REW, Room Correction, Signal Generation

I feel like this should be obvious, but I am stumped. I am trying to use REW to do some room correction of some wireless Roon ready speakers (Devialet Phantom II’s). The only real way to get a signal to them is wirelessly via roon. So, any tips on how to generate the test signal in REW and send it to the speakers to be measured?

1 Like

The Generator and Measure features of REW allow saving test signals as files. Do so, then load the file into Roon, then play. For Measure, be sure to add a timing signal to the sweeps.


Here’s a video I created recently that demonstrates how to do this.


Thanks. I found a second (non-free, and non-generalizable) solution which is the Devialet Arch. It takes spdif (or analog) input and streams it to the speakers. I connected to it via a usb->spdif box and that worked fine.

1 Like

This may not be according to Hoyle but I did something similar. I saved pink noise generated with REW and played it back with Roon. Then I used REW’s RTA (real time analyzer) and captured a minute or two — then I generated filters for each sample frequency and generated a convolution filter.

When I played the 20-20k pink noise with the filter active …it was pretty flat.

Then I repeated the exercise but this time I used a sweep but only 10-600Hz. . I thought the pink noise would be better since you can move the mic around and the sweep is a fixed location. But switched between the two Roon convolution filters and can’t tell the difference, but both are better than no filter

I played from files since I have a RP4/Ropieee as the endpoint, it only shows up as Airplay for REW output, I suppose I could bypass the RP4 just for the room eq and go straight into the DAC. Maybe next time. When I used Airplay the volume was significantly reduced

But in the end, it did a pretty good job of “fixing” the room nodes

1 Like

Pretty cool. I’m not surprised by the results, but I’d be curious to know how your changes affected time-domain performance. If you can use the sweep approach that I shared and record sweeps before and after engaging your filters, we should be able to look at the Step response (%) to get a sense of what’s going on.

I did record 10-600Hz sweeps before and 20-20k after…but I don’t know what Step response % is.

Are you suggesting I repeat the sweep recording with 20-22k as in your video?