Ok, didn't have time to make any measurements, but had a chance to do a quick listening session.
I used the new REW sweeps to do a left and right PCM file for DRC Designer and produced a bunch of filters, to compare them with gmad's tweaked filter.
Firstly, the sound of my own filters got so much better, pretty much in the same ball park as gmad's filter, although mine were completely flat from 20 to 20,000.
Here's the thing though....
With my own filters, I can clearly hear the left and right speakers arriving at different times. On a nice DSD recording, I hear the snare hit arriving at different times, creating a double hit. Plus, there's a side shift in intensity to the left.
With gmad's aligned filter, everything got back to the center and the sounds are sharp, arriving at the same time.
What puzzles me is the time shift between the Left and Right speakers. There is the exact same length of cables throughout the whole signal path between the two. Could it be my sound interface? I have to hook up things in a different way to find out.
I used the new REW sweeps to do a left and right PCM file for DRC Designer and produced a bunch of filters, to compare them with gmad's tweaked filter.
Firstly, the sound of my own filters got so much better, pretty much in the same ball park as gmad's filter, although mine were completely flat from 20 to 20,000.
Here's the thing though....
With my own filters, I can clearly hear the left and right speakers arriving at different times. On a nice DSD recording, I hear the snare hit arriving at different times, creating a double hit. Plus, there's a side shift in intensity to the left.
With gmad's aligned filter, everything got back to the center and the sounds are sharp, arriving at the same time.
What puzzles me is the time shift between the Left and Right speakers. There is the exact same length of cables throughout the whole signal path between the two. Could it be my sound interface? I have to hook up things in a different way to find out.
So it did sound better this time? About that shift gmad applied, I hardly think it would be enough to create a double hit sound. It could just be your mind playing tricks on you. Sometimes all your brain needs is knowing it is there to hear it. It is an awfully small shift. Gmad's filter might be more gentle than most of the filters you generated. Less forced, 4 cycles correction from high to low is a relatively light correction.
Though your filters are flat DRC uses the boost/cut limit to extend some of the low frequencies but it won't make it flat to 20 Hz. Your speakers are ported right? That high pass gmad put in will protect from over excursion.
Hope you can measure the result soon so we can all see what's going on. Isn't it fun that we can solve this while we are all so far apart geographically?
Though your filters are flat DRC uses the boost/cut limit to extend some of the low frequencies but it won't make it flat to 20 Hz. Your speakers are ported right? That high pass gmad put in will protect from over excursion.
Hope you can measure the result soon so we can all see what's going on. Isn't it fun that we can solve this while we are all so far apart geographically?
Yes, I didn't get the crazy wall bricking frequencies I had with the first run.
But it is not my mind playing tricks about the double hit.
There is something going on and I can only guess the whole process just multiplies the problem...
I will record the resulting sound, you will definitively hear like an echo of each hits. It's very obvious with the snare's attack, and even with a fill of toms, it just sounds fuzzy.
Kinda like hitting the dsp "Hall" effect on a cinema receiver.
And yes, the internet is a mixed bag of blessings and a terrible time wasting place! 🙂
I love the fact that we can find info and help each other the way we are now... but sometimes, I have information overload! 🙂
But it is not my mind playing tricks about the double hit.
There is something going on and I can only guess the whole process just multiplies the problem...
I will record the resulting sound, you will definitively hear like an echo of each hits. It's very obvious with the snare's attack, and even with a fill of toms, it just sounds fuzzy.
Kinda like hitting the dsp "Hall" effect on a cinema receiver.
And yes, the internet is a mixed bag of blessings and a terrible time wasting place! 🙂
I love the fact that we can find info and help each other the way we are now... but sometimes, I have information overload! 🙂
oh boy!
We are getting deeper and I feel I am running after my tail!
So, using gmad's filter seems to clean up the sound and center it. Great! something to build on... so I thought...
Decided to fire up Star Trek 2009, as I know the intro very well, using it to test changes to the sound system.
What cleaned up the "echos" in normal audio listening, now is introducing "echos" in movie mode. The soundtrack seems to be duplicated and not in sync.
Will need to spend some quality time with my system to try and pin it down, as of right now, I have no idea.
Maybe I should start another thread, as we are veering quite far the wonderful arrays you built, wesayso. Or do you want us to stay on track here?
We are getting deeper and I feel I am running after my tail!
So, using gmad's filter seems to clean up the sound and center it. Great! something to build on... so I thought...
Decided to fire up Star Trek 2009, as I know the intro very well, using it to test changes to the sound system.
What cleaned up the "echos" in normal audio listening, now is introducing "echos" in movie mode. The soundtrack seems to be duplicated and not in sync.
Will need to spend some quality time with my system to try and pin it down, as of right now, I have no idea.
Maybe I should start another thread, as we are veering quite far the wonderful arrays you built, wesayso. Or do you want us to stay on track here?
The delay is there... convolution can take up some time, though very little. You play the movie with JRiver? You can set the timing of audio video in the Video settings under advanced. Playback options -> Video -> Advanced.
I agree we're drifting further away but I have no problem solving this first. It might encourage others to try out/play with DRC.
Do you use JRiver's JRSS mixing from multiple channels to stereo? The two speakers are the only ones playing? These are the Tangband speakers right?
By the way, I use a lip sync video with a spinning beam and click sound to get proper sync of my audio/video play. I guess I got that one from a Disney movie setup disk. This one might work if you can download it: https://www.youtube.com/watch?v=bCPEidaVzQU
I agree we're drifting further away but I have no problem solving this first. It might encourage others to try out/play with DRC.
Do you use JRiver's JRSS mixing from multiple channels to stereo? The two speakers are the only ones playing? These are the Tangband speakers right?
By the way, I use a lip sync video with a spinning beam and click sound to get proper sync of my audio/video play. I guess I got that one from a Disney movie setup disk. This one might work if you can download it: https://www.youtube.com/watch?v=bCPEidaVzQU
Last edited:
I am using JRiver for the video as well... for now.
Had a beer, settled down and fired it up again.
If I am in 2.1 (my 2 TB up front), it sounds fine, if the track is 5.1, the center and surrounds engage, and I hear the huge delay in the surrounds that's causing the echo.
The convolution, does it only happen in the front two speakers? Can it be used for 5.1 surround?
Had a beer, settled down and fired it up again.
If I am in 2.1 (my 2 TB up front), it sounds fine, if the track is 5.1, the center and surrounds engage, and I hear the huge delay in the surrounds that's causing the echo.
The convolution, does it only happen in the front two speakers? Can it be used for 5.1 surround?
Yes, you can use convolution for a 5.1 system. You have to look up how on the JRiver forums though. I have no experience with that (yet).
I do not get the JRiver for video "for now" comment. I just love that part of JRiver. If you use Red October HQ at least. It's rendering is done by MadVR if you choose the HQ playback. I played and tinkered with it extensively and love the picture quality I was able to get. I love it so much it has become my player for all video sources and had me abandon separate players (my DVD player has done nothing since). It's very powerful with the right settings for your hardware and works very well for me. Combined with all audio tools it is what made JRiver worthwhile for me.
It takes some time to properly setup MadVR for maximum quality but the rewards are huge I.M.H.O. Provided your PC and graphics card can handle it. My old Core 2 Quad 9300processor could do it and was from 2008. So most will be able to do just fine.
I have separate settings for DVD and Blu-Ray to be able to maximise both.
I do not get the JRiver for video "for now" comment. I just love that part of JRiver. If you use Red October HQ at least. It's rendering is done by MadVR if you choose the HQ playback. I played and tinkered with it extensively and love the picture quality I was able to get. I love it so much it has become my player for all video sources and had me abandon separate players (my DVD player has done nothing since). It's very powerful with the right settings for your hardware and works very well for me. Combined with all audio tools it is what made JRiver worthwhile for me.
It takes some time to properly setup MadVR for maximum quality but the rewards are huge I.M.H.O. Provided your PC and graphics card can handle it. My old Core 2 Quad 9300processor could do it and was from 2008. So most will be able to do just fine.
I have separate settings for DVD and Blu-Ray to be able to maximise both.
Last edited:
Sorry, JRiver has been getting on my nerve since day 1.
I don't mind all the extra settings, but it just seems a bit all over the place... maybe it's because I didn't get JRiver yet. It's a beast with all the tweaks available, just not very user friendly (in my modest view) when it comes to using more than just the stop / play button.
I am using the HQ playback, at least, it was available with my i3-4150 with integrated graphics, and I do not get artefacts doing so. I did not get into settings deeper than just enabling HQ.
I'll take a stroll to the JRiver forum to see what happening between the convolution and the surrounds.
I don't mind all the extra settings, but it just seems a bit all over the place... maybe it's because I didn't get JRiver yet. It's a beast with all the tweaks available, just not very user friendly (in my modest view) when it comes to using more than just the stop / play button.
I am using the HQ playback, at least, it was available with my i3-4150 with integrated graphics, and I do not get artefacts doing so. I did not get into settings deeper than just enabling HQ.
I'll take a stroll to the JRiver forum to see what happening between the convolution and the surrounds.
OK, a few things:
-Since the correction filter is a PCM stereo file it can only be used with PCM stereo (no 5.1 and no DSD).
-My filter is "centering" the sound because I compensated for an imbalance of the two channels (left channel was ~4db louder with the correction filter applied).
-The sound is most likely more focused with my filter because I had to show DRC where the impulse center was for the left channel.
-The filter can't cause (or correct) a time delay between channels as long as the filter length is the same for both channels. Whether or not I performed a slight time shift on the left channel IR is insignificant; what matters is that I told DRC where the impulse center was on the left channel since it was not the loudest peak in the waveform.
-The correction strength I used is less than any of the standard presets, and will cause less artifacts when listened to from outside the "sweet spot".
-Since the correction filter is a PCM stereo file it can only be used with PCM stereo (no 5.1 and no DSD).
-My filter is "centering" the sound because I compensated for an imbalance of the two channels (left channel was ~4db louder with the correction filter applied).
-The sound is most likely more focused with my filter because I had to show DRC where the impulse center was for the left channel.
-The filter can't cause (or correct) a time delay between channels as long as the filter length is the same for both channels. Whether or not I performed a slight time shift on the left channel IR is insignificant; what matters is that I told DRC where the impulse center was on the left channel since it was not the loudest peak in the waveform.
-The correction strength I used is less than any of the standard presets, and will cause less artifacts when listened to from outside the "sweet spot".
Had another short window with my wife taking a shower upstairs! 🙂
Here's the REW sweeps for L-R and stereo, with gmad's filter.
https://www.dropbox.com/s/2z6b3tqvppfdukv/rew with conv.mdat?dl=0
Here's the REW sweeps for L-R and stereo, with gmad's filter.
https://www.dropbox.com/s/2z6b3tqvppfdukv/rew with conv.mdat?dl=0
JRiver offers you the use of different filters for different sources. With a configuration file you could setup JRiver to use the correct file for a 5.1 setup and a different one for stereo.
It will be a lot of work though to setup using DRC. Audiolense and Acourate as well as Dirac already have build in multi channel support. In DRC it would all be manual labour and that means a lot of tinkering.
Too bad you don't get along that well with JRiver Perceval, I loved it from the start for all the different things one can do with it. It takes some time to set it all up but once that's done it is as easy to press play and stop 😉. Remote control with Gizmo on my android devices helps too as well as streaming audio and or video from and to different devices in my network.
It will be a lot of work though to setup using DRC. Audiolense and Acourate as well as Dirac already have build in multi channel support. In DRC it would all be manual labour and that means a lot of tinkering.
Too bad you don't get along that well with JRiver Perceval, I loved it from the start for all the different things one can do with it. It takes some time to set it all up but once that's done it is as easy to press play and stop 😉. Remote control with Gizmo on my android devices helps too as well as streaming audio and or video from and to different devices in my network.
I will do my very best to try and love JRiver.
I just downloaded Dirac's demo that I will try tomorrow.
I previously downloaded Audiolense, but the demo does not generate filters, instead, it generates a file that attenuates the sound by 10dB, and by that, you are supposed to trust that it can do its job!
Unfortunately, they are both pricey solutions.
I'm mostly interested to see how they would handle the weird impulses I get in my room.
I just downloaded Dirac's demo that I will try tomorrow.
I previously downloaded Audiolense, but the demo does not generate filters, instead, it generates a file that attenuates the sound by 10dB, and by that, you are supposed to trust that it can do its job!
Unfortunately, they are both pricey solutions.
I'm mostly interested to see how they would handle the weird impulses I get in my room.
There is something seriously wrong with that. It should be very close to my virtual correction...
Tomorrow, I'll bring down my good old MacBookPro and redo the initial sweeps with REW, on the Mac's side.
I've been using my sound interface with the Presonus drivers on the PC side, but I think they are a bit sketchy... many people have reported some problems with them. This way, I can compare if it is hardware, software related, or just my room being terrible for sound reproduction.
I've been using my sound interface with the Presonus drivers on the PC side, but I think they are a bit sketchy... many people have reported some problems with them. This way, I can compare if it is hardware, software related, or just my room being terrible for sound reproduction.
Well that's where I would start to investigate. Those pulses look strange and rough to me. The quick measurement with convolution didn't help. They look worse than what you started with. I can't say it another way but you need quite some time to get the hang of this. My first measurements didn't go so well either. But the quality of the measurements determine the amount and quality of correction you'll get in the end.
It might not be unwise to start a new thread and take it from the top. If you're willing of coarse. My start would be to see if the pulses you are getting are that rough due to placement or some other factor.
First thing I would do is get one speaker out in the middle of the room and try to correct that first from about 2 meter with the microphone. Verifying each step along the way.
Sadly you can't rush into this and count on everything to work. I've made lots of mistakes along the way and that has thought me more each time than the successful tries.
Dirac is the strangest solution of the 3 I mentioned. It is a combination of convolution and EQ. Audiolense and Acourate come at a hefty price and that's why I'm using DRC for now.
And maybe longer than that as it has proven to be very powerful once you get the hang of it.
Don't expect either one of these programs to perform miracles out of the box. It's up to the user to make it perform miracles and with a bit of guidance it can actually do that.
Right now I'm saddened I am in a different part of the world and can't look over your shoulder. Don't give up, do not jump into multi channel correction before we can prove the 2 channel correction to work.
It might not be unwise to start a new thread and take it from the top. If you're willing of coarse. My start would be to see if the pulses you are getting are that rough due to placement or some other factor.
First thing I would do is get one speaker out in the middle of the room and try to correct that first from about 2 meter with the microphone. Verifying each step along the way.
Sadly you can't rush into this and count on everything to work. I've made lots of mistakes along the way and that has thought me more each time than the successful tries.
Dirac is the strangest solution of the 3 I mentioned. It is a combination of convolution and EQ. Audiolense and Acourate come at a hefty price and that's why I'm using DRC for now.
And maybe longer than that as it has proven to be very powerful once you get the hang of it.
Don't expect either one of these programs to perform miracles out of the box. It's up to the user to make it perform miracles and with a bit of guidance it can actually do that.
Right now I'm saddened I am in a different part of the world and can't look over your shoulder. Don't give up, do not jump into multi channel correction before we can prove the 2 channel correction to work.
I'd love to have a DRC guru looking over my shoulder and pointing to my mistakes! 🙂
If you ever plan a trip to Asia! ha ha!
I've taken enough of your towers real estate in this thread, so yes, I will start a new thread, and it might indicate to others the steps required to reach DRC nirvana, or close enough anyway!
Thanks so much for being there, guys!
If you ever plan a trip to Asia! ha ha!
I've taken enough of your towers real estate in this thread, so yes, I will start a new thread, and it might indicate to others the steps required to reach DRC nirvana, or close enough anyway!
Thanks so much for being there, guys!
For all of the followers of this DRC trail perceval started a thread for his journey here:
http://www.diyaudio.com/forums/full-range/272870-drc-trials-failures-successes.html
No idea how many are actually watching this journey, but if we can help perceval to achieve better sound we might just convince others to try DRC. At least that's what I'm hoping for. Let us know if you have additional questions 🙂. This really is fun stuff! Very powerful and definitely worth a shot. Don't be shy and speak up. With the generous help of gmad I'm sure we can get to the bottom of this and achieve a successful implementation.
http://www.diyaudio.com/forums/full-range/272870-drc-trials-failures-successes.html
No idea how many are actually watching this journey, but if we can help perceval to achieve better sound we might just convince others to try DRC. At least that's what I'm hoping for. Let us know if you have additional questions 🙂. This really is fun stuff! Very powerful and definitely worth a shot. Don't be shy and speak up. With the generous help of gmad I'm sure we can get to the bottom of this and achieve a successful implementation.
Last edited:
A repost from perceval's new thread but as it covers the EQ of my line array I figured to have it in here also:
Let's see a simplified presentation of what DRC actually does for my Line Arrays.
For this example I used measurements of the left channel of my Line Arrays. It does not show everything that's going on but makes it a bit clearer to see what it is doing.
A side note: the imported correction file isn't exactly the same as the file used in the measured signal, but close enough for this example.
From top to bottom:
Rough measurement of left channel taken at listening position. (base signal)
Impulse correction for left channel generated by DRC imported back into REW (actual correction)
DRC correction applied to the rough measurement (Red signal + Green signal)/2. (predicted response)
Actual measurement of the corrected left channel response at the listening position using convolution. (actual measurement of response)
It shows that it does a lot of corrections to the frequency response that would be very hard to replicate with (P)EQ. This simple graph leaves a lot of the factors out like phase correction etc. but it clearly demonstrates it's power. The "Maximum correction Boost" in this example was high at 9.25 dB. Not recommended for use on a single point source. But I use DRC to shape the response of my 25 driver/side Line Arrays.
It clearly shows there aren't any boosts at small dips taking place, even with this large Maximum correction Boost factor.
I used an edited template as base for this correction. The windowing done in DRC is also different from the standard templates. The correction seen in this graph is today's favourite 😉. Subject to change of coarse.
Let's see a simplified presentation of what DRC actually does for my Line Arrays.
For this example I used measurements of the left channel of my Line Arrays. It does not show everything that's going on but makes it a bit clearer to see what it is doing.
A side note: the imported correction file isn't exactly the same as the file used in the measured signal, but close enough for this example.
From top to bottom:
Rough measurement of left channel taken at listening position. (base signal)
Impulse correction for left channel generated by DRC imported back into REW (actual correction)
DRC correction applied to the rough measurement (Red signal + Green signal)/2. (predicted response)
Actual measurement of the corrected left channel response at the listening position using convolution. (actual measurement of response)
It shows that it does a lot of corrections to the frequency response that would be very hard to replicate with (P)EQ. This simple graph leaves a lot of the factors out like phase correction etc. but it clearly demonstrates it's power. The "Maximum correction Boost" in this example was high at 9.25 dB. Not recommended for use on a single point source. But I use DRC to shape the response of my 25 driver/side Line Arrays.
It clearly shows there aren't any boosts at small dips taking place, even with this large Maximum correction Boost factor.
I used an edited template as base for this correction. The windowing done in DRC is also different from the standard templates. The correction seen in this graph is today's favourite 😉. Subject to change of coarse.
Last edited: