REW eq to VST plug-in

[bowl_actually]

Thanks ola1 for taking the time to put all that together. I am very experienced in all the details of building a proper studio, including room treatment, monitoring solutions, etc. So I am not trying to take a cheap pair of speakers in a small bedroom and try to make them sound like JBL M2s in a world class facility

I actually have two objectives here, first, even if you have JBL M2's in a world class studio, what you will find if you shoot the room with REW, is that there are still small errors. These can very effectively be treated with room correction. I currently do this with DIRAC in my studio. By turning off the DIRAC and using a VST plugin, I can effectively determine if the VST is working properly.

If I can make the VST/Convolution Reverb work properly, then I can finally get to my final objective.

My final objective is to develop many IR's that represent many speaker systems. Small near field monitors, home speakers, maybe some car audio systems, etc. Many of the speakers frequency response curves can be found online. So I can eventually find a way to accurately turn those into IRs, then I can put a plugin on the output of my DAW and "simulate" the sound of other speakers.

As a mixing engineer, this would allow me to make better decisions. If I can find an accurate way to generate an IR that represents the EQ curve I want, and accurately implement on a VST plugin, then I can do this.

Actually I spent much of the day yesterday working on this. I would say at the moment I'm getting very close to making this work very well. Thanks for teaching me the basics of this method, I think over time it's going to prove very valuable for me.

 

[Michael Boeker]

True, except if you use an ASIO interface, which all the pro interfaces are, then my understanding is that it gets bypassed as ASIO talks from your DAW direct to the hardware by passing the OS an all those tools.

Not to mention I want it on my DAW bus so I can easily turn it on/off inside my DAW as I'm mixing.

JRiver supports both ASIO input and output. If you have an audio device with an ASIO multi client driver, then it will work. It works fine with my 3 MOTU and 2 RME audio devices. This method supports 32 channels.

You can activate JRiver as a WDM driver, which is essentially the same as hardware direct. This lets the DAW select JRiver as the soundcard and then JRiver outputs via ASIO to hardware. This is limited to 8 channels.

 

[bowl_actually]

JRiver supports both ASIO input and output. If you have an audio device with an ASIO multi client driver, then it will work. It works fine with my 3 MOTU and 2 RME audio devices. This method supports 32 channels.

You can activate JRiver as a WDM driver, which is essentially the same as hardware direct. This lets the DAW select JRiver as the soundcard and then JRiver outputs via ASIO to hardware. This is limited to 8 channels.

Ah, that's great to know, thanks for that info. I use RME exclusively in my studio, UFX for the studio and BabyFace for remote recording, so I need to go look into JRiver. I've never used it before, but have heard many great things about it.

 

[Michael Boeker]

By the way, JRiver also supports convolution files and you can setup Zones that simulate other speakers with each Zone having a different convolution file loaded.

 

[bowl_actually]

By the way, JRiver also supports convolution files and you can setup Zones that simulate other speakers with each Zone having a different convolution file loaded.

Oh wow, that does sound pretty nice, might be better than a VST, although have to go outside my DAW to see what's applied. Maybe I can handle that, you know how audio guys are, I'm a bit forgetful some times.

 

[bowl_actually]

By the way, JRiver also supports convolution files and you can setup Zones that simulate other speakers with each Zone having a different convolution file loaded.

So looking at their web site, I assume it's the JRiver media center? Unless I'm looking at the wrong thing, it looks like that's all they make? At $59 I'll get it just to see what it's all about.

Thanks.

 

[bowl_actually]

Just a follow up here in case anyone else is interested in this. First, I gave JRiver a try, not at all what I wanted and really not something to use for a recording studio. Great media server, with lots of great features, but it's really for a different purpose than a recording studio.

I did actually find exactly what I was looking for in a product called "Hang Loose Convolver" from Accurate Sound that is a VST3 plugin. It allows you to install up to 6 different IRs and seamlessly switch between them without and pop/glitches.

With the latest version of REW, you can easily export an IR either from a filter or from a frequency response curve. This easily allows you to build an IR of a speakers frequency response. That IR can then be loaded into Hang Loose as a VST3 plugin inside your DAW.

It works flawlessly and I have now implemented multiple "speaker models" into my DAW.

 

[moedra]

I know the last post here was a few months shy of a year ago, but if any of you or anyone you know are still looking to use correction filters generated with REW, contact me. I have been working on this subject for two years and I am looking for anyone interested in helping me test my filters. All you need to do is take a L/R pair of speaker measurements from your main listening position and post the mdat file here.

Alternatively, if you are using Windows, you can sign up for Focus Fidelity's newsletter and get free access to Impala. Impala is fantastic at taking measurements. It's fast and efficient, and it can measure both speakers in one pass with clock drift correction. You can export your measurements from Impala as .wav and those can be imported into REW for processing. You can either post the exported wav files or the Impala ffm file and I can export the measurements myself.

 

[JStewart]

I know the last post here was a few months shy of a year ago, but if any of you or anyone you know are still looking to use correction filters generated with REW, contact me. I have been working on this subject for two years and I am looking for anyone interested in helping me test my filters. All you need to do is take a L/R pair of speaker measurements from your main listening position and post the mdat file here.

Alternatively, if you are using Windows, you can sign up for Focus Fidelity's newsletter and get free access to Impala. Impala is fantastic at taking measurements. It's fast and efficient, and it can measure both speakers in one pass with clock drift correction. You can export your measurements from Impala as .wav and those can be imported into REW for processing. You can either post the exported wav files or the Impala ffm file and I can export the measurements myself.

I'd welcome a go at this, but I can't take a measurement until Friday. Is that Ok?

 

[jtalden]

I know the last post here was a few months shy of a year ago, but if any of you or anyone you know are still looking to use correction filters generated with REW, contact me. I have been working on this subject for two years and I am looking for anyone interested in helping me test my filters. All you need to do is take a L/R pair of speaker measurements from your main listening position and post the mdat file here.

Hi,
Thanks for that generous offer. It's an opportunity to compare another EQ solution to my own.
Attached is an LP measurement of my FL and FR channels with my current EQ removed. I also included my current house curve is case you want to use that. If you have another you prefer to use that is fine also.

If this is an FIR filter solution it would be helpful to provide filters at both 44.1 and 96 kHz sample rates. I do all measurement and analysis at 96kHz but use 44.1kHz for my music files.

 

[JStewart]

I know the last post here was a few months shy of a year ago, but if any of you or anyone you know are still looking to use correction filters generated with REW, contact me. I have been working on this subject for two years and I am looking for anyone interested in helping me test my filters. All you need to do is take a L/R pair of speaker measurements from your main listening position and post the mdat file here.

Alternatively, if you are using Windows, you can sign up for Focus Fidelity's newsletter and get free access to Impala. Impala is fantastic at taking measurements. It's fast and efficient, and it can measure both speakers in one pass with clock drift correction. You can export your measurements from Impala as .wav and those can be imported into REW for processing. You can either post the exported wav files or the Impala ffm file and I can export the measurements myself.

I found some time after all. (Have to keep the priorities straight)
Thanks again!
Be interesting to compare to Focus Fidelity filters. I use a 0.6dB/Octave Slope for the house "curve".

 

[moedra]

I'd welcome a go at this, but I can't take a measurement until Friday. Is that Ok?

Whenever you can. No rush, no problem.

 

[moedra]

Hi,
Thanks for that generous offer. It's an opportunity to compare another EQ solution to my own.
Attached is an LP measurement of my FL and FR channels with my current EQ removed. I also included my current house curve is case you want to use that. If you have another you prefer to use that is fine also.

If this is an FIR filter solution it would be helpful to provide filters at both 44.1 and 96 kHz sample rates. I do all measurement and analysis at 96kHz but use 44.1kHz for my music files.

My prefferred curve is just a sloped linear target. I will use that and when I get time I'll use your curve as well so you can compare things.
REW works internally at 48kHz. I will use what you have here and we'll see what happens. You may need to measure at 48kHz, though. Not sure about this matter.

 

[moedra]

I found some time after all. (Have to keep the priorities straight)
Thanks again!
Be interesting to compare to Focus Fidelity filters. I use a 0.6dB/Octave Slope for the house "curve".

Oh ok. Nice. I have Focus Fidelity also. If you can, try sending me Impala measurements also. If you can align the pulses when measuring, the results often come out better. Check the preview tab for impulse response. If the left pulse is before the blue one, inch your mic a little to left and try again. Same goes if the blue pulse is before the green one. This means the mic is closer to the left speaker and you should inch it to the right and remeasure. Check it until you have them aligned/overlapping, as in my screenshot.

 

[JStewart]

Oh ok. Nice. I have Focus Fidelity also. If you can, try sending me Impala measurements also. If you can align the pulses when measuring, the results often come out better. Check the preview tab for impulse response. If the left pulse is before the blue one, inch your mic a little to left and try again. Same goes if the blue pulse is before the green one. This means the mic is closer to the left speaker and you should inch it to the right and remeasure. Check it until you have them aligned/overlapping, as in my screenshot.
View attachment 58750

Wave files from Impala attached along with export log.

Left/Right Impulse Alignment:from same.

Thanks again!

 

[jtalden]

My prefferred curve is just a sloped linear target. I will use that and when I get time I'll use your curve as well so you can compare things.

Just go ahead and use your normal target. I can probably adjust I little as needed for direct comparison.

REW works internally at 48kHz. I will use what you have here and we'll see what happens. You may need to measure at 48kHz, though. Not sure about this matter.

The recent early release beta versions of REW provides a popup to select the sample rates desired for the filters after "Export filters impulse responses as wave" is selected.

 

[ddude003]

Measurements should be taken in the full range of 0Hz to 24,000Hz (Nyquist one-half the 48kHz sampling rate) for REW's accurate vector math calculations, regardless of the final selectable output filter .wav(s) file(s) sample rate(s)...

 

[moedra]

Hi,
Thanks for that generous offer. It's an opportunity to compare another EQ solution to my own.
Attached is an LP measurement of my FL and FR channels with my current EQ removed. I also included my current house curve is case you want to use that. If you have another you prefer to use that is fine also.

If this is an FIR filter solution it would be helpful to provide filters at both 44.1 and 96 kHz sample rates. I do all measurement and analysis at 96kHz but use 44.1kHz for my music files.

OK looking at your measurements... what is it that you've done with the trace arithmetic operations? I'm not sure that's necessary. All I need are raw sweeps from each speaker...

 

[jtalden]

OK looking at your measurements... what is it that you've done with the trace arithmetic operations? I'm not sure that's necessary. All I need are raw sweeps from each speaker...

The 2 provided impulse measurements are equivalent to a simple measurement of FL and FR channels. They were calculated from individual measurements of; SWgroup + FLmidwoofer + FLtweeter for the left channel and SWgroup + FR midwoofer + FRtweeter for the right channel. I don't normally measure without any EQ or at this high a level but I had already done this on the individual driver sections recently. I have confirmed the SPL, phase and other responses are consistent with my other full channel measurements. I thought the extra 8 dB of SPL would be a good idea. I can make new full channel measurements for you if this is a concern.

 

[moedra]

Wave files from Impala attached along with export log.

Left/Right Impulse Alignment:from same.

View attachment 58751

Thanks again!

Alright try these out. The only thing concerning me at the moment is whether or not your calibration file made it into the exported measurements. If it didn't, and since I don't have your cal file, these filters will not be accurate. Let me know what you find out. Measure them in REW with your cal file in place and let's see if we nailed it. The measured response should be flat.

 

[moedra]

The 2 provided impulse measurements are equivalent to a simple measurement of FL and FR channels. They were calculated from individual measurements of; SWgroup + FLmidwoofer + FLtweeter for the left channel and SWgroup + FR midwoofer + FRtweeter for the right channel. I don't normally measure without any EQ or at his high a level but I had already done this on the individual driver sections recently. I have confirmed the SPL, phase and other responses are consistent with my other full channel measurements. I thought the extra 8 dB of SPL would be a good idea. I can make new full channel measurements for you if this is a concern.

I'd try that, yes. Also the sweeps only need to be between 75dB and 80dB SPL in the room, which is loud but not peircing. Also the bass should not distort. You want nice clean sweeps. When they measure into REW, that level should match on the magnitude scale to the left.

 

[JStewart]

Alright try these out. The only thing concerning me at the moment is whether or not your calibration file made it into the exported measurements. If it didn't, and since I don't have your cal file, these filters will not be accurate. Let me know what you find out. Measure them in REW with your cal file in place and let's see if we nailed it. The measured response should be flat.

Many thanks!

Attached are Left/Right measurements with and without filters applied.
The 48kHz filter was used in Roon. Used REW 's Measure with File Playback method with timing reference.
Mic has been, and was for these measurements, in same location since the original measurements used for filter generation were taken using Impala.

Now that this is done, I'm going to move the mic out of the way and have a listen

 

[moedra]

Many thanks!

Attached are Left/Right measurements with and without filters applied.
The 48kHz filter was used in Roon. Used REW 's Measure with File Playback method with timing reference.
Mic has been, and was for these measurements, in same location since the original measurements used for filter generation were taken using Impala.

Now that this is done, I'm going to move the mic out of the way and have a listen

Oh it looks alright. Seems as though I was concerned for no reason. That should sound good.

Remember you can pack those filters up into a .zip archive and load the zip file directly into Roon. It should then switch to the appropriate filter based on playback sample rate.

Also, you can measure filters directly with Impala just as you did with REW. When you begin a new measurement file you can load a filter into the first parameter page.

 

[moedra]

Just go ahead and use your normal target. I can probably adjust I little as needed for direct comparison.

The recent early release beta versions of REW provides a popup to select the sample rates desired for the filters after "Export filters impulse responses as wave" is selected.

Let's see how this goes. I am curious as to how these compare to filters made from full-range sweeps.

 

[jtalden]

Let's see how this goes. I am curious as to how these compare to filters made from full-range sweeps.

Thanks - Job well done.
Below are a couple of charts to show the impact of these new filters on new LP measurements of FL and FR at my LP. I went ahead and did the new measurements to show there would be only the minor expected differences seen between the calculated responses you worked on and these new measurements. There has been 2 or 3 months difference between them, and the mic position was no doubt in a slightly different position. The variation found was negligible in my opinion, 1 or 2 dB in several spots. I then applied my soundcard cal and mic cal to the impulses so they would also be included. This only impacts the extreme ends of range a little bit and accounts for a 2-3 dB SPL difference around 10k due to the broad mic peak in that area.

The charts show the main findings. The SPL was effectively smoothed out to a house curve very near what I would target. The phase rotation was only impacted a little by the effect of PEQ EQ filters and shows the minor differences at the extremes that reflect the cal files I added. There is no indication that you applied any FIR phase correction even in the higher frequencies. Several of us are doing that with help of rePhase, but its practical value is questionable. I think you were wise to address the large SPL issues and not try to correct for all the minor fluctuations. My working file is quite large, so I just attached the new measurements and the final calculated result, so you better review the differences in more detail. You can also import the original calculated mdat file you worked on to compare the differences resulting from these new measurements.

I will also try to listen to them, but it won't mean much because I will likely assume any differences are just due to the different house curve. This new filter set is bound to be noticeably brighter than I am used to.