• Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Room Measurement Tutorial for Dummies Part 1

When measurement range starts below 20Hz, REW sweep always includes linear sweep from DC to 10Hz. This is a documented feature, but why does it do that? Is there a way to avoid this?
@JohnPM
 
When measurement range starts below 20Hz, REW sweep always includes linear sweep from DC to 10Hz. This is a documented feature, but why does it do that? Is there a way to avoid this?
REW's sweep starts at half the start frequency, so a measurement from 20 Hz starts its sweep from 10 Hz. As sweeps follow an exponential frequency profile the sweep spends equal time in each octave, maintaining an exponential (aka log) sweep with a start lower than 10 Hz would spend excessive time at very low frequencies, so REW substitutes a linear portion to reach 10 Hz then exponential thereafter, ensuring the sweep has sufficient low frequency energy without devoting too much of its total energy to that range.
 
REW's sweep starts at half the start frequency, so a measurement from 20 Hz starts its sweep from 10 Hz. As sweeps follow an exponential frequency profile the sweep spends equal time in each octave, maintaining an exponential (aka log) sweep with a start lower than 10 Hz would spend excessive time at very low frequencies, so REW substitutes a linear portion to reach 10 Hz then exponential thereafter, ensuring the sweep has sufficient low frequency energy without devoting too much of its total energy to that range.
But if I select 15Hz as the start frequency, REW sweep seems to start from DC instead of 7.5Hz. Is that correct?
This near DC part seems to trigger some amplifiers protection circuit at the start. Not in my case but a fellow user reported this on his system.
 
Hi,

I'm starting to explore using REW, but I'm not sure if my current setup will even allow me to do a room measurment. First of all, I only have a basic Surface Pro laptop. I also have an old Omnimic V2 microphone that I can connect to the Surface Pro via USB.

I don't have the ability (no HDMI, audio outs, etc.) to connect the Surface Pro to my stereo system in order to play the measurement sweeps back through the speakers. I'm using Logitech Media Server (now Lyrion) and an RPI4 as a streamer connected via USB to an RME ADI-2 dac which is, in turn, connected to a Benchmark AHB-2 amplifier via XLRs.

My question is: Can I save/export the REW sweep file and place it on the LMS/Lyrion server, play it back through my speakers and then, with the Omnimic plugged into the Surface Pro have the ability for REW to do a room measurement??

Not sure if I've provided all the info needed to see if this is possible, so if any additional information is needed, please ask. Help is much appreciated!

Rick
 
Hi,

I'm starting to explore using REW, but I'm not sure if my current setup will even allow me to do a room measurment. First of all, I only have a basic Surface Pro laptop. I also have an old Omnimic V2 microphone that I can connect to the Surface Pro via USB.

I don't have the ability (no HDMI, audio outs, etc.) to connect the Surface Pro to my stereo system in order to play the measurement sweeps back through the speakers. I'm using Logitech Media Server (now Lyrion) and an RPI4 as a streamer connected via USB to an RME ADI-2 dac which is, in turn, connected to a Benchmark AHB-2 amplifier via XLRs.

My question is: Can I save/export the REW sweep file and place it on the LMS/Lyrion server, play it back through my speakers and then, with the Omnimic plugged into the Surface Pro have the ability for REW to do a room measurement??

Not sure if I've provided all the info needed to see if this is possible, so if any additional information is needed, please ask. Help is much appreciated!

Rick

In principle, what you want is an "offline measurement" - https://www.roomeqwizard.com/help/help_en-GB/html/offlinemeasurements.html
 
Ah, thanks. So, if I understand correctly, for the offline measurment procedure, I would first generate a sweep file from the REW signal generator, save it, and then copy the saved file to LMS/Lyrion. Then, with the mic hooked up and REW loaded, play the sweep and save the result as a response file (?). Then I would import the original sweep file as the stimulus, import the saved response file as the "response" and proceed from there?

Is there any way to do this live---in other words just play the sweep file that I saved to LMS/Lyrion and have the room response measurment done real time without needing to do the offline measurment procedure?
 
Connect your Surface Pro to your ADI-2 via USB. Connect to your Omnimic V2 via another USB port. If you don't have enough USB ports, use a port expander.

Set REW to Java mode. Set the ADI-2 as the output, and the Omnimic as the input. Play a test tone SOFTLY to confirm that all is well, then increase volume, and check levels. Set REW to use an acoustic timing reference, then do a sweep.

(edit) Oh yes, the Omnimic comes with a mini tripod. Throw it away and get a proper mic tripod if you are serious about taking good measurements.
 
Connect your Surface Pro to your ADI-2 via USB. Connect to your Omnimic V2 via another USB port. If you don't have enough USB ports, use a port expander.

Set REW to Java mode. Set the ADI-2 as the output, and the Omnimic as the input. Play a test tone SOFTLY to confirm that all is well, then increase volume, and check levels. Set REW to use an acoustic timing reference, then do a sweep.

(edit) Oh yes, the Omnimic comes with a mini tripod. Throw it away and get a proper mic tripod if you are serious about taking good measurements.
Thank you, Keith. Yeah, the Surface only has one USB port, but I do have an expander. Appreciate the advice on the trip. I will definitely give this a try.
 
Why does it start from DC instead of half of 15Hz i.e. 7.5Hz?
@JohnPM I'm still wondering this, could you please explain. If I select linear sweep and set it to start from 7Hz, it starts from 7Hz. And if I select logarithmic sweep, it also starts from 7Hz.
So why doesn't measurement sweep start from e.g. 8Hz if I select 16Hz start frequency? Why does it always start from DC i.e. 0Hz when start frequency is set to below 20Hz?
 
why doesn't measurement sweep start from e.g. 8Hz if I select 16Hz start frequency? Why does it always start from DC i.e. 0Hz when start frequency is set to below 20Hz?
That's the way I designed the measurement sweep to ensure sufficient low frequency energy to obtain a good impulse response and hence an accurate low frequency result.
 
Thank you, Keith. Yeah, the Surface only has one USB port, but I do have an expander. Appreciate the advice on the trip. I will definitely give this a try.
Turns out the 10-yr old Omnimic had either gone defective (the likely problem, I suspect) or it just would not work with REW---even using the proper mic cal file. So, I replaced it with a UMIK-1 which worked perfectly with REW using the ADI-2 dac connected to the USB port of the Surface (which actually does have two USB ports!). Thanks again for putting me onto the solution. Along with my tripod mic stand I was able to get some excellent room measurements.
 
I have a question regarding the soundcard calibration. I am under the impression that the input gain should be set to what it needs for the measurement mic to register certain externally measured SPL, e.g., 80dB. This in my setup means having the gain at 90% and calibrating that way gave some harsh peaks on the soundcard calibration measurement and the actual measurements were buggy as well after using that distorted soundcard calibration file.
I tried to calibrate the soundcard with less gain and boosting the monitoring level from my interface until input-output aligned and it worked fine but still, wondering if I'm doing things wrong. This part is not too clear on the manual.

Anyways, great to have some measurements. Having tailored low end EQ makes it sound so much better! Pic related, my measurements, LR averaged 1/12 smoothing (sorry, didn't know how to set the spl limits). Right speaker has a nasty dip at 88hz and smaller one at 200, and left one has one big at 230 but overall pretty good! I will measure again tomorrow with the suggested EQ which seems to correct the boominess in low end well based on listening tests.

Also, I noticed that EQ Apo lets you have separate EQs for speakers, so should I EQ each speaker separately or create filters based on the LR average? I can't currently have symmetrical placement of speakers in the room (left one's in the corner and right one's in the middle of the short wall) which probably plays a big role in having such weird dips at different places.
 

Attachments

  • average.png
    average.png
    14 KB · Views: 17
Last edited:
I have a question regarding the soundcard calibration. I am under the impression that the input gain should be set to what it needs for the measurement mic to register certain externally measured SPL, e.g., 80dB. This in my setup means having the gain at 90% and calibrating that way gave some harsh peaks on the soundcard calibration measurement and the actual measurements were buggy as well after using that distorted soundcard calibration file.
I tried to calibrate the soundcard with less gain and boosting the monitoring level from my interface until input-output aligned and it worked fine but still, wondering if I'm doing things wrong. This part is not too clear on the manual.

It sounds like this is what you did:

- Plug microphone in, play an 80dB sound, adjust the gain on the interface so that REW measures 80dB SPL
- Remove the microphone, connect output from interface to mic input (HOPEFULLY you remembered to turn off 48V phantom power as well!)
- Run REW calibration

If this is what you did, you would have clipped the mic input producing that peaky looking calibration file. REW should have given you a warning that you clipped the measurement? Anyway, the correct way to do it is like this:

- Turn the DAC output and mic input gains to zero
- Connect the DAC output to mic input with a cable
- Slowly turn up the DAC output and mic input gains using REW's signal generator playing noise or a tone and watch the clipping indicator on your mixer (if you have one).
- Run the soundcard calibration.

Also, I noticed that EQ Apo lets you have separate EQs for speakers, so should I EQ each speaker separately or create filters based on the LR average? I can't currently have symmetrical placement of speakers in the room (left one's in the corner and right one's in the middle of the short wall) which probably plays a big role in having such weird dips at different places.

This is a bit difficult to explain but I will try. For low frequencies, below 150-200Hz, speakers should be EQ'ed individually to give the smoothest bass response possible. Above this, speakers should be corrected according to the anechoic measurement - meaning, you need to be able to take a measurement free of reflections. This may or may not be possible depending on the size of your speakers and your room. If you have problems from asymmetric speaker placement, the most I would do is apply very broad correction to high frequencies and not try to chase every peak and dip.
 
It sounds like this is what you did:

- Plug microphone in, play an 80dB sound, adjust the gain on the interface so that REW measures 80dB SPL
- Remove the microphone, connect output from interface to mic input (HOPEFULLY you remembered to turn off 48V phantom power as well!)
- Run REW calibration

If this is what you did, you would have clipped the mic input producing that peaky looking calibration file. REW should have given you a warning that you clipped the measurement? Anyway, the correct way to do it is like this:

- Turn the DAC output and mic input gains to zero
- Connect the DAC output to mic input with a cable
- Slowly turn up the DAC output and mic input gains using REW's signal generator playing noise or a tone and watch the clipping indicator on your mixer (if you have one).
- Run the soundcard calibration.



This is a bit difficult to explain but I will try. For low frequencies, below 150-200Hz, speakers should be EQ'ed individually to give the smoothest bass response possible. Above this, speakers should be corrected according to the anechoic measurement - meaning, you need to be able to take a measurement free of reflections. This may or may not be possible depending on the size of your speakers and your room. If you have problems from asymmetric speaker placement, the most I would do is apply very broad correction to high frequencies and not try to chase every peak and dip.

Thanks for the reply! Yes that is what I basically did. Trying to match the set gain of mic input with monitoring level being 1-2% up. It did clip and abort when I accidentally tried to move the monitoring volume too fast up. Good to know.

I have read how you should not EQ room above the 200-500Hz range, but the speakers. There are quasi-anechoic measurements for the speakers (LSR305) provided by @dominikz (ty!) available, so I will probably try to mess with those a bit and keep measuring using 90degree, 0 degree, moving mic method etc. I have some 10cm deep early reflection panels set up but it's indeed not symmetrical placement and most likely can't bother to measure outside seeing there are measurements already available. Having EQ Apo with filters active in the background does probably work? I noticed REW captures audio playback from the system so just wondering that. I'm too busy to measure today but here I have added my LR measurements which show the extreme irregular dips with 1/12 smoothing. REW generated these for averaged LR already with limit from 43Hz to 500Hz and they do seem to correct some of the problems:
Filter 1: ON PK Fc 60.00 Hz Gain -5.60 dB Q 2.805
Filter 2: ON PK Fc 139.0 Hz Gain -8.70 dB Q 5.617
Filter 3: ON PK Fc 186.5 Hz Gain -3.60 dB Q 11.120

I will continue measuring and try to generate sub 200-300Hz filters for each of the speakers, and probably try to compensate for the dip at 430Hz and see how it sounds and maybe see how broad eq of -3.5dB or so at 15.5kHz reacts. I will also try to mess with some of the quasi-anechoic measurements for the upper register to see if it's any use but to me above 500Hz looks pretty good averaged anyways (or is it that the upper register measurements really aren't valid and that is why you should EQ the speakers and not the room there, well.. We will see).
 

Attachments

  • lr.png
    lr.png
    41.8 KB · Views: 20
Last edited:
Thanks for the reply! Yes that is what I basically did. Trying to match the set gain of mic input with monitoring level being 1-2% up. It did clip and abort when I accidentally tried to move the monitoring volume too fast up. Good to know.

I have read how you should not EQ room above the 200-500Hz range, but the speakers. There are quasi-anechoic measurements for the speakers (LSR305) provided by @dominikz (ty!) available, so I will probably try to mess with those a bit and keep measuring using 90degree, 0 degree, moving mic method etc. I have some 10cm deep early reflection panels set up but it's indeed not symmetrical placement and most likely can't bother to measure outside seeing there are measurements already available. Having EQ Apo with filters active in the background does probably work? I noticed REW captures audio playback from the system so just wondering that. I'm too busy to measure today but here I have added my LR measurements which show the extreme irregular dips with 1/12 smoothing. REW generated these for averaged LR already with limit from 43Hz to 500Hz and they do seem to correct some of the problems:
Filter 1: ON PK Fc 60.00 Hz Gain -5.60 dB Q 2.805
Filter 2: ON PK Fc 139.0 Hz Gain -8.70 dB Q 5.617
Filter 3: ON PK Fc 186.5 Hz Gain -3.60 dB Q 11.120

I will continue measuring and try to generate sub 200-300Hz filters for each of the speakers, and probably try to compensate for the dip at 430Hz and see how it sounds and maybe see how broad eq of -3.5dB or so at 15.5kHz reacts. I will also try to mess with some of the quasi-anechoic measurements for the upper register to see if it's any use but to me above 500Hz looks pretty good averaged anyways (or is it that the upper register measurements really aren't valid and that is why you should EQ the speakers and not the room there, well.. We will see).

You need your own quasi-anechoic measurements of your speaker. There is a difference between "speaker correction" and "room correction". Speaker correction means anechoic or quasi-anechoic measurement, free of reflections. You can examine the ETC (Energy-Time Curve) to see how early the reflections are. Most of us hobbyists are able to obtain a quasi-anechoic measurement to a lower limit determined by the size of your room, proximity of reflective surfaces/furniture, and size of your speakers. If you are lucky (like me) I am able to obtain a quasi-anechoic measurement down to about 300Hz. I have a large enough room and I don't have to measure the entire speaker, only individual drivers.

"Room correction" means correcting for anomalies created by the room. This means fixing all the peaks below 200Hz and broadly fixing egregious volume differences between L/R speaker above that.

Your measurement is too zoomed out to see anything. Do this:

1752234462985.png

We typically show a y-axis spanning 50dB. Yours spans 220dB, this flattens the curve and makes it look better than it really is.

Also, suggest you read this thread. If you need more help, feel free to start another thread.
 
Install REW and run it.
Nice thread that I just discovered.
But especially for dummies, the pictures from the start post might be important to understand.
Can you perhaps adapt the start post and use the current program version?

1000035330.png
 
Back
Top Bottom