Sound energy is virtually absent between 50-150ms. Vocal intelligibility has never been greater. For the first time, I am getting a sense of what LEV is. Never heard it before in this room so it is new to me and may take a while to reach firm conclusions. But my initial impressions are positive and something to be pursued further.
Maybe David knows what he is talking about ? 😀
Great acquisition, Jim! It will probably take some time to dial in, now that you can adjust every parameter.
And yes, I do believe David knows a thing or two about audio 😀.
And yes, I do believe David knows a thing or two about audio 😀.
In my own quest I've setup a system similar to the one used by Prof. Edgar Choueiri in his ambiophonic days. Still lower in level than the original files, as I make them up as I go in JRiver.
Meaning I add an inverted copy of left to the right channel, delayed 0.27 ms. And do the same for the right to the left channel. Basically the same I used before but this time I'm also injecting it into the sides.
This definitely improves the 3D stage, but it's still early. Didn't have quite enough time to do a lot of experimenting as I wasted a lot of time getting all the settings right.
Meaning I add an inverted copy of left to the right channel, delayed 0.27 ms. And do the same for the right to the left channel. Basically the same I used before but this time I'm also injecting it into the sides.
This definitely improves the 3D stage, but it's still early. Didn't have quite enough time to do a lot of experimenting as I wasted a lot of time getting all the settings right.
Another day of listening, well, only about an hour...
Boy this one is getting the job done so far. Very entertaining to listen to all kinds of songs, very involving!
Boy this one is getting the job done so far. Very entertaining to listen to all kinds of songs, very involving!
Hurry up listening : )
Towns streets will soon sing into big TT Assen happy party and potent high rpm engine noise
Too late, it's already happening... 5 days of partying this year, the festivities started yesterday. Once every year this otherwise sleepy town comes to life for a few days. 😀
No (real) Line Array time until next week, Tuesday I guess... but enough distraction!
No (real) Line Array time until next week, Tuesday I guess... but enough distraction!
Continuing listening sessions with some mild adjustments. This cross talk cancelation is a lot of fun! Never had this much room within my stage before. Way too much adjustability to get it right but I can see my test zone in JRiver making it into production 🙂.
I've played various songs from Ayreon to Zappa and they all convey a bigger, deeper space. Playtime is almost over again, but I even played some Led Zeppelin and that's saying a lot for people who know me 😉. (I wouldn't dream throwing some of that on unless I'm convinced the setup is doing 'something' right. I waited months after finishing the speakers before I actually played anything by Led Zep)
I've played various songs from Ayreon to Zappa and they all convey a bigger, deeper space. Playtime is almost over again, but I even played some Led Zeppelin and that's saying a lot for people who know me 😉. (I wouldn't dream throwing some of that on unless I'm convinced the setup is doing 'something' right. I waited months after finishing the speakers before I actually played anything by Led Zep)
A few of you might be wondering what this all looks like inside JRiver,
My chain of processing looks somewhat like this:
Copying left and right to new channels, delay them, invert and attenuate...
Band pass on the XTC channel (the channels I just filled above)
Mixing the channels back to left and right, after the convolution step, notice
the left going to the right and vice versa...
FIR filter for convolution of the first 2 pairs of channels, Left & Right and the newly created Center & Sub are identical, the ambience has it's own FIR filter. The reason for the Pré EQ to be after the mid/side EQ is to not have any channel specific EQ edit messing with the mid/side processing.
I left out a few PEQ settings that don't have anything to do with this XTC experiment.
I could also show the ambient chain, as the base signal for the ambience is build up within JRiver's Parametric Equalizer... If anyone wants to know...🙂
I have another (test)zone without the convolution, Pré-EQ and PEQ settings (as those are specific to the speaker/room) that I use with Dirac pulses to loop it to REW to see what it does with a perfect signal. I create the REW pictures with that chain.
My chain of processing looks somewhat like this:
Copying left and right to new channels, delay them, invert and attenuate...
Band pass on the XTC channel (the channels I just filled above)
Mixing the channels back to left and right, after the convolution step, notice
the left going to the right and vice versa...
FIR filter for convolution of the first 2 pairs of channels, Left & Right and the newly created Center & Sub are identical, the ambience has it's own FIR filter. The reason for the Pré EQ to be after the mid/side EQ is to not have any channel specific EQ edit messing with the mid/side processing.
I left out a few PEQ settings that don't have anything to do with this XTC experiment.
I could also show the ambient chain, as the base signal for the ambience is build up within JRiver's Parametric Equalizer... If anyone wants to know...🙂
I have another (test)zone without the convolution, Pré-EQ and PEQ settings (as those are specific to the speaker/room) that I use with Dirac pulses to loop it to REW to see what it does with a perfect signal. I create the REW pictures with that chain.
You know, I was asking myself that exact question: I wonder how it all looks like in JRiver..
Thanks for the screenshots, and the spaghetti!
Do you mind if I try your setup of mixing R and L channels? ... without all the spaghetti though.
You are way past what most mortals do for their mortal ears.
Thanks for the screenshots, and the spaghetti!
Do you mind if I try your setup of mixing R and L channels? ... without all the spaghetti though.
You are way past what most mortals do for their mortal ears.
As long as you include a band pass on that XTC signal (the sub and center).
I have it set 2 pole high pass at 600 Hz and 2 pole low pass at 8000 Hz.
The EQ's with number 6, 7 13 and 8 are all linear phase. So the band pass I use is linear too...
By the way, let me know how you get on with it, I hope I didn't miss a setting somewhere (lol). This should be it though, as far as I see. Aside from all of the mid/side EQ etc... but that's just a few dB's here and there to get an even tonal balance across the stage.
Let us know how you "perceive" this twist on Stereo.
Let us know how you "perceive" this twist on Stereo.
Didn't fare so well for my short time allowed to listening.
I had to reinstall the OS, and had trouble with setting up some of that.
I finally got it going, but when I tried to do the copy-mix thing, I got a strange signal, kinda like the ET signal in the movie Contact. A little surreal. By that time, music time was up.
I have a feeling it was my audio interface introducing that signal... will have to investigate next time.
So, sorry, no review yet!
I had to reinstall the OS, and had trouble with setting up some of that.
I finally got it going, but when I tried to do the copy-mix thing, I got a strange signal, kinda like the ET signal in the movie Contact. A little surreal. By that time, music time was up.
I have a feeling it was my audio interface introducing that signal... will have to investigate next time.
So, sorry, no review yet!
I didn't hear ET entering my home, so something must be off.
An important step that I didn't mention yet, channels should be set to 2 channels in a 5.1 container to have the other channels available.
The JRSS mixing only kicks in when a multichannel track is being played.
As said, you need a band pass on the delayed signal. But if you use a linear phase EQ to set that band pass, you need to be sure the delay on all other channels is kept the same. That's why I have that row of EQ's that each channel goes trough within Metaplugin.
Another advantage of using the Metaplugin is I can listen, even on headphones what each channel gets/does as a signal just by re-wiring those spaghetti wires. It's very easy to mess up somewhere along the chain.
Especially if you also have convolution as part of it. Convolution will also alter the time delay per channel so each channel you use has to have a same length FIR filter, with the IR peak at exactly the same spot or you get one big mess.
As the copied channels are direct copies of Left and Right signals, that are added to the original Left and Right channel, they should get the same FIR filter as the Left and Right get.
I have the following convolution chain in a TXT file:
As you see I'm using a 44.100 sample rate FIR filter, no change in the sample rate in the Output Format tab either.
Left and right get filtered by "Custom_44_linPC_lowPEQ.wav" and
Center and Sub get filtered by the same "Custom_44_linPC_lowPEQ.wav".
The ambient channels SL and SR get their own FIR filter.
If I'm not using a FIR filter on one of the channels, I substitute that filter with a pure Dirac impulse of the same length as my processing chain filters. Do it any other way and the timing is messed up. Unless you fix the difference in time delay in another way of coarse.
As we are playing with time difference it is of the highest importance that those timing difference are what we want.
An important step that I didn't mention yet, channels should be set to 2 channels in a 5.1 container to have the other channels available.
The JRSS mixing only kicks in when a multichannel track is being played.
As said, you need a band pass on the delayed signal. But if you use a linear phase EQ to set that band pass, you need to be sure the delay on all other channels is kept the same. That's why I have that row of EQ's that each channel goes trough within Metaplugin.
Another advantage of using the Metaplugin is I can listen, even on headphones what each channel gets/does as a signal just by re-wiring those spaghetti wires. It's very easy to mess up somewhere along the chain.
Especially if you also have convolution as part of it. Convolution will also alter the time delay per channel so each channel you use has to have a same length FIR filter, with the IR peak at exactly the same spot or you get one big mess.
As the copied channels are direct copies of Left and Right signals, that are added to the original Left and Right channel, they should get the same FIR filter as the Left and Right get.
I have the following convolution chain in a TXT file:
Code:
44100 6 6 0
0 0 0 0 0 0
0 0 0 0 0 0
C:\DRCDesigner\Convolverfilters\Custom_44_linPC_lowPEQ.wav
0
0.0
0.0
C:\DRCDesigner\Convolverfilters\Custom_44_linPC_lowPEQ.wav
1
1.0
1.0
C:\DRCDesigner\Convolverfilters\Custom_44_linPC_lowPEQ.wav
0
2.0
2.0
C:\DRCDesigner\Convolverfilters\Custom_44_linPC_lowPEQ.wav
1
3.0
3.0
C:\DRCDesigner\Convolverfilters\amb_44_lin.wav
0
4.0
4.0
C:\DRCDesigner\Convolverfilters\amb_44_lin.wav
1
5.0
5.0As you see I'm using a 44.100 sample rate FIR filter, no change in the sample rate in the Output Format tab either.
Left and right get filtered by "Custom_44_linPC_lowPEQ.wav" and
Center and Sub get filtered by the same "Custom_44_linPC_lowPEQ.wav".
The ambient channels SL and SR get their own FIR filter.
If I'm not using a FIR filter on one of the channels, I substitute that filter with a pure Dirac impulse of the same length as my processing chain filters. Do it any other way and the timing is messed up. Unless you fix the difference in time delay in another way of coarse.
As we are playing with time difference it is of the highest importance that those timing difference are what we want.
Last edited:
If I remove all of my speaker correction (my second test setup in JRiver) and run a Dirac pulse trough JRiver with the setting I displayed here, output that in JRiver to file, single out the left channel in Audacity or Cool edit and import it into REW I get the following impulse:
You see the normal IR peak, followed by the negative band passed XTC signal at a level of -10 dB.
If we look at the FR curve we do see a bump at ~1850 Hz, a dip at ~3700 Hz etc...
Which counters the Dirac pulses summing at our ears out of sync...
Disclaimer: these IR plots have my Mid EQ (from mid/side EQ) applied to them. 😉
If we compare the summing at the ear of the XTC vs Dirac we get this:
In all of these examples I simulate the signal "leaking" to the opposing ear as being the same volume. In real life it probably isn't.
That's why a small difference in signal can go a long way.
You see the normal IR peak, followed by the negative band passed XTC signal at a level of -10 dB.
If we look at the FR curve we do see a bump at ~1850 Hz, a dip at ~3700 Hz etc...
Which counters the Dirac pulses summing at our ears out of sync...
Disclaimer: these IR plots have my Mid EQ (from mid/side EQ) applied to them. 😉
If we compare the summing at the ear of the XTC vs Dirac we get this:
In all of these examples I simulate the signal "leaking" to the opposing ear as being the same volume. In real life it probably isn't.
That's why a small difference in signal can go a long way.
Last edited:
Due to having way less time available it is taking me longer to find a good compromise.
I was quite happy with my tonal balance before I started these experiments. Now I'm happy about other parts, but the tonal balance has shifted somewhat. It's hard to get that right again as no CD has been recorded with a good set of standards. So what sounds right for one CD might not be the key for another.
It took me a long while to get it to my liking the last time, now it's going to take even longer. I got a job offer and happily accepted that. It will still leave me some room to continue these experiments but I will be moving a little slower (even slower than usual).
I was quite happy with my tonal balance before I started these experiments. Now I'm happy about other parts, but the tonal balance has shifted somewhat. It's hard to get that right again as no CD has been recorded with a good set of standards. So what sounds right for one CD might not be the key for another.
It took me a long while to get it to my liking the last time, now it's going to take even longer. I got a job offer and happily accepted that. It will still leave me some room to continue these experiments but I will be moving a little slower (even slower than usual).
I know we all like to read about your experiments, but the news of the accepted job offer is very pleasing as well!
Congrats and good luck with the new job!
Congrats and good luck with the new job!
Thanks! I'm pretty stoked about it myself!
I'll be a coach/teacher for people with autism. Actually, I started about 3 months ago on a voluntary base. It's a very good fit. I get to do all kinds of cool projects with them like 3D design and 3D printing and might even start experimenting with audio within my classes.
I'll be a coach/teacher for people with autism. Actually, I started about 3 months ago on a voluntary base. It's a very good fit. I get to do all kinds of cool projects with them like 3D design and 3D printing and might even start experimenting with audio within my classes.
Congradulations!
Fired Up, Wesayso! 🙂
Glad someone could see the gifts You have.
(finally🙄)
You have consistently amazed me how far You have taken this project. I look forward to see how it progresses now that You will soon have some $$$ to play with! 😀
Fired Up, Wesayso! 🙂
Glad someone could see the gifts You have.
(finally🙄)You have consistently amazed me how far You have taken this project. I look forward to see how it progresses now that You will soon have some $$$ to play with! 😀