Thanks for the help!
Well yeah it was a bit of a selfish experiment I admit haha. But you see if I could somehow put an allpass at key pressue points to somewhat linearize the output of my speakers this might help me "see into the recordings" better and therefore help me mix the song better. I could always take the de-emphasis out of the chain before the final mixdown/render.
The more I think about the idea the more it makes no sense in a real world haha. But it's one of those things I have seen people talk about trying with DSP that was not exactly possible without a great deal of effort and loss with analog means.
I have some maliable all pass dsp that I can mess around with and actually try the concept out. Most experiments that I have read about where they attempt this the results seem to prove that phase linearity is actually negligble at a certain point and that phase relationship of the drivers despite the linearity of the group delay is what is really important. I guess I was attempting to prove this to myself one way or the other.
Well yeah it was a bit of a selfish experiment I admit haha. But you see if I could somehow put an allpass at key pressue points to somewhat linearize the output of my speakers this might help me "see into the recordings" better and therefore help me mix the song better. I could always take the de-emphasis out of the chain before the final mixdown/render.
The more I think about the idea the more it makes no sense in a real world haha. But it's one of those things I have seen people talk about trying with DSP that was not exactly possible without a great deal of effort and loss with analog means.
I have some maliable all pass dsp that I can mess around with and actually try the concept out. Most experiments that I have read about where they attempt this the results seem to prove that phase linearity is actually negligble at a certain point and that phase relationship of the drivers despite the linearity of the group delay is what is really important. I guess I was attempting to prove this to myself one way or the other.
Last edited:
Hey Key,
This might be of some interest to you, Gunness Focus
http://www.eaw.com/info/EAW/Technical_Papers/NT_Whitepaper.pdf
and this: http://www.eaw.com/technology/proprietary/GunnessFocusing/
This might be of some interest to you, Gunness Focus
http://www.eaw.com/info/EAW/Technical_Papers/NT_Whitepaper.pdf
and this: http://www.eaw.com/technology/proprietary/GunnessFocusing/
Last edited:
Okay so this is the closest I could get with my clunky improvised all-pass network.
http://img12.imageshack.us/i/spectruma.png/
It's like a rubix cube trying to get that crazy steep slope on both sides of the crossover with the network I am using.
I thought I was hearing a slightly extended transient response at first. I tried to ABX the two different versions twice and failed. Both times though on the first two choices I could hear the difference and the ABX said I got the answer correct. But both times by the time I got to the 3rd try with ABXing it became very difficult to distinguish a difference between the two and would resort to guessing and getting it wrong.
I would think that even though the filter is obviously not a perfect match for my crossover that I should still be able to tell a difference. But I will keep this idea in the back of my head until I come across a better convolution solution to maybe import the inverse impulse response into a VST. At this point it seems like it doesn't really make much of a difference from my perspective.
http://img12.imageshack.us/i/spectruma.png/
It's like a rubix cube trying to get that crazy steep slope on both sides of the crossover with the network I am using.
I thought I was hearing a slightly extended transient response at first. I tried to ABX the two different versions twice and failed. Both times though on the first two choices I could hear the difference and the ABX said I got the answer correct. But both times by the time I got to the 3rd try with ABXing it became very difficult to distinguish a difference between the two and would resort to guessing and getting it wrong.
I would think that even though the filter is obviously not a perfect match for my crossover that I should still be able to tell a difference. But I will keep this idea in the back of my head until I come across a better convolution solution to maybe import the inverse impulse response into a VST. At this point it seems like it doesn't really make much of a difference from my perspective.
Last edited:
Hey Key,
Here's a screenshot of a measurement of 6 x KF730 line array modules flown under 2 x SB730 at about 35 feet trim, measured about 50 feet back.
The top trace is the Phase response, the bottom trace is Magnitude, (red trace is coherence).
The interesting thing is Gunness focus has been applied from about 250Hz and up to limit latency and there is a passive crossover between the MF/HF around 1200Hz but the phase response is reasonably flat.
Here's a screenshot of a measurement of 6 x KF730 line array modules flown under 2 x SB730 at about 35 feet trim, measured about 50 feet back.
The top trace is the Phase response, the bottom trace is Magnitude, (red trace is coherence).
The interesting thing is Gunness focus has been applied from about 250Hz and up to limit latency and there is a passive crossover between the MF/HF around 1200Hz but the phase response is reasonably flat.
Yeah I am still kinda new to the impulse response thing so I don't know how to make an inverse all pass that closely matches my speakers and it's active crossover. I think I know some development tools for VST which might work but I have to buy them first.
I cheated a little and just ran the file through the all-pass in reverse for one of the steps so there wasn't a way I could just take the measurement of the whole chain in real time. That last pic was just the all-pass phase de-emphasis isolated and not added into the actual speakers' crossover.
I assume the curve should look a lot like Guness's logo where you have that vertical slope on the crossover point. This seemed to be beyond what I was capable of doing with my adjustable all-pass. I think the best I could hope for was an approximate slope around the 1.8kHz range. I was doing it all with trial and error so I couldn't tweak the filters in real time either. So yeah a lot of things working against me to where I basically need a better method or tools to properly experiment with the idea.
It just seemed like there was some shift though and that my brain was working to make the regular file match the one that had the phase de-emphasis or vice versa. But yeah I should do the experiment better before reaching any conclusions.
I cheated a little and just ran the file through the all-pass in reverse for one of the steps so there wasn't a way I could just take the measurement of the whole chain in real time. That last pic was just the all-pass phase de-emphasis isolated and not added into the actual speakers' crossover.
I assume the curve should look a lot like Guness's logo where you have that vertical slope on the crossover point. This seemed to be beyond what I was capable of doing with my adjustable all-pass. I think the best I could hope for was an approximate slope around the 1.8kHz range. I was doing it all with trial and error so I couldn't tweak the filters in real time either. So yeah a lot of things working against me to where I basically need a better method or tools to properly experiment with the idea.
It just seemed like there was some shift though and that my brain was working to make the regular file match the one that had the phase de-emphasis or vice versa. But yeah I should do the experiment better before reaching any conclusions.
Hey Key,
You might want to look into Convolution techniques, for instance Gratisvolver at: http://www.catt.se/download_area.htm (Scroll down) and there's a bunch of VST implementations out there.
If you get a good impulse response of your Speakers you could do an inverse.
You might want to look into Convolution techniques, for instance Gratisvolver at: http://www.catt.se/download_area.htm (Scroll down) and there's a bunch of VST implementations out there.
If you get a good impulse response of your Speakers you could do an inverse.
How does one set up for testing a loudspeaker?
I downloaded the software but am not sure how to set up the connections of the computer to my loudspeaker I want to test.
With no instructions with the software I thought I'd ask here how one sets things up.
Please explain in detail.
Thanks in advance
I downloaded the software but am not sure how to set up the connections of the computer to my loudspeaker I want to test.
With no instructions with the software I thought I'd ask here how one sets things up.
Please explain in detail.
Thanks in advance
I just use an M-Audio Mobile Pre USB preamp with a Behringer ECM8000 mic. The preamp provides phantom power to the mic. XLR from mic to preamp, 1/4" from mic preamp to either XLR or RCA input on preamp/power amp to speakers. Setup HOLMImpulse to use the USB preamp input and output drivers and you're good to go. I'll leave it to experts to tell you how to measure speakers properly
Hopefully that helps get you started.
Tom
I just use an M-Audio Mobile Pre USB preamp with a Behringer ECM8000 mic. The preamp provides phantom power to the mic. XLR from mic to preamp, 1/4" from mic preamp to either XLR or RCA input on preamp/power amp to speakers. Setup HOLMImpulse to use the USB preamp input and output drivers and you're good to go. I'll leave it to experts to tell you how to measure speakers properly
Hopefully that helps get you started.
Tom
Thanks Tom for your response. I'm still confused however, I also have a Behringer ECM8000 mic and an Art Audio Tube MP mic preamp and am currently using basic ATB Pro software. However, ATB doesn't do impulse response.
You wrote to go from the mic preamp to speakers? The mic is supposed to be picking up the speakers output rather than providing it?
I think I need to get from my computer that's putting out the test signal to an amp, then to the speakers. Isn't that the correct way? Then the mic's output to the mic preamp then to the computer to measure the speaker's response to the signal.
Let me know if I'm off base here.
thanks again.....
All of the signal I/O for the external M-Audio device to computer is over USB. The M-Audio has 2 inputs for line and/or mic. It has 2 outputs to send measuring signal to the amplifier and speakers. That is what I was trying to describe, just not clear without pictures.
Last edited:
I have the book and reviewed the chapter on impulse response. Nice reading for a EE graduate student IMO.
IMO, simple terms would be something like this:
A good impulse response looks like "this"
A bad impulse response looks like "this" and show graphical examples.
Instead Joe D. delves into complex equations and transforms but doesn't really explain the utility of the measurement. I believe he assumed we already understood that. You know what happenes when you break up 'assume' into it's component parts? You make an *** out of U and ME.
I Agree
I Agree, so why don't you try making a measurement?
At least a loopback test of your soundcard?
I Agree, so why don't you try making a measurement?
At least a loopback test of your soundcard?
I will be posting some measurements soon. Just starting making my first measurements today. I'm using an older laptop generating the test signal out thru the headphone jack to a small 40 WPC amp then into the test speaker. Measuring with a Behringer ECM8000 mic (using REW .cal file) passing thru an Art Audio TUBE MP mic preamp back into the laptop.
Is that hook up okay?
I work mostly with older loudspeakers and test them after repairing/upgrading them. Currently using ATB PC Pro basic version.
BTW soongsc, I'm not an EE either. HOwever, I do like the international flavor of this forum!
Last edited: