Why is REW using pic of impulse for time reference

[Dominique-Tanguy]

I am quite a long time user of REW, that I like very much...

I am using a Tascam which is very convenient for measurement with loopback and time reference.

My question is related to the information provided by REW in relation with "flight time" and distance. REW is using the pic of the impulse while most of my audiophile colleagues are using the foot or beginning of the impulse as time reference.

This make quite a significant difference in term of physical alignement (more or less 10 cm) between the woofer and the 1505B horn.

Listening the two configurations, I tend to prefer the configuration aligned on the pic, but I am interested to here your opinions on these two options.

Kind regards,

Dominique

 

[Bernard]

Hello,
Which version of REW are you using?
The last one is V5.20 beta 22.

 

[Dominique-Tanguy]

Hello,
Which version of REW are you using?
The last one is V5.20 beta 22.

No, I am using the V5.19.

Had some stability issues with the beta.

Dominique

 

[John Mulcahy]

With the acoustic timing reference REW uses either the estimated IR delay or the IR start time, depending on the time from the IR start to the IR peak. If that time is short (less than 1 ms) the estimated delay is used.

With a loopback timing reference the peaks of the measurement and loopback are used, partly because the loopback gets the same signal as the measurement and the location of the peak depends on the measurement bandwidth. At some point I'll implement something more like the acoustic timing reference.

 

[Dominique-Tanguy]

Thank you so much for taking the time to answer... So for my better understanding, this "acoustic timing reference" could be located at The beginning of the impulse, or just slightly ofset from the pic?

For exemple, first illustration, measurement with Time reference on the pic at 15,312:

And position of the time reference after shifting the IR, the Time reference is not on the pic anymore:

Of course, the effect is more spectacular with the low frequencies...


Dominique

 

[John Mulcahy]

Yes, estimate IR Delay uses cross correlation with a minimum phase version of the response, if the response bandwidth is close to half the sample rate the t=0 time will be very close to the peak.

 

[Dominique-Tanguy]

It tooks me stupidely many years to realize that these two values are different: