First REW measurement, please let me know what I can do better!

[Waldemar-Johann]

Good day everyone,
I have now measured for the first time.
Looks the measurement ok?

Question here about how I can improve the room
The way the monitoring system is set up now, it has to stay that way.

I look forward to professional feedback and help!

All the best and a great day,
Waldemar

 

[Keith_W]

I have asked a mod to split your old thread and merge it with this one, so all the replies don't get lost.

Before we go any further:
1. Please read this thread: How to ask for help with REW in ASR
2. Your five files have duplicate measurements. In the future, delete all duplicate measurements and only post relevant measurements.
3. Give your measurements proper names. All I know is that it's L and R. Any subwoofers?

Since you asked about how to improve your room, the relevant measurements are the last three files which contain timing information. MMM measurements can't be used to look at the room. Of the 18 measurements you posted, I threw away 16 of them. Here are a couple that I threw away:

What is going on here?!?!? I am guessing that these are L/R speaker MMM's taken at 75dB and 80dB. They are very different to your single point logsweeps. So these are clearly nonsense measurements, they need to be thrown away. If it looks strange to you, delete them and try again.

Red/Blue = left/right speaker. Yellow = L and R speaker measured together. Notice how the combined measurement has a treble boost from 6kHz up (circled). This measurement is clearly flawed and should be rejected.

As I pointed out in my post in the other thread, comparison between your measurement and published measurements of your speakers suggests you have a huge cancellation between 55Hz - 110Hz. This is normal for all speakers in all rooms, but you may be able to improve your situation by repositioning the speakers or listening position, or both.

If we look at your RT60 (here, only the right speaker) we see that your RT60 is on the "wet" side above 100Hz. Read more about the RT60 here. For a very small room like yours, these are not true numbers since these are specular reflections and not reverberant fields.

The interesting thing is - it really shoots through the roof below 100Hz. If you see this, we normally think "room mode". But with a room mode, you expect to see a big spike in the freq response graph at 50Hz (since that is where the RT60 seems to peak). But in this case, there is no spike in the FR graph. So why is there so much "reverberation" at 50-60Hz?

Answer: poor signal to noise ratio / high noise floor. This is your waterfall graph, which I extended to 1000ms and ticked the "normalise to peak at each frequency" option. It is likely that if you chose a quieter time to measure, your "RT60" will improve.

There are some good things though. Your Energy-Time Curve (ETC) shows there are no reflections louder than -15dB in the first 20ms. Probably because you are sitting so close to your speakers.

 

[Waldemar-Johann]

Hello Keith,
Thank you very much for your continued help and information.

1. Please read this thread: How to ask for help with REW in ASR

I have now done this again and hope that I have understood everything correctly and can now provide you with the correct data.

I have now ordered a MINIDSP UMIK-1.
Furthermore, my speaker pair is the Quested V2108.
I also have a subwoofer, Quested SB10R, but it is not connected at the moment.

As suggested, I have rearranged the speakers and my workspace.

In the Room.zip file, you will find the dimensions of the room, the speakers and a 3D image.
I am also adding the link from GIK Acoustic to the Room Acoustic Visualiser, https://www.roomle.com/t/cp/?config...otTag[]=gikacoustics_eur_root&api=false&state. mode=room&buttons.requestplan=false&id=ps_kpbxhc2bxj8dc1xowetqyd7gb1pqksj&locale=en&usePriceService=false

The room height is 3.03 m, but Room Acoustic Visualizer displays it incorrectly.

The measurement appendix contains a picture of the settings for the measurement, in case it is important.

I hope that all the information is there now and that I have measured correctly.

Thank you!

All the best and a great day,
Waldemar

 

[Keith_W]

With reference to my previous post, I would like to draw your attention to these two measurements that you just posted:

Once again, very high noise floor making the bass RT60 shoot through the roof. Is it very noisy where you are? If you want to improve the signal to noise ratio, you can try:

- choose a longer sweep, e.g. 1M or 4M.
- take multiple sweeps and vector average them
- you can try playing your speakers louder, but in general this is not recommended because you might drive them to distortion.
- choose a quieter time to measure

Also, i'm not sure what question you are asking or why you are taking these measurements.

(EDIT)

I just saw your other ZIP file with the images. This is a very bad idea! There should be no obstructions between your high freq drivers and your ears. Ideally, no obstructions at all between your speakers and your ears. Having a table which obstructs the entire speaker is the worst solution. If you must have your table there, I suggest you sell your floorstanding speakers and use bookshelf speakers instead. Put them on your desk at ear height.

 

[Waldemar-Johann]

Good evening Keith,

thank you for coming back.

Once again, very high noise floor making the bass RT60 shoot through the roof. Is it very noisy where you are? If you want to improve the signal to noise ratio, you can try:

No, it was not noisy. But of course I will do it later on with other parameter? I wasn't in the room when the measurement was taken either.

- choose a longer sweep, e.g. 1M or 4M.

I used 1m but can try 4m.

- take multiple sweeps and vector average them

Do you mean instead of Single measurement, Repeated measurements, see attach file Mode?

you can try playing your speakers louder, but in general this is not recommended because you might drive them to distortion.

I have set the level to -30.00 dBFS. I can adjust it to -20.00 dBFS if that is sufficient?

Also, i'm not sure what question you are asking or why you are taking these measurements.

I did make a mistake ;-)
My question is, can I possibly improve the problems with a DSP?

I just saw your other ZIP file with the images. This is a very bad idea! There should be no obstructions between your high freq drivers and your ears. Ideally, no obstructions at all between your speakers and your ears. Having a table which obstructs the entire speaker is the worst solution. If you must have your table there, I suggest you sell your floorstanding speakers and use bookshelf speakers instead. Put them on your desk at ear height.

My second mistake!
The speakers are on proper speaker stands, which I couldn't draw in.

All the best,
Waldemar

 

[kemmler3D]

My question is, can I possibly improve the problems with a DSP?

I'm not as sharp as Keith when it comes to REW, but the answer here is "yes".

If you use EQ you can flatten out the peaks in the bass, which should give a strong audible benefit. REW can generate the EQ curves for you - you just need a DSP capable of implementing them. If you are listening on a Windows PC, EQAPO is a free solution that works well. On Mac OS, you have Soundsource. And then for standalone equipment, MiniDSP and WiiM have these capabilities.

 

[Waldemar-Johann]

I'm not as sharp as Keith when it comes to REW, but the answer here is "yes".

If you use EQ you can flatten out the peaks in the bass, which should give a strong audible benefit. REW can generate the EQ curves for you - you just need a DSP capable of implementing them. If you are listening on a Windows PC, EQAPO is a free solution that works well. On Mac OS, you have Soundsource. And then for standalone equipment, MiniDSP and WiiM have these capabilities.

Thank you very much for your feedback and information.
That sounds good.
I just think that the measurement I took needs to be improved, as I understand Keith.
Once that has been exhausted, I will do exactly that at this point.

All the best,
Waldemar

 

[Keith_W]

Do you mean instead of Single measurement, Repeated measurements, see attach file Mode?

I have set the level to -30.00 dBFS. I can adjust it to -20.00 dBFS if that is sufficient?

-30dBFs or -20dBFs does not tell me the actual volume your speakers are reproducing. That depends on the gain structure on the rest of your system, sensitivity of speakers, volume control setting, etc. What matters is the whether the speakers are being measured in its linear range. One way to tell if it's in its linear range is to look for frequency response anomalies like this:

Or by looking for high distortion:

On the left: distortion in your current measurements. On the right: distortion from the measurements you posted 2 weeks ago.

Comparison of your current measurements with your previous set tells me that this new set was taken at higher SPL and you are pushing your speakers out of its linear range. Remember that speakers are nonlinear devices. There are mechanical limitations to how loud speakers can go. Voice coils can heat up, particularly if speakers are played loudly for a long time. Some speakers are worse than others, and whether you encounter those limitations depends on how loud you play your speakers.

My recommendation is to measure at the same volume that you listen, not at some arbitrary volume (e.g. REW recommends 76dB). If you see a lot of distortion, you need new speakers. If the room noise floor is very high at your listening volume, you need to do something about it.

For now, the poor signal to noise ratio is sending your "RT60" through the roof. Yes, you can take 5 sweeps with the setting that you posted. Use 1M. Then select all the measurements in REW in the "ALL SPL" tab and then right click and vector average.

My question is, can I possibly improve the problems with a DSP?

DSP only helps if you can take proper measurements that represent reality. If your measurements are all over the place, you will be correcting phantom problems and very likely introducing new problems that were not there before. So your first priority should be to learn to take proper measurements and understand what you are doing.

The speakers are on proper speaker stands, which I couldn't draw in.

I still have major concerns about how you have set your speakers up. Would it be possible to show us a photo of your speakers and your desk?

 

[Waldemar-Johann]

-30dBFs or -20dBFs does not tell me the actual volume your speakers are reproducing. That depends on the gain structure on the rest of your system, sensitivity of speakers, volume control setting, etc. What matters is the whether the speakers are being measured in its linear range. One way to tell if it's in its linear range is to look for frequency response anomalies like this:

View attachment 466471

Or by looking for high distortion:

View attachment 466467View attachment 466478

On the left: distortion in your current measurements. On the right: distortion from the measurements you posted 2 weeks ago.

Comparison of your current measurements with your previous set tells me that this new set was taken at higher SPL and you are pushing your speakers out of its linear range. Remember that speakers are nonlinear devices. There are mechanical limitations to how loud speakers can go. Voice coils can heat up, particularly if speakers are played loudly for a long time. Some speakers are worse than others, and whether you encounter those limitations depends on how loud you play your speakers.

My recommendation is to measure at the same volume that you listen, not at some arbitrary volume (e.g. REW recommends 76dB). If you see a lot of distortion, you need new speakers. If the room noise floor is very high at your listening volume, you need to do something about it.

For now, the poor signal to noise ratio is sending your "RT60" through the roof. Yes, you can take 5 sweeps with the setting that you posted. Use 1M. Then select all the measurements in REW in the "ALL SPL" tab and then right click and vector average.



DSP only helps if you can take proper measurements that represent reality. If your measurements are all over the place, you will be correcting phantom problems and very likely introducing new problems that were not there before. So your first priority should be to learn to take proper measurements and understand what you are doing.



I still have major concerns about how you have set your speakers up. Would it be possible to show us a photo of your speakers and your desk?

Hello Keith,
Thank you very much for your feedback.

Comparison of your current measurements with your previous set tells me that this new set was taken at higher SPL and you are pushing your speakers out of its linear range. Remember that speakers are nonlinear devices. There are mechanical limitations to how loud speakers can go. Voice coils can heat up, particularly if speakers are played loudly for a long time. Some speakers are worse than others, and whether you encounter those limitations depends on how loud you play your speakers.

To be honest, I don't understand some of the measurements at all, which is why I'm here.
But I'll try it, provided the measurement is correct!

My recommendation is to measure at the same volume that you listen, not at some arbitrary volume (e.g. REW recommends 76dB). If you see a lot of distortion, you need new speakers. If the room noise floor is very high at your listening volume, you need to do something about it.

Thank you very much for that. I'll read it the first and am grateful for any input!

For now, the poor signal to noise ratio is sending your "RT60" through the roof. Yes, you can take 5 sweeps with the setting that you posted. Use 1M. Then select all the measurements in REW in the "ALL SPL" tab and then right click and vector average.

Attach you will find the file. Hope it will be correct.

DSP only helps if you can take proper measurements that represent reality. If your measurements are all over the place, you will be correcting phantom problems and very likely introducing new problems that were not there before. So your first priority should be to learn to take proper measurements and understand what you are doing.

I made another mistake.
If I get the measurements right, and you should never give up hope, the next question is: does it make sense to integrate the subwoofer?
Then the next question is: what acoustic treatment can I do if possible?

I still have major concerns about how you have set your speakers up. Would it be possible to show us a photo of your speakers and your desk?

Attached is the speaker pair + the height.

All the best and a nice evening
Waldemar

 

[Keith_W]

All right, it looks as if your speakers are positioned properly after all.

Here are 3 measurements. In thick yellow, is your latest measurement. In green and red are your previous measurements from 15 Jul and 29 Jul. If you want to DSP, you need to be able to take consistent measurements.

Here, we can see that the treble bump between 6kHz - 10kHz seems to move between measurements. This is because you are measuring at different volumes, thus producing different amounts of harmonic distortion. You can see this if you click on the "Distortion" tab. If the top end seems more shrill / more sibilant as you turn up the volume, then it is a problem. Nothing will fix it - not room treatment, not DSP. The only solution is to buy a better speaker. You can decide if it bothers you enough to do that.

On the other hand, your bass measurements, especially below 100Hz, are quite consistent. This is what you will be attempting to fix with DSP. So to answer your question - yes, you would benefit from DSP.

 

[Waldemar-Johann]

All right, it looks as if your speakers are positioned properly after all.

View attachment 466645

Here are 3 measurements. In thick yellow, is your latest measurement. In green and red are your previous measurements from 15 Jul and 29 Jul. If you want to DSP, you need to be able to take consistent measurements.

Here, we can see that the treble bump between 6kHz - 10kHz seems to move between measurements. This is because you are measuring at different volumes, thus producing different amounts of harmonic distortion. You can see this if you click on the "Distortion" tab. If the top end seems more shrill / more sibilant as you turn up the volume, then it is a problem. Nothing will fix it - not room treatment, not DSP. The only solution is to buy a better speaker. You can decide if it bothers you enough to do that.

On the other hand, your bass measurements, especially below 100Hz, are quite consistent. This is what you will be attempting to fix with DSP. So to answer your question - yes, you would benefit from DSP.

Good morning Keith,
Thank you very much for your quick response and the information.

Here are 3 measurements. In thick yellow, is your latest measurement. In green and red are your previous measurements from 15 Jul and 29 Jul. If you want to DSP, you need to be able to take consistent measurements.

Does that mean I should repeat the measurement I took yesterday several times?
I have made it much quieter, as you recommended.

Here, we can see that the treble bump between 6kHz - 10kHz seems to move between measurements. This is because you are measuring at different volumes, thus producing different amounts of harmonic distortion. You can see this if you click on the "Distortion" tab. If the top end seems more shrill / more sibilant as you turn up the volume, then it is a problem. Nothing will fix it - not room treatment, not DSP. The only solution is to buy a better speaker. You can decide if it bothers you enough to do that.

Thank you for your honest answers; that's what I love and live by myself. However, I am very surprised that you are telling me to ‘buy better speakers’. Shouldn't I first contact the manufacturer to find out whether the measurements of the speakers themselves are OK?
I hope my English is easy to read and that you are not offended?

On the other hand, your bass measurements, especially below 100Hz, are quite consistent. This is what you will be attempting to fix with DSP. So to answer your question - yes, you would benefit from DSP.

Wouldn't it be better to test what I wrote last time first?

I made another mistake.
If I get the measurements right, and you should never give up hope, the next question is: does it make sense to integrate the subwoofer?
Then the next question is: what acoustic treatment can I do if possible?

Thank you once again for help and answer.

All the best and a great day,
Waldemar

 

[Keith_W]

I just checked the price of your speakers - Quested V2108. Holy cow!!! Not cheap!! At that price you could have bought a nice pair of Genelecs! I was unable to find any online measurements of your speakers.

If you are going to contact the manufacturer of your speakers to complain, you should try to obtain really good measurements to confirm the problem. If you have your UMIK-1, do a loudspeaker compression test. This is how:

1. Set your speaker up with no table in the way and your mic on a proper tripod. Put the speaker in the middle of the room, away from any furniture or boundaries that may cause reflections. It would be even better if you could take your speaker outside. The mic should be pointing at the speaker with the correct cal file loaded, exactly 1m away.

2. Using one speaker only, play pink noise in REW and adjust the output until it reads 65dB. Take a single 4M sweep. Label this "65dB"

3. Increase the electrical output in REW by 10dB. Do not adjust the output with pink noise, just increase the output by 10dB. Take another 4M sweep. Label this "75dB"

4. Increase the electrical output by 10dB again and take another 4M sweep. Label this "85dB"

5. Repeat with another 10dB increase. Label this "95dB".

6. Post the MDAT with all 3 measurements to ASR.

The goal is to compare the frequency response at different volumes to see if it stays flat. A perfect speaker will product the same FR no matter how loud or soft it plays. The actual formal test is 3 volumes starting at 75dB, but I suspect your speakers distort quite early which is why I am asking you to start at 65dB.

 

[Waldemar-Johann]

I just checked the price of your speakers - Quested V2108. Holy cow!!! Not cheap!! At that price you could have bought a nice pair of Genelecs! I was unable to find any online measurements of your speakers.

If you are going to contact the manufacturer of your speakers to complain, you should try to obtain really good measurements to confirm the problem. If you have your UMIK-1, do a loudspeaker compression test. This is how:

1. Set your speaker up with no table in the way and your mic on a proper tripod. Put the speaker in the middle of the room, away from any furniture or boundaries that may cause reflections. It would be even better if you could take your speaker outside. The mic should be pointing at the speaker with the correct cal file loaded, exactly 1m away.

2. Using one speaker only, play pink noise in REW and adjust the output until it reads 65dB. Take a single 4M sweep. Label this "65dB"

3. Increase the electrical output in REW by 10dB. Do not adjust the output with pink noise, just increase the output by 10dB. Take another 4M sweep. Label this "75dB"

4. Increase the electrical output by 10dB again and take another 4M sweep. Label this "85dB"

5. Repeat with another 10dB increase. Label this "95dB".

6. Post the MDAT with all 3 measurements to ASR.

The goal is to compare the frequency response at different volumes to see if it stays flat. A perfect speaker will product the same FR no matter how loud or soft it plays. The actual formal test is 3 volumes starting at 75dB, but I suspect your speakers distort quite early which is why I am asking you to start at 65dB.

Hello Keith,

thank you for coming back and the information.

1. Set your speaker up with no table in the way and your mic on a proper tripod. Put the speaker in the middle of the room, away from any furniture or boundaries that may cause reflections. It would be even better if you could take your speaker outside. The mic should be pointing at the speaker with the correct cal file loaded, exactly 1m away.

Could I also place the loudspeaker on the balcony?
Should the microphone be set to the same height as it is now, i.e. at ear level?

2. Using one speaker only, play pink noise in REW and adjust the output until it reads 65dB. Take a single 4M sweep. Label this "65dB"

What does ‘pink noise’ mean?
In the generator, under the Noise tab, I have either ‘Pink random’ or ‘Pink periodic’.

Thank you for help and answer.

All the best,
Waldemar

 

[Keith_W]

Could I also place the loudspeaker on the balcony?
Should the microphone be set to the same height as it is now, i.e. at ear level?

The idea is that the speaker should be in a free-field, i.e. no reflections. If you can't get "no reflections", then as few reflections as possible. I don't know what your balcony looks like. The microphone should be placed on axis to the speaker. You could place your mic on the balcony and the speaker in your apartment, that way there will be no reflections from behind the mic. All depends.

What does ‘pink noise’ mean?
In the generator, under the Noise tab, I have either ‘Pink random’ or ‘Pink periodic’.

Pink random noise. All you are doing with it is setting the output level for REW.

 

[Waldemar-Johann]

Hello Keith,
I have now taken a measurement with the right speaker, but I am uncertain if it was done correctly.
It was very loud on the third run!

 

[Keith_W]

Well, that's surprising.

You did not label your graphs with the SPL's. So I took a guess and labelled them 75dB/85dB/95dB.

The really surprising thing is that when you align the SPL's (All SPL, right click, align SPL at 1kHz with 2 octaves) the responses overlay each other perfectly. This is unbelievably good performance, I don't think that even KEF's do that.

The other interesting thing that blows my mind is that the quietest measurement has more distortion than the loudest measurement, and the pattern of distortion is completely unchanged.

This is your current measurement (blue) compared to your previous measurement (red). The bass is different (as you would expect from moving your speaker to a different position), so we will ignore that. But what is surprising is that there is no upwards bump from 6kHz onwards. Maybe the high treble bump is due to table bounce, but I would have thought that would occur at a lower frequency.

The other possibility is that your previous measurement is of both speakers together, and it is your left speaker which is causing the high frequency bump.

Examination of your previous measurement from 15 Jul (yellow = left speaker, green = right speaker) shows that the left speaker has a very small bump above 12kHz. I have no idea what SPL this was measured at, or if it is even real.

Anyway, there are some very strange things going on here:

1. I am not aware of ANY speaker on the market that has such linearity at different SPL's.
2. Your speaker must be the only one on the market that has more distortion at low volume than at higher volume
3. It is possible that your left speaker is the culprit.

Because (1) and (2) are so unbelievable, I am inclined to think that you took your measurements incorrectly. It almost looks as if you took the measurements 3 times and then increased the SPL in REW.

Watch this video by Erin on how he does compression testing:

 

[kemmler3D]

the quietest measurement has more distortion than the loudest measurement

This would tend to be due to noise, right?

 

[Keith_W]

This would tend to be due to noise, right?

Nah the noise floor looks about the same in the distortion measurements. You can download it and take a look.

BTW another explanation for the rising treble response above 10kHz is microphone orientation / improper calibration file. @Waldemar-Johann please confirm pointing your mic at the speakers in ALL measurements, and with the correct calibration file loaded?

 

[Sokel]

Nah the noise floor looks about the same in the distortion measurements. You can download it and take a look.

BTW another explanation for the rising treble response above 10kHz is microphone orientation / improper calibration file. @Waldemar-Johann please confirm pointing your mic at the speakers in ALL measurements, and with the correct calibration file loaded?

It's a setting, you can toggle this to see the distortion relative to the fundamental.
Example:

 

[marin.weigel]

BTW another explanation for the rising treble response above 10kHz is microphone orientation / improper calibration file.

I've been lurking all along, trying to stay out of and not getting sucked into the discussion, but there's just one comment I'd like to add.

Looking at all frequency responses overlayed, instead we're seeing cancellation in the L+R measurement between ~3.5-9kHz ...

... caused by a ~85us time of flight difference between individual speakers' impulse response peaks:

See both speakers' individually measured impulse responses overlayed for comparison:

When not physically perfectly mirroring both speakers in 3D space relative to the microphone, measurement results are incorrect in the high frequencies and should be discarded for any sort of correction.

The takeaway from this is to better take individual measurements and work from there.