RoomFit Moving Mic Measurement (beta) - improvements

[slartibartfast]

I guess they just want to reduce the number of variables to see if that helps pin-point the cause of the issue you are seeing.


Regarding this part of their response - I agree that some small variability is expected, but not the amount you saw in your measurements.
In my experience the match between predicted and actual responses is generally very good (in both RoomFit and REW).

Exactly and the results I posted were using filters generated by REW so that removes all the variables associated with RoomFit as effectively I am only using PEQ and checking the results using REW and WiiM evaluation. I got the impression that the filters generated by REW gave similar results to those generated by RoomFit. Both gave a similar non flat response when checked with REW. So it seems to me Roomfit is calculating the correct filters but they aren't being applied correctly. I suppose performing the RoomFit should provide all the info about the raw waveforms etc which isn't available if I manually enter filters. It would be a lot easier to provide the info asked for if it was possible to export from the WiiM app.

Edit. How do WiiM receive system logs if I just provide results by uploading images to the ticket? I would need to provide feedback from the app again and that would open a new ticket.

 

[Newzpix]

As @slartibartfast said, "Individual Channel RoomFit" already covers that case for the calibration step.

In case you are referring to the evaluation step (which always uses combined L+R measurement signal), then a potentially easier way (than physically disconnecting speakers) may be to use the L/R balance slider prior to evaluation - but note that in that case you need to do two separate evaluations to measure both channels.

Would never disconnect a speaker while the amp is on

 

[Newzpix]

Exactly and the results I posted were using filters generated by REW so that removes all the variables associated with RoomFit as effectively I am only using PEQ and checking the results using REW and WiiM evaluation. I got the impression that the filters generated by REW gave similar results to those generated by RoomFit. Both gave a similar non flat response when checked with REW. So it seems to me Roomfit is calculating the correct filters but they aren't being applied correctly. I suppose performing the RoomFit should provide all the info about the raw waveforms etc which isn't available if I manually enter filters. It would be a lot easier to provide the info asked for if it was possible to export from the WiiM app.

Edit. How do WiiM receive system logs if I just provide results by uploading images to the ticket? I would need to provide feedback from the app again and that would open a new ticket.

I'm not an audio engineer, I'm a photojournalist. We speak different languages.

 

[slartibartfast]

I'm not an audio engineer, I'm a photojournalist. We speak different languages.

This wasn't addressed to you

 

[Newzpix]

This wasn't addressed to you

You sound like my Mom

 

[slartibartfast]

I guess they just want to reduce the number of variables to see if that helps pin-point the cause of the issue you are seeing.


Regarding this part of their response - I agree that some small variability is expected, but not the amount you saw in your measurements.
In my experience the match between predicted and actual responses is generally very good (in both RoomFit and REW).

This is driving me mad. I've done stereo sweep, stereo mmm and L/R sweep in Roomfit. Each time REW measurements agree with WiiM and the response looks fine. As soon as I do a L/R mmm measurement in Roomfit the REW measurement is not even close to flat and is exactly the same as the measured response when using the filters generated by REW.
So the issue only occurs with mmm L/R mode in Roomfit but the REW generated filters are totally independent of Roomfit but actually exactly the same.

I guess they just want to reduce the number of variables to see if that helps pin-point the cause of the issue you are seeing.


Regarding this part of their response - I agree that some small variability is expected, but not the amount you saw in your measurements.
In my experience the match between predicted and actual responses is generally very good (in both RoomFit and REW).

I have measured the response in REW using mmm with the microphone stationary.
1. WiiM Roomfit Sweep individual channel
2. WiiM Roomfit Sweep stereo
3. WiiM Roomfit MMM stereo
4. Wiim Roomfit MMM individual channel
5. REW MMM individual channel

You can see the first three lie on top of each other but the last two are not even close but similar to each other. I haven't got a clue how to explain this to WiiM.



Edit. If I use Roomfit in MMM individual channel mode but leave the microphone in the same place throughout the measurement then I get a good flat result. What does that mean?

 

[dominikz]

This is driving me mad. I've done stereo sweep, stereo mmm and L/R sweep in Roomfit. Each time REW measurements agree with WiiM and the response looks fine. As soon as I do a L/R mmm measurement in Roomfit the REW measurement is not even close to flat and is exactly the same as the measured response when using the filters generated by REW.
So the issue only occurs with mmm L/R mode in Roomfit but the REW generated filters are totally independent of Roomfit but actually exactly the same.


I have measured the response in REW using mmm with the microphone stationary.
1. WiiM Roomfit Sweep individual channel
2. WiiM Roomfit Sweep stereo
3. WiiM Roomfit MMM stereo
4. Wiim Roomfit MMM individual channel
5. REW MMM individual channel

You can see the first three lie on top of each other but the last two are not even close but similar to each other. I haven't got a clue how to explain this to WiiM.

View attachment 32142

Can you also show the filters for each of these correction attempts?
They should be visibly different as well if the measured responses are this different - maybe that will give some clues?

 

[slartibartfast]

Can you also show the filters for each of these correction attempts?
They should be visibly different as well if the measured responses are this different - maybe that will give some clues?

Definitely different

First the stationary mic



Then the moving microphone

 

[dominikz]

This is driving me mad. I've done stereo sweep, stereo mmm and L/R sweep in Roomfit. Each time REW measurements agree with WiiM and the response looks fine. As soon as I do a L/R mmm measurement in Roomfit the REW measurement is not even close to flat and is exactly the same as the measured response when using the filters generated by REW.
So the issue only occurs with mmm L/R mode in Roomfit but the REW generated filters are totally independent of Roomfit but actually exactly the same.


I have measured the response in REW using mmm with the microphone stationary.
1. WiiM Roomfit Sweep individual channel
2. WiiM Roomfit Sweep stereo
3. WiiM Roomfit MMM stereo
4. Wiim Roomfit MMM individual channel
5. REW MMM individual channel

You can see the first three lie on top of each other but the last two are not even close but similar to each other. I haven't got a clue how to explain this to WiiM.

View attachment 32142

Edit. If I use Roomfit in MMM individual channel mode but leave the microphone in the same place throughout the measurement then I get a good flat result. What does that mean?

View attachment 32143

Hmm, how fast do you move the mic when doing MMM, and how wide an area do you cover? How far from you body do you hold the mic?

I usually move the mic pretty slowly (I'd say about 4 inch per second), within an area of about 2ft cross-section, and at an arms-length from my body (while making sure I'm never blocking the direct sound path between the speakers and mic).

 

[slartibartfast]

Hmm, how fast do you move the mic when doing MMM, and how wide an area do you cover? How far from you body do you hold the mic?

I usually move the mic pretty slowly (I'd say about 4 inch per second), within an area of about 2ft cross-section, and at an arms-length from my body (while making sure I'm never blocking the direct sound path between the speakers and mic).

Fairly slowly. Mainly in a line covering the seating area of small sofa so about 3-4 feet.
I can understand a single point measurement wouldn't be flat using filters generated by mmm but I would expect the same moving mic measurement to be reasonably flat since the mmm measurements are very repeatable. The stereo mmm doesn't suffer the same issue.

 

[slartibartfast]

Hmm, how fast do you move the mic when doing MMM, and how wide an area do you cover? How far from you body do you hold the mic?

I usually move the mic pretty slowly (I'd say about 4 inch per second), within an area of about 2ft cross-section, and at an arms-length from my body (while making sure I'm never blocking the direct sound path between the speakers and mic).

If I move over a smaller area about 2 feet wide I get a better result. this shows a raw large movement, the smaller movement and the result of the small movement. Green is the smaller movement. I am not sure where the dip at 35 Hz comes from. It isn't there with sweeps IIRC

 

[dominikz]

If I move over a smaller area about 2 feet wide I get a better result. this shows a raw large movement, the smaller movement and the result of the small movement. Green is the smaller movement. I am not sure where the dip at 35 Hz comes from. It isn't there with sweeps IIRC

View attachment 32149

I was just about to suggest to try measuring over a smaller area, looks like you had the same idea. Great!

The 35Hz dip looks like it might be caused by suboptimal correction filter allocation - I've seen this in the past with RoomFit. You might be able to do better with REW calculated filters.

 

[slartibartfast]

I was just about to suggest to try measuring over a smaller area, looks like you had the same idea. Great!

The 35Hz dip looks like it might be caused by suboptimal correction filter allocation - I've seen this in the past with RoomFit. You might be able to do better with REW calculated filters.

That dip is present in the raw mmm measurement but not with the sweep
purple is the sweep

 

[slartibartfast]

Maybe the dip is caused by the back of the leather sofa. Moving the mic into the seating position rather than above the sofa back gives this