Thanks Wesayso. I feel like without seeing the response that needs to be corrected, I'm shooting in the dark. Sure I can look at the measured response in REW, but that is not what DRC is seeing. And this is of utmost importance around and below the room transition frequency where traditional gating does not work anymore and where the potential of frequency dependent windowing can be used to the maximum.
I'm just guessing here, do you want to output the windowed impulse to REW and export to RePhase by any chance?
Haha... yes, I was thinking of that, but DRC does have some psychoacoustic filters, correct? I might try those, or I might try correcting it in rePhase myself.
How does a windowed impulse work? I think what would need to be output is the frequency response itself after windowing, which can then be read by Holm or REW. It's the same process in Holm, where you can export the response after windowing. Now, we must get DRC to output it for us.
Someone could write the code for it in REW or Holm. I mean Holm does offer three separate windows. What's to stop someone from doing a continuously sliding window? If I had more time, and knew the language, I would write something myself. (Hint, hint, to others in this thread
). If Holm or REW can have this sliding window ability, then we would measure directly in REW/Holm and go to rePhase. I think I'll drop a line to pos, who is the author of rePhase.
Last edited:
Look at the MPPFOutFile variable in DRC. When you set that to output to file it probably has something similar to what you want to look at.
From DRC documentation:
MPPFOutFile - Output file for the minimum phase component after band windowing. Usually not used (commented out).
http://drc-fir.sourceforge.net/doc/drc.html#htoc77
Note that it has separate windows for the minimum phase filtering and excess phase.
Did you see the new smoothing window in the new REW beta? No idea what kind of windowing is applied yet but it uses a Psychoacoustic windowing filter, with dip limiting and some smoothing is being done. See: http://www.diyaudio.com/forums/full-range/273971-group-delay-questions-analysis-39.html#post4355384
From DRC documentation:
MPPFOutFile - Output file for the minimum phase component after band windowing. Usually not used (commented out).
http://drc-fir.sourceforge.net/doc/drc.html#htoc77
Note that it has separate windows for the minimum phase filtering and excess phase.
Someone could write the code for it in REW or Holm. I mean Holm does offer three separate windows. What's to stop someone from doing a continuously sliding window? If I had more time, and knew the language, I would write something myself. (Hint, hint, to others in this thread
Did you see the new smoothing window in the new REW beta? No idea what kind of windowing is applied yet but it uses a Psychoacoustic windowing filter, with dip limiting and some smoothing is being done. See: http://www.diyaudio.com/forums/full-range/273971-group-delay-questions-analysis-39.html#post4355384
Last edited:
I see, I think my trouble comes from the way I have to use DRC. I am on a Mac, and I can run DRC in Wine, but I can't just double click the .bat file and expect it to work.
So, instead, I copied the 4Cycles drc file into the strong drc file, and selected the strong filter in the Generate Standard Filters. It gave me a wav file.
From DRC documentation:
This looks like what we are after for looking at windowed IR.
So far so good, but test scripts don't give output, however, default file rtc.pcm is in 'samples', with right channel convolution result. Hmmm???
After setting PCOutfile to test.pcm (instead of rtc.pcm) in my template (removing the "#" in front of it) I successfully got output in the samples directory, a test.pcm file was created. I ran only a test script for my left channel to get it.
I could read that in REW. Looks like it had some work done on it though. Not a raw response of the windowed IR.
From Dennis Sbragion: rtc.pcm is the test convolution and should be used only to check the correction results.
Source here
Just tried to extract the rmppf.pcm output to REW and that looks like the windowed IR, without any work done. That's the one I think ra7 was looking for.
Edit: still not sure if this is a pure filtering of the impulse and nothing more. It looks better than I expected it to look.
Last edited:
Well, that was interesting.
I tried a few of the filters in Greg's (gmad) first post.
I have been using "Private investigations" from Dire Straits, because that track has a lot of dynamics, precise spatial information, and is clear enough to distinguish the different instruments, from guitar to piano, along xylophone and drums!
Mark Knoffler's voice is right up front and centre, with the main acoustic guitar, then all the other instruments are spread around.
Using the different filters, I had a few different experiences. Sometimes, Mark's voice would grow to be coming from everywhere and omnipresent, looking like he's 20 feet tall, and other filters would pan it to a side or the other. Differences in tonal shifts were also noticed, with some filters allowing more high frequencies through, and others, less.
Sadly, none of the filters were as good as the one I settled on a month or so ago. That one has everything I want to hear. Nice thundering drums, beautifully rendered vocals right in the centre, pinpoint accuracy of the different instruments around, and shimmering highs (but not fatiguing) on cymbals and presence.
It sounds like all would be perfect with that custom filter I came up with... but there's a hitch.
I forgot how I made it!
It was part of half a dozen filters I had made, testing the custom filters of DRCDesigner. I didn't take any notes, thinking I would remember... but here I am, a month later, and I forgot which settings I used with my "custom-3" filter!
And since I have tried a few others after making that one, I do not have a .bat nor .txt file showing me the windows and frequencies of that particular filter, since DRCDesigner overwrites the last one made.
So, I can only backup that filter, or try to reproduce the magic....
For now, I am just enjoying listening sessions!
I tried a few of the filters in Greg's (gmad) first post.
I have been using "Private investigations" from Dire Straits, because that track has a lot of dynamics, precise spatial information, and is clear enough to distinguish the different instruments, from guitar to piano, along xylophone and drums!
Mark Knoffler's voice is right up front and centre, with the main acoustic guitar, then all the other instruments are spread around.
Using the different filters, I had a few different experiences. Sometimes, Mark's voice would grow to be coming from everywhere and omnipresent, looking like he's 20 feet tall, and other filters would pan it to a side or the other. Differences in tonal shifts were also noticed, with some filters allowing more high frequencies through, and others, less.
Sadly, none of the filters were as good as the one I settled on a month or so ago. That one has everything I want to hear. Nice thundering drums, beautifully rendered vocals right in the centre, pinpoint accuracy of the different instruments around, and shimmering highs (but not fatiguing) on cymbals and presence.
It sounds like all would be perfect with that custom filter I came up with... but there's a hitch.
I forgot how I made it!
It was part of half a dozen filters I had made, testing the custom filters of DRCDesigner. I didn't take any notes, thinking I would remember... but here I am, a month later, and I forgot which settings I used with my "custom-3" filter!
And since I have tried a few others after making that one, I do not have a .bat nor .txt file showing me the windows and frequencies of that particular filter, since DRCDesigner overwrites the last one made.
So, I can only backup that filter, or try to reproduce the magic....
For now, I am just enjoying listening sessions!
Perceval,
In the end, all that matters is that you find a filter that makes you happy. With my package, in the beginning at least, try to keep the correction window 5 cycles or less to better get a sense of what is being corrected (this should really be enough correction for long term enjoyment anyway imho). Also, I highly recommend using my files in the manner I outlined in my guide (which would require a PC).
As I remember it, there were a couple of issues with your system that needed special attention:
-Firstly, the impulse centers of your measurements were not the loudest points of the waveforms. I have tried to explain how to diagnose and deal with this problem in my guide, and unless this is addressed, the correction may be faulty.
-Secondly, there was a channel balance issue. Even still, I have found that the channel balance always needs adjustment after the filters are generated (also outlined in my guide) as DRC does not do this part automatically, and that could explain the panning effects you're experiencing while comparing filters.
If you want to send my your mdat, I could make a filter for you to see what differences there are...
In the end, all that matters is that you find a filter that makes you happy. With my package, in the beginning at least, try to keep the correction window 5 cycles or less to better get a sense of what is being corrected (this should really be enough correction for long term enjoyment anyway imho). Also, I highly recommend using my files in the manner I outlined in my guide (which would require a PC).
As I remember it, there were a couple of issues with your system that needed special attention:
-Firstly, the impulse centers of your measurements were not the loudest points of the waveforms. I have tried to explain how to diagnose and deal with this problem in my guide, and unless this is addressed, the correction may be faulty.
-Secondly, there was a channel balance issue. Even still, I have found that the channel balance always needs adjustment after the filters are generated (also outlined in my guide) as DRC does not do this part automatically, and that could explain the panning effects you're experiencing while comparing filters.
If you want to send my your mdat, I could make a filter for you to see what differences there are...
Yep,
I got most of it fixed.
The impulse centers... that was a back wall reflection (cement wall right behind the listening spot).... fixed that with some absorption on the back wall.
At this moment, there are no channel imbalance. Not sure what happened in the previous tries, but right now, it sounds pretty sweet.
I need to redo my whole setup. I seem to have lost my left/right/stereo initial mdat file. The trouble with doing 5 projects at the same time... I have a terrible memory, and a tendency to lose things. I only have the resulting sweep I got from my latest "custom-3", which looks pretty sweet (included below) which does not take into account my two subs taking care of the low end of things.
Good news is I have the house to myself tomorrow morning. I'll try to get the whole thing done again, and work from that... and this time, keep copies of my progress! But I'm not throwing away that "custom-3" filter... that thing is pure gold! That will be the base to judge any other filters.
I still have the L/R pcm files I use in DRC though... would that be ok? No way to tell how new filters would work though...
I got most of it fixed.
The impulse centers... that was a back wall reflection (cement wall right behind the listening spot).... fixed that with some absorption on the back wall.
At this moment, there are no channel imbalance. Not sure what happened in the previous tries, but right now, it sounds pretty sweet.
I need to redo my whole setup. I seem to have lost my left/right/stereo initial mdat file. The trouble with doing 5 projects at the same time... I have a terrible memory, and a tendency to lose things. I only have the resulting sweep I got from my latest "custom-3", which looks pretty sweet (included below) which does not take into account my two subs taking care of the low end of things.
Good news is I have the house to myself tomorrow morning. I'll try to get the whole thing done again, and work from that... and this time, keep copies of my progress! But I'm not throwing away that "custom-3" filter... that thing is pure gold! That will be the base to judge any other filters.
I still have the L/R pcm files I use in DRC though... would that be ok? No way to tell how new filters would work though...
Attachments
Last edited:
Left/right pcm files are fine as long as they're representative of your current setup. Also, tell me if there is a particular target frequency response you're shooting for.
If the severe channel imbalance I'm remembering was some sort of measurement error, there is still the issue of the different filters skewing the balance one way or the other, at least slightly (although it's possible to just luck out in this regard).
If the severe channel imbalance I'm remembering was some sort of measurement error, there is still the issue of the different filters skewing the balance one way or the other, at least slightly (although it's possible to just luck out in this regard).
I believe these are the pcm files I used. I have two sets, but I think these are the right ones.
https://www.dropbox.com/s/qycui73f6bfk28a/pcm files.zip?dl=0
https://www.dropbox.com/s/qycui73f6bfk28a/pcm files.zip?dl=0
I know that happened to me a couple of times. Right now if I really enjoy the sound I get I rename the filter to the settings I used in DRC Designer and include an image of my target curve under the same name.
Yes, I still use DRC Designer. With a template based on Greg's set posted in the first post of this thread copied as soft44100.drc. I changed a couple of things though after some experimenting with previews of the results in REW.
I'll upload it for people to try it...
This isn't a linear phase template though, this is minimum phase all the way and should work fine for single full range speakers.
I tried the REW psychoacoustic smoothing and it is giving me strange results. I see a comb filter response all the way from 20 kHz. Unless there is a strong early reflection, and by early I mean really early, I don't see how the response is correct. ERB and variable smoothing give more reliable results.
I'll try gmad's DRC guide next when I get time.
I'll try gmad's DRC guide next when I get time.