use REW speaker cal signal". Is that the one that you used?


- Stuart -

Trying to get the speakers calibrated to 80db, is this what im aiming for? I used a sound meter on my phone to see if i was getting it right, a ball park figure
It gave me higher readings than REW, I know the app is probably not as accurate as the REW software so...am i good to go? Should i get testing? Also wanted to ask. What kind of sweep settings should i be using ie
One more question; I'm not sure I can use both speakers at once for testing....at least not with the asio driver. Do i need to use the java drivers for that?


Also
I realise i have answered most of those questions you asked but to round up:

1) How did you have the mic set up? (location and angle)

Pointing directly up, placed around where my head would usually be, at ear/tweeter height

2) What pre-amp (if any) did you use, and how did you have the controls set on that?

Roland Studio Capture Interface Preamp, I turned the gain on the preamp and the volume of the interface up til they were matching up with the level checker in REW and to get 80db readings ended up with mic gain at 100%, and monitor volume knob at 50%. While doing this i noticed my right speaker was around 3dbs higher than the left one at the test frequency anyway, so i adjusted the speaker volume knob slightly to match it as closely as possible to 80db

3) What console (if any) did you use, and how did you have the controls set on that?

None

4) What interface (sound card) did you use, and how did you have the controls set on that?

Roland Studio Capture. 44.1khz, 24bit. Direct monitoring off. Output on the software mixer set to -6.5db

5) What speakers did you use, and how did you have the controls set on those?

Tannoy Reveal 501 Actives, Volume knobs at approx 50%, with the HF adjust switch at neutral.

6) If they are passive speakers, what amp did you use, and how did you have the controls set on that?

n/a

7) What are the exact dimensions of the room? From what I can see on the graphs, it's a small room: Maybe 5m x 3m x 2.5m, at a very rough guess?

4.9m x 2.75m x height 2.8m

If everything is looking good let me know and i can get the testing/more tweaking done on Monday

It gave me higher readings than REW, I know the app is probably not as accurate as the REW software so...am i good to go?

No, because you are misunderstanding the REW calibration process! You do not calibrate your hand-held meter to REW! It's the other way around: you calibrate REW to your hand-held meter!

REW has no idea at all how loud things are in your room- Nothing. It cannot possibly know... unless you tell it. All REW knows is that it is getting a digital signal, but it has no idea how that relates to the actual sound level in the room. YOU have to TELL it what the real level is. That's the entire purpose of calibrating! REW needs to know, and you have to tell it. So that's why you START by setting up your hand-held meter in the room, at the mic position, then set up your sound system, USING YOUR HAND-HELD SOUND LEVEL METER (NOT REW), such that EACH speaker is producing exactly 80 dB when REW is playing its test signal. When you have your sound system set up like that, and you are seeing 80 dB on your hand-held meter, THEN you tell REW what that level is.

Your levels are clearly much too high. You are seeing clipping at 88 dB, which is no use at all.

Once again, before you do any of the above you MUST set all your gain structure correctly. The fact that your sound system is clipping already at a level of 88 dB is proof positive that you do not have your gains set up correctly.

So one more time:

1) Set up your gains correctly on all of your gear (mic pre amp, console, sound card, software, etc.) -20 dBFS for digital, 0 dBu for analog.
2) Set up each of your speakers to produce 80 dBC at the mix position, using the REW test signal going out through your sound system with all the gains and faders set to 0 dBu (0r -20 dBFS), and adjust the physical speaker volume control knob (or amplifier volume control, if you are using passive speakers), until you see 80 dBC ON YOUR HAND HELD METER.
3) Calibrate REW.

It's that simple. But it won't work if you don't set your gain structure properly.

(mic) Pointing directly up, placed around where my head would usually be, at ear/tweeter height

I prefer to set the mic pointing up at 60°, not 90° as you have it. Many people do say 90°, and there's nothing wrong with that, but I have my reasons for preferring 60°.

Also, ear height and tweeter height are two different things. Ear height should be the height of the acoustic axis of the speaker, which is usually not the same as tweeter height. It's normally close: within an inch or two, but not necessarily the same. The speaker should be set up with the acoustic axis at 120cm above the floor, which is the "standard" height, or maybe a bit higher.

Roland Studio Capture Interface Preamp, I turned the gain on the preamp and the volume of the interface up til they were matching up with the level checker in REW and to get 80db readings ended up with mic gain at 100%,

Then you did it backwards! That's why you are getting clipping. FIRST you get the speaker to produce exactly 80 dbC MEASURED ON YOUR HAND HELD METER, NOT ON REW. THEN you set your mic pre-amp gain such that that incoming level produces a level of -20 dBFS MEASURED ON THE ROLAND INTERFACE ITSELF. Only then do you tell REW what the real level is in the room.

and monitor volume knob at 50%.

Once again, that's backwards. You set the speaker level (volume control) so that when you feed a 0 dBu signal into it, it produces 80 dBC in the room. You do not use the speaker controls to set the mic input level...


- Stuart -

I have been tweaking the setup as per your recommendations.

My mic input set at 38 db gain on the pre is giving me -20dbfs on my interface. Did you mean -20 within REW? (average not peak)
I used my SPL app to calibrate each speaker to 80dbs (yep, ive been using speaker cal and not sub cal sound), placing my phone very close to the superlux
90 degree angle (for now, ill do multiple tests once im sure ive got everything set up correctly)
Volume knob on interface is now at around 40%
I just measured the height of my monitors from the floor, 120cm gets me just under the tweeter....that cool?

My main concerns are that the level checking seems to be off....as you can see in this screenshot
I have my interface output level at -6.5 in the interfaces software mixer purely because i get clipping through my DAW otherwise. I could turn it down within my DAW but just happened to do it this way. Is that an issue?

If i turn the mic gain or the speakers up to match input and output levels it means going over -20dbfs.... im slightly confused about how to get these levels to match now....or if i need to? Is "C" the correct selection on the calibration window?

Also, which sweep settings are best to use? longer or shorter, how many passes?

I can't seem to set it up for sound to come out of both speakers at the same time....the only way i managed was switching to the Java driver, but when measured it gave me an overall lower output on my spl app of around 75db, so i assume the drivers have something to do with that, surely meaning it wouldnt be comparable? i played with the phase invert button that didnt change anything either....

Thanks again Stuart, i feel much much closer to getting this all set up right and tested

Hey Stuart, just wanted to leave a little bump here, am really eager to assess the situation at the studio

Cheers

My mic input set at 38 db gain on the pre is giving me -20dbfs on my interface.

Assuming that the meters on your interface are set to read digital, then that's correct.

placing my phone very close to the superlux 90 degree angle

Phone??? You can't use a cell phone app for this!!! As I said from the start, you need a proper hand-held sound level meter. A cell phone does not have a precision, calibrated omni-directional mic with flat response! It does not have a calibrated pre-amp and high quality electronics...

I just measured the height of my monitors from the floor, 120cm gets me just under the tweeter....that cool?

That's probably about right, yes.

My main concerns are that the level checking seems to be off....as you can see in this screenshot

Why? I don't see what is wrong there.

I have my interface output level at -6.5 in the interfaces software mixer purely because i get clipping through my DAW otherwise. I could turn it down within my DAW but just happened to do it this way. Is that an issue?

As long as it is not close to clipping anywhere, and you are getting good headroom, then that's fine.

If i turn the mic gain or the speakers up to match input and output levels

I don't understand: Why would you want to match the input and output levels? For what reason?

im slightly confused about how to get these levels to match now

I'm slightly confused about why you would want to do that!!! ?

Is "C" the correct selection on the calibration window?

YES! You need BOTH your sound level meter AND ALSO REW set to "C" and "SLOW". I'm pretty sure I already mentioned this a couple of times. ALL of your tests must be done using ONLY "C" weighting.

Also, which sweep settings are best to use? longer or shorter, how many passes?

Set to: Start=18, End=22,000, level=whatever you have it at now, Length=256k, Sweeps=4

I can't seem to set it up for sound to come out of both speakers at the same time...

???? Huh? That makes no sense. You DO have your system wired for stereo, correct? You do have two channels of output from your interface going to two channels of input on the console, and those two channels are panned fully left and fully right, correct? If you can't get stereo sound out like that, then you have an even bigger problem somewhere....

i played with the phase invert button that didnt change anything either...

It actually changed a lot, but you won't hear any change if you are only listening to one speaker! But DON'T play with phase inversion. Make sure that BOTH channels have no phase inversion applied.

- Stuart -