• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. 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!

REW measurement tones/noises

5-pot-fan

Member
Joined
Nov 5, 2021
Messages
67
Likes
55
Hi,
Having played with REW for a few days I have one initial query before going further.
I am using REW v5.20.13 + Umik-1 mic with 90deg cal file and decided to test both Sweep tones and Pink noise , which have generated the 2 curves in the screengrab below. Both curves are averages of 3 runs, all output from L+R speakers, from almost identical mic positions, no smoothing.
The sweep curve (in red) looks moderately good, while the PN curve would require rather more work, if this was the 'correct' curve to work from.
Can anyone advise me which test tones are most appropriate, and therefore which type of curve to work from?
The room is approx 2.5m w x 3.3m l x 2.1 m high, speakers on short wall, no subs, no room treatment, no EQ, carpeted, with bookshelves, and a thick curtain over the 1 window, if relevant.
I will provide more background info if needed....
Thanks in anticipation


1698325077929.png
 

staticV3

Master Contributor
Joined
Aug 29, 2019
Messages
8,315
Likes
13,386
The sweep curve in red just shows the mic's noise floor.
If it's meant to show the in-room frequency response of a pair of loudspeakers, then something went wrong during the measurement.

Same story for measurements 1, 4, 6, 8, and 10 in your screenshot. None of these seem to have captured any usable data. Only noise.
 

Philbo King

Addicted to Fun and Learning
Joined
May 30, 2022
Messages
669
Likes
882
To get a useful curve from pink noise the result must be heavily averaged over quite a long period.
 

staticV3

Master Contributor
Joined
Aug 29, 2019
Messages
8,315
Likes
13,386
To get a useful curve from pink noise the result must be heavily averaged over quite a long period.
If you look closely, only the pink noise measurements look successful. It's the sweep measurements that did not work.
 
OP
5-pot-fan

5-pot-fan

Member
Joined
Nov 5, 2021
Messages
67
Likes
55
Thank you both.
I can run a pink noise for longer and see what this produces.
Also, will look into why the sweep appears to be relatively meaningless.
Is there any reason why a pn curve is more/less useful than a sweep as the basis for eq?
 

Sokel

Master Contributor
Joined
Sep 8, 2021
Messages
6,294
Likes
6,440
There's something wrong with the pink noise too,can't be that flat with it,highs should decline considerably.
In this measurement is the opposite.
 

staticV3

Master Contributor
Joined
Aug 29, 2019
Messages
8,315
Likes
13,386
There's something wrong with the pink noise too,can't be that flat with it,highs should decline considerably.
Only if the speaker is anechoically flat.

Also, OP's choice of 90dB vertical range is compressing the measurement, making it look flatter than it is.
 

Sokel

Master Contributor
Joined
Sep 8, 2021
Messages
6,294
Likes
6,440
Only if the speaker is anechoically flat.

Also, OP's choice of 90dB vertical range is compressing the measurement, making it look flatter than it is.
Can it also be Mic placement?
For example some speakers acoustic center is lower than the tweeter,usually at the mid's upper ring,measuring such a speaker at tweeter's level would seem as the highs are higher than it is,wouldn't be?
 

staticV3

Master Contributor
Joined
Aug 29, 2019
Messages
8,315
Likes
13,386
Can it also be Mic placement?
For example some speakers acoustic center is lower than the tweeter,usually at the mid's upper ring,measuring such a speaker at tweeter's level would seem as the highs are higher than it is,wouldn't be?
Depending on the speakers' directivity, that can affect the measurements slightly, but to a degree that would likely be invisible at OP's current graph settings.

Furthermore, spatial averaging using MMM will take care of that as well.
 
OP
5-pot-fan

5-pot-fan

Member
Joined
Nov 5, 2021
Messages
67
Likes
55
Thanks again for your replies - such a great forum!
1) Though using the 90deg cal file I had the mic horizontal at the LP. Now changed so it is vertical .....
2) Sweep measurements were faulty because I used the Generator tab to create them. Using the Measurement tab and stopping the recording before the sweep finishes now gives me curves with shapes similar to the Pink Noise ones, and therefore probably useful.
3) Have also used recorded PN and Sweep files which I then played back through Moode Audio (no EQ) and obtained roughly similar curves.

I won't put any of these up yet until I have tried to understand what I am seeing, which will include some movement of speakers and some EQ tests using Moode/Camilla, at which point I hope to come back with more questions !
 
OP
5-pot-fan

5-pot-fan

Member
Joined
Nov 5, 2021
Messages
67
Likes
55
OP
5-pot-fan

5-pot-fan

Member
Joined
Nov 5, 2021
Messages
67
Likes
55
And another followup...
Just seen the YT "Room calibration at EchoBar studios" which is very helpful regarding the use of REW as well as in taking measurements. I don't have a Scarlett or aMinidsp but may have enough other tools to make some headway, so back to the beginning and try again!
Thanks for all your tips so far.
 
OP
5-pot-fan

5-pot-fan

Member
Joined
Nov 5, 2021
Messages
67
Likes
55
I have now successfully made measurements in REW, with repeatable results, and am now experimenting with using REW to provide EQ filters for my 2.0 system, especially following an upgrade to v5.20.75 (beta).
My current difficulty is transferring the filter EQ settings into Moode/CamillaDSP. I can do it manually (laborious) for the pipeline filters, but is there a way to convert the data between the 2 systems? The version of .yml and .txt output fromREW are not directly compatible with Camilla, nor viceversa.
I am also trying to get .wav files fromREW but am having trouble getting the filter levels into the correct region. I will perhaps adopt the wav/convolution route if I can get it to work, but would still like to know about the conversion issue mentioned.
Thanks
 
Top Bottom