Impulse Response - How to?

[jsalinasii]

I have been using the RTA features on REW for a couple of years now and feel comfortable doing that. One thing that I am having a very difficult time understanding or performing is a time alignment check using impulse responses in REW. Quick and dirty I simply don't know how to set it up or perform it. I am simply looking for a second verification that my tape measured DSP delays are correct.

I have many questions and I don't want to waste any persons time but the way I understand it is I send a signal from my computer output (toslink, or analog) through my system, I have my UMIK-1 set at signal point and that mic is supposed to pick up the first signal. Since the computer is sending the signal it has a reference as to when it was sent and the microphone knows when it was received (pitcher/catcher scenario). What I don't understand is why the numbers I receive from REW are so large ie 20 MS of delay for a car cab. I am clearly doing it all wrong so:

1. Do I use the loopback as timing selection or something else?
2. Do I use a frequency range of like 1 hertz?
3. Can I even do this with only a USB mic and my laptop's output?

Any help is appreciated.

 

[jsalinasii]

Typo: I have my UMIK-1 set at a SINGLE point/location

 

[John Mulcahy]

To make timing measurements you need a timing reference. With a USB mic that requires REW's acoustic timing reference, but it may not be reliable in a car as there are so many strong reflections. An alternative for aligning the individual drivers for a channel is to look at the wavelet spectrogram, as discussed in this post.

 

[jsalinasii]

John,

First great name. Second thank very much for taking the time to respond. To best use wavelet I watched the video and I think I set it up the way Matt has it setup. I have crossovers at 5500, 200, and 70 hertz. Here is the wavelet graph I get from my previous all speakers firing sweep.

It appears to me (who knows nothing about this) that the upper crossover at 5500 is very close in time but my 200 hz crossover point and then sub are NOT right. Is this a correct understanding of that wavelet graph? If the peak energy time line goes to the positive numbers does that mean the frequency range is OVER or UNDER delayed (arriving late or early)?

Thanks,
John

 

[John Mulcahy]

Looks pretty good until 200 Hz or so, easier to see if you select the option to normalise the response. Positive numbers show the peak energy is taking longer to arrive, so needs to be made earlier or everything else needs to be delayed. It is normal to see a general curve to the right as the low end rolls off though, and audibility of delays down there seems limited.

 

[IWC Doppel]

John,

First great name. Second thank very much for taking the time to respond. To best use wavelet I watched the video and I think I set it up the way Matt has it setup. I have crossovers at 5500, 200, and 70 hertz. Here is the wavelet graph I get from my previous all speakers firing sweep.

It appears to me (who knows nothing about this) that the upper crossover at 5500 is very close in time but my 200 hz crossover point and then sub are NOT right. Is this a correct understanding of that wavelet graph? If the peak energy time line goes to the positive numbers does that mean the frequency range is OVER or UNDER delayed (arriving late or early)?

Thanks,
John

Could you reference the video I'm still learning and determined to become usefully knowledgeable on all of this !

 

[John Mulcahy]

The video is in this post.

 

[jsalinasii]

John,

I use the REW PN Generator for consistency in my measurements. Is there a way within it to flip between correlated (MONO) pink and uncorrelated (stereo) pink noise? I am being told to measure my pair speakers using stereo PN and mono PN for individual speakers but I don't know how to do it in REW.

Thank you again.
John II

 

[John Mulcahy]

Afraid not, REW does not have an uncorrelated noise signal. I'm not clear what useful information would come from measuring a speaker pair with uncorrelated noise signals, perhaps someone could enlighten me?