Exporting IR wav's

bowl_actually

Member
Thread Starter
Joined
Mar 4, 2020
Messages
29
When you export an EQ IR, it pops up a dialog allowing you to define the bit rate and sample frequency of the export.

However, if you export a measurement IR, it gives you a dialog for the bit depth but not a dialog for the sample frequency. For a measurement that was acquired through REW, it looks like it exports as the sample frequency of whatever it was when it was captured. So if it was captured at 48Khz (maybe by someone else), even though your machine and your REW preferences are set to 44.1K, it still exports at 48Khz.

Also, if you import a frequency graph from a TX file, with columns of frequency and SPL, then generate a minimum phase for that graph, then export the IR wav. It seems to export as an 88.2Khz sampling rate. I can't find any way to change that. My machine is set to 44.1Khz and my preferences in REW are set to 44.1Khz. There is no information in the imported text file regarding either bit depth or sample frequency, it's just two columns with frequency and dbSPL.

Is there any way to set the export sample rate? If not, a feature request would be to add the same dialog like there is in the EQ IR export to allow you to set both the bit depth and the sample frequency.
 

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
7,212
The export rate for measurements is the rate used when the measurement was made, so the data is as it was acquired. Minimum phase responses are generated at a higher sample rate to reduce the errors at high frequencies. I can look at resampling the response for IR export to allow other sample rates.
 

bowl_actually

Member
Thread Starter
Joined
Mar 4, 2020
Messages
29
Thanks John. Yea I'm using REW for some unconventional things. Let me explain.

Of course REW, the awesome program, was written for analyzing and correcting room acoustics. But over the years it has evolved to serve many uses. Of course speaker builders are now using it to measure drivers, crossovers, box resonances, and overall speakers.

But the goal of all of these uses is to "correct" things to be as close to flat or some house curve. To that end, it can create filters to correct a response, export those filters as IRs to be used in DSP or other devices.

While all this was going on, in the music industry, there has been a revolution in "modeling" things that attempts not to make things be flat or "corrected" but to make something sound like a known "cool" thing.

The way this is done is by taking an impulse response of the "cool" thing and using it to impart it's sound on something else. So instead of taking all the flaws of something and trying to correct them, we want to capture those flaws and use them.

A major use of this has been in the Guitar world, where you make an IR of a guitar amp or amp cabinet or combination. In the recording world you sometimes record "direct in" which means you just plug the guitar directly into the recording interface. But what if you then wanted that recording to sound like it was played through a Marshall amp with a 4 speaker cabinet? Just use the IR and load it into a convolution engine that adds that as an effect.

There are several programs out there to capture these IRs, one of the most used is Voxengo Deconvolver.

But I have learned that REW might just be the best IR capturing program out there! It's actually easier than the room correction, just measure an amp or speaker cabinet with REW and export the IR.

But what if you don't actually have the cabinet you want to measure? If you find a frequency response graph of that cabinet online, there are programs to trace the graph, create a txt frequency response graph, import that into REW, create a minimum phase plot, and export it as an IR. Of course it's not as good as being able to measure it, but it's better than you might think.

So in this instance, exporting the IR of the measurement or even an imported frequency graph is what I'm using. And for the most part it's brilliant!! Thanks.
 
Top Bottom