Home Theater Forum and Systems banner

1 - 6 of 6 Posts

·
Registered
Joined
·
261 Posts
Discussion Starter #1
I have 2 suggestions that come directly from Holm but do make that app a bit less fiddly to use (for speaker design use) + 1 other suggestion I've dreamt up

1) provide "curtains" on the IR that you can drag left and right to set the left/right window positions
2) provide an "auto detect" window size that looks for the 1st reflection and sizes accordingly
3) add another option that functions like the "apply windows to all" button except that it applies a fixed window width on the detected ref time rather than just overwriting all the values

TIA!
 

·
Premium Member
Joined
·
6,711 Posts
1) provide "curtains" on the IR that you can drag left and right to set the left/right window positions
That's been on the todo list for a long time, afraid it will be there some while longer

2) provide an "auto detect" window size that looks for the 1st reflection and sizes accordingly
Bit dubious about this one. How big does a peak have to be to count as a reflection? What's the minimum time after the peak before a reflection counts?

3) add another option that functions like the "apply windows to all" button except that it applies a fixed window width on the detected ref time rather than just overwriting all the values
If by "detected ref time" you mean keep what the measurement's ref time currently is, I've done that for the next build.
 

·
Registered
Joined
·
261 Posts
Discussion Starter #3
That's been on the todo list for a long time, afraid it will be there some while longer
I thought that might be the case as it seems a bigger ui change. Is there a cheap option though? For example another keyboard modifier that locks the y zoom values to "full scale"? Bit of a hack but still


Bit dubious about this one. How big does a peak have to be to count as a reflection? What's the minimum time after the peak before a reflection counts?
Typically means a few ms otherwise you have a user error. Algorithmically it looks something like

From ref time
Walk samples until the impulse is effectively zero (e.g. x sample moving average is within some of error of 0... I don't know the best way to capture that, one of those things that is obvious visually) == start reflection search
Continue until samples spike to " min reflection threshold "
Set right window to the value that reduces this to near zero

I don't think this has to be perfect, close is good enough. I could then set this as a default mode and know that I get a decent 1st cut quasi anechoic measurement. Fine tuning then comes later (manually)

I can supply some IRs if you like.

If by "detected ref time" you mean keep what the measurement's ref time currently is, I've done that for the next build.
Great, thanks.
 

·
Registered
Joined
·
261 Posts
Discussion Starter #4
As an aside, it is quite amazing how much you have put into this software for such a long period of time. I have been using it for ~10yrs now and the stream of features that have accumulated really sets it apart. Obviously you enjoy developing it but the fruits of your labour are certainly v much appreciated.
 

·
Registered
Joined
·
261 Posts
Discussion Starter #6
OK here's some examples. These are 2 pairs of measurements where pair means a measurement of a woofer & a tweeter with everything left the same except which driver plays.

The 1st pair (BMS4550 and TD12M) are indoor measurements, 1.1m measurement distance and ~1.2m to any boundary.
The 2nd pair (woofer and tweeter 1m) are outdoor measurements, 1m mic distance and at least 2m to each boundary but this is in SE London so not exactly quiet at the best of times :)

My comments;

BMS4550: arguably the correct 1st reflection is 7.45ms but it's tiny so you could reasonably decide to aim at the 8ms one instead to eke out a bit more frequency resolution. The 1st boundary reflection is pretty clearly identified by the spiky oscillation from an another pretty flat line (after the initial noise).
TD12M: this has a less obvious flat line after the initial impulse and a "softer" transition to the same initial reflection which makes it less clear where to set the window, nevertheless you can still see a larger/wider spike around the big reflection. This measurement is from the same conditions as the previous one.

Woofer: this has a fairly obvious reflection at ~7.5ms after an initial impulse that has "settled down"
Tweeter: this is much noisier and it indicates a reflection at about 3.7ms, there's not even anything obvious at 7.5ms. You could even argue for 2.25ms as a 1st spike. I would argue that an algorithm that spots ~1ms as a 1st reflection is probably flawed (the algorithm in holm does choose that point).
 

Attachments

1 - 6 of 6 Posts
Top