Suggested improvement of the DCR part
This all very impressive work and very interesting.
Let me suggest that you take a look at the Denis Sbragions DCR tool called "DCR".
There is a wiki at: http://www.duffroomcorrection.com/wiki/Main_Page
And the tool itself: http://drc-fir.sourceforge.net/
What the tool does is it produces a Impulse response that tries to correct as much as possible of the rooms negative influences both in time and frequency domain - the Proposed methode in this thread only deals with correction in the frequency domain.
The Impulse response must be convolved with the main audio stream prior to X-over process. The X-over must stll be given great care and thought in order to reach a god result.
If Foobar 2K is used to feed the console program the convolver in foobar 2K can be used.
Let the ideas flow ....
This all very impressive work and very interesting.
Let me suggest that you take a look at the Denis Sbragions DCR tool called "DCR".
There is a wiki at: http://www.duffroomcorrection.com/wiki/Main_Page
And the tool itself: http://drc-fir.sourceforge.net/
What the tool does is it produces a Impulse response that tries to correct as much as possible of the rooms negative influences both in time and frequency domain - the Proposed methode in this thread only deals with correction in the frequency domain.
The Impulse response must be convolved with the main audio stream prior to X-over process. The X-over must stll be given great care and thought in order to reach a god result.
If Foobar 2K is used to feed the console program the convolver in foobar 2K can be used.
Let the ideas flow ....
Re: Suggested improvement of the DCR part
Good tip Henckel!
I've messed around with DCR and ACXO before I found all this.
Like you said you can use a convolution plugin in winamp/foobar to great effect. The downside of this is that your strictly limited to music only and no external sources or DVD decoding aswell as games filtering.
There's actually some very powerful professional convolvers that are available for this setup. I haven't mentioned them so far because I didn't want to confuse and muddy things. I'll save that till afterwards and if folks like what they hear then they can try other little suggestions out.
The one I use is called Voxengo Pristine Space and actually can convolve a stereo inverse reverbs. Though you need two mics to capture such a thing!
The main problem with impulse responses is the fact that unless you use a very high quality mic and pre-amp to capture the room and also use a suitable sound, usually a gun shot or electric 'crack', to trigger the rooms impulse then the results are often inferior to the method I'm describing here. The main problem is that the very tonal qualities of the sound are altered in quite a dramatic fashion with a convoluter and its easy to imagine that a less that perfect impulse response may actually make things worse.
Back in my perceive v1 construction thread we talked about this and I actually hired an Earthworks M1 microphone and a decent preamp from a local company The results were good but it was difficult to get a good impulse response.
The method outlined already is superior 90% of the time from my experience messing around.
Henckel said:
Good tip Henckel!
I've messed around with DCR and ACXO before I found all this.
Like you said you can use a convolution plugin in winamp/foobar to great effect. The downside of this is that your strictly limited to music only and no external sources or DVD decoding aswell as games filtering.
There's actually some very powerful professional convolvers that are available for this setup. I haven't mentioned them so far because I didn't want to confuse and muddy things. I'll save that till afterwards and if folks like what they hear then they can try other little suggestions out.
The one I use is called Voxengo Pristine Space and actually can convolve a stereo inverse reverbs. Though you need two mics to capture such a thing!
The main problem with impulse responses is the fact that unless you use a very high quality mic and pre-amp to capture the room and also use a suitable sound, usually a gun shot or electric 'crack', to trigger the rooms impulse then the results are often inferior to the method I'm describing here. The main problem is that the very tonal qualities of the sound are altered in quite a dramatic fashion with a convoluter and its easy to imagine that a less that perfect impulse response may actually make things worse.
Back in my perceive v1 construction thread we talked about this and I actually hired an Earthworks M1 microphone and a decent preamp from a local company The results were good but it was difficult to get a good impulse response.
The method outlined already is superior 90% of the time from my experience messing around.
Audiophilenoob said:
The Antelope is supposed to be the best clock in existance but its not cheap and whether its worth it is anyones opinion. I haven't tested it against others clocks so couldn't give an opinion as to how it performs against others.
We really need RyanC or Vil to interject here because they have more knowledge than me regarding such things.
Illusus said:
I'm loathed to recommend it as I feel its likely you'll try to compare it to this setup. But foobar with plugins is the cheapest option by far for simple 2 way.
I tried this method though and actually prefered the DCX, so you may be better served elsewhere if you don't fancy the hassle and expense of this setup.
If you do try the foobar method please don't compare it to this one because, as I've said, its not a patch on it.
m0tion said:
Audiophilenoob said:
I've compared it to the following all on the same speakers and in the same room. Only thing that changed was the XO.
First was passive:
I prototyped the original XO using just cheap wirewound resistors, air cored inductors and cheap polypropylene caps.
After I was happy and settled on proper values I went for Audyn + caps, Goertz copper foil inductors and Mills non-inductives. I wasted a fair amount of money on all these boutique components and even went to the trouble of ordering the Goertz & Mills from the US.
Then analogue active:
I used Rod Elliots P09 boards to create a 3-way XO. Here I also spent decent money and the best components I could get such as black gates etc.
Then Behringer DCX:
I actually prefered this to the analogue active but not to the passive setup.
Finally what I have now and detailed in this thread:
Comparisons between all the previous attempts showed an audible improvement by a decent magnitude when setup with DRC and fine tuned.
I've simplified things here but there was actually postive and negative points to each setup. There wasn't one that was better in all area's and that includes cost. There was a clear winner where it matters the most though and that was sound quality.
Re: Re: Suggested improvement of the DCR part
On the measurering of the impulse response in the listening position you must use a log sine sweep and the appropriate invers filter in order to get the IR. The log sweep methode yields by far the best and most reliable measurements of the IR, all other know exitations signals has lower S/N
http://support.supermegaultragroovy.com/wiki/index.php/Log_Sweeps_vs_MLS ... with further links
And the actual tools for the PC platform:
http://www.ramsete.com/aurora/
ShinOBIWAN said:
On the measurering of the impulse response in the listening position you must use a log sine sweep and the appropriate invers filter in order to get the IR. The log sweep methode yields by far the best and most reliable measurements of the IR, all other know exitations signals has lower S/N
http://support.supermegaultragroovy.com/wiki/index.php/Log_Sweeps_vs_MLS ... with further links
And the actual tools for the PC platform:
http://www.ramsete.com/aurora/
This is quite the interesting thread. I myself own a DEQX, that admittedly I haven't had much time to play with. I was looking into a PC solution before buying my DEQX but felt that it would require too much effort for the time being and took the easy way out for now.
I do want to eventually experiment further in this realm of things. I wonder if the equivalent to all your applications is available in Linux, I know that FIR XOs have been done but not sure to what extent the DRC has been done. I think the resources required could be minimized under Linux, but I am not sure about that.
I also think if video wasn't really a requirement you could get away with a much less powerful PC by using a much larger buffer, so hiccups wouldn't be a problem. Again I could be wrong, but I have to think it has to be possible.
Ultimately, the real reason I intent to eventually move from DEQX to PC based DSP is for ambiosonics. A friend of mine has a proof of concept system designed around an ambiosonic presentation. I have also heard SACD & DVD-A multichannel presentation on a $150,000 system. Let's just say DVD-A/SACD Mch missed the mark. Not only is it more convincing, it works with mono/stereo/mch recordings you already have, so one doesn't have to go out and find those 2 elusive mch recordings done right that you actually like.
I do want to eventually experiment further in this realm of things. I wonder if the equivalent to all your applications is available in Linux, I know that FIR XOs have been done but not sure to what extent the DRC has been done. I think the resources required could be minimized under Linux, but I am not sure about that.
I also think if video wasn't really a requirement you could get away with a much less powerful PC by using a much larger buffer, so hiccups wouldn't be a problem. Again I could be wrong, but I have to think it has to be possible.
Ultimately, the real reason I intent to eventually move from DEQX to PC based DSP is for ambiosonics. A friend of mine has a proof of concept system designed around an ambiosonic presentation. I have also heard SACD & DVD-A multichannel presentation on a $150,000 system. Let's just say DVD-A/SACD Mch missed the mark. Not only is it more convincing, it works with mono/stereo/mch recordings you already have, so one doesn't have to go out and find those 2 elusive mch recordings done right that you actually like.
Well, I broke out all the software I had left over from when I tried this the first time and installed it on my Athlon64 3000+ (running at 2250MHz) box this time. I also have my Audiotrak Prodigy 7.1LT installed in this box.
So far I haven't had any problem getting all of the stuff to work and even running 2 instances of CurveEQ, 6 instances of Waves LinEQ, and 4 instances of the Simple Delay my CPU utilization doesn't go over 40%. I can even use an external analog input and have it processed as well (either by itself or on top of audio playing from my PC). I'm giving some serious thought to giving this thing another go considering I still haven't found an active crossover to go with my almost complete active 3-way system.
I'll keep you guys updated if I end up implementing it. I can't wait for the next parts of Shin's guide so I can get a better idea of how to configure some of these plugins, especially CurveEQ. They seem so simple, but some of them I've found to be pretty complicated. Below is a screenshot of what I've got going on now. The plugins mark "High-L" and the like are all Waves LinEQ.
http://www.whatisrazar.com/console.jpg
So far I haven't had any problem getting all of the stuff to work and even running 2 instances of CurveEQ, 6 instances of Waves LinEQ, and 4 instances of the Simple Delay my CPU utilization doesn't go over 40%. I can even use an external analog input and have it processed as well (either by itself or on top of audio playing from my PC). I'm giving some serious thought to giving this thing another go considering I still haven't found an active crossover to go with my almost complete active 3-way system.
I'll keep you guys updated if I end up implementing it. I can't wait for the next parts of Shin's guide so I can get a better idea of how to configure some of these plugins, especially CurveEQ. They seem so simple, but some of them I've found to be pretty complicated. Below is a screenshot of what I've got going on now. The plugins mark "High-L" and the like are all Waves LinEQ.
http://www.whatisrazar.com/console.jpg
Ok, very strange.
After I just had that pretty wonderful experience with all of the software we've been discussing I loaded up Console (v1.4.0) again. I'm not sure what I changed or what about my system changed, but when I hit the power on button in Console my PC got so slow it took me about 2 minutes to end the Console task. Tried it again and got pretty much the same thing.
I've now got Console loaded with just 1 instance of CurveEQ and even though it doesn't seem to be taking up more than 15% or so of my cpu (which seems really high) every time I click on a new window, or open up some thing new I get horrible clicking and popping. This is the Console experience I'm used to, now I just want to know how to get back to when everything was working great! =)
Seriously, I can't think of one thing that I've changed other than the slopes inside of the LinEQ plugins and I have no idea why it's preforming so poorly with just 1 CurveEQ instance. Suggestions welcome.
After I just had that pretty wonderful experience with all of the software we've been discussing I loaded up Console (v1.4.0) again. I'm not sure what I changed or what about my system changed, but when I hit the power on button in Console my PC got so slow it took me about 2 minutes to end the Console task. Tried it again and got pretty much the same thing.
I've now got Console loaded with just 1 instance of CurveEQ and even though it doesn't seem to be taking up more than 15% or so of my cpu (which seems really high) every time I click on a new window, or open up some thing new I get horrible clicking and popping. This is the Console experience I'm used to, now I just want to know how to get back to when everything was working great! =)
Seriously, I can't think of one thing that I've changed other than the slopes inside of the LinEQ plugins and I have no idea why it's preforming so poorly with just 1 CurveEQ instance. Suggestions welcome.
Heh, ok, I'm not 100% sure what the solution is, but I've noticed that my CPU usage changes drastically depending on what sample rate I choose in Console. I guess this makes sense because it determines how many times per second it has to perform these calculations. At 96KHz my system falls apart, at 48 it's usable, but the cpu utilization goes up to about 70% with all of the filters I show in my screenshot. What sample rate do you use Shin?
Not sure what's wrong motion.
I get less than that running my 5.1 multichannel XO:
EDIT: Missed you last post, I use native where possible 44Khz for CD's 48Khz for DVD and 96Khz for DVD-A.
The CPU load above is with 48Khz. With 96Khz this roughly doubles to around 60-70%
I get less than that running my 5.1 multichannel XO:
An externally hosted image should be here but it was not working when we last tested it.
EDIT: Missed you last post, I use native where possible 44Khz for CD's 48Khz for DVD and 96Khz for DVD-A.
The CPU load above is with 48Khz. With 96Khz this roughly doubles to around 60-70%
Ok, i've read through the CurveEQ docs. Are you capturing your room response in CurveEQ or importing it from another program? I'm ready to give this aspect a shot, but I'm pretty confused as to how to get my room's response into CurveEQ. I know once I get it in there I can just hit the 'I' to invert it and then it should give me flat response.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.