Delayed First Reflections

[Tim Link]

I've been thinking about the need for reflections coming from about 60 degrees off axis for a sense of acoustic envelopment, which I have, but I also understand that they should have more delay time than I get from my roughly 13' wide room. It occurred to me that I could bounce the reflections around a bit to get about 10ms delay using some old 2' x 4' gobos I've had for years.

The result sounds really nice. Clear, natural tone and spaciousness are happening. This is the first time I've found a meaningful use for these gobos! I was already getting good sound without any treatment but now is noticeably nicer.

This is a variation on a concept I was taught as ASC. TubeTraps can be set up this way as well using their half reflect / half absorptive feature. If your rear wall is flat and featureless on either side of the listening position it can be used as an excellent rear reflector. My room has too much stuff going on and it's not symmetrical back there, with an opening into the kitchen on one side and a piano and a wet bar.

My speaker arrangement is unusual, being a 3 speaker array with speakers placed very close together in the middle of the room and fed a matrix mix to create crosstalk reduction. With a standard 2 speaker setup you'd just have to change the position of the reflectors accordingly, and you might need another set of absorbers for the first reflection point since it might not be where you want to hear the delayed reflections come from.

I used a mirror to position and angle the reflectors.

First put the mirror on the rear reflectors and view from the speaker. You should see the side reflector in the mirror.

Then put the mirror on the side reflector. View the side reflector from the rear reflector. You should see your listening chair in the mirror.

Another potential good thing about this is you are absorbing reflections coming laterally off the speaker and replacing them with reflections that are closer to on axis from the speaker - more like the sound that reached you directly from the speaker.

 

[dasdoing]

hehe, that's kind of cool. have you meassured this before and after? would love to see the ETC

 

[Duke]

It occurred to me that I could bounce the reflections around a bit to get about 10ms delay using some old 2' x 4' gobos I've had for years.

Very clever!

The result sounds really nice. Clear, natural tone and spaciousness are happening.

Ime one can simultaneously achieve both clarity with spaciousness by introducing a significant time gap (like your 10 milliseconds) between the first-arrival sound and the strong onset of reflections...

Another potential good thing about this is you are absorbing reflections coming laterally off the speaker and replacing them with reflections that are closer to on axis from the speaker - more like the sound that reached you directly from the speaker.

... and by minimizing the spectral discrepancy between the direct sound and the reflections, natural timbre is preserved. Brilliant!

 

[Tim Link]

hehe, that's kind of cool. have you meassured this before and after? would love to see the ETC

I haven't but I'll try to do that. While I really like the way this sounds it is a bit awkward to have the room set up like this. I've put it away for now so when I get unlazy I'll take some measurements, put the arrangement up, and measure again.

 

[Tim Link]

Very clever!



Ime one can simultaneously achieve both clarity with spaciousness by introducing a significant time gap (like your 10 milliseconds) between the first-arrival sound and the strong onset of reflections...



... and by minimizing the spectral discrepancy between the direct sound and the reflections, natural timbre is preserved. Brilliant!

It's nice to try something like this and verify it for onself. It's kind of hard to describe the effect but when I hear it I can say with confidence that I really do like it. Getting that time gap to happen a little better really does a nice thing that you and others have been on to. I've been thinking about a version of live end / dead end room with a bit of a divider wall, maybe a couple feet deep, off the side walls and ceiling separating the dead front end of the room where the speakers are from the live end where the listener is, blocking all the first reflections. That way you can get a real clean first arrival followed by lots of lateral reflections off those baffle walls to open things up and maintain an adequate RT60. If you put one on the floor as well it'd remind me of stepping through hatches on ships.

I think a dead front end without a reflective room baffle wall on the sides just ends up killing the whole room too quickly, and without providing any lateral 60 degree delayed reflections.

 

[OCA]

fed a matrix mix to create crosstalk reduction

Is this some Audacity matrix or are you talking about a RACE (Recursive Ambiophonic Crosstalk Elimination) filter? I am asking because I've been looking for a good RACE algorithm for some time.

 

[dasdoing]

the idea is somewhat simliar to a so called Haas-kicker, look it up. those where used in some control rooms for some times but where abandoned pretty fast, probably because of comb filtering effects

 

[dasdoing]

there is a guy on another forum that used two rear/side speakers to mirror a delayed signal to achieve an artificial late reflection in a otherwise pretty much reflection free room

 

[Emlin]

the idea is somewhat simliar to a so called Haas-kicker, look it up. those where used in some control rooms for some times but where abandoned pretty fast, probably because of comb filtering effects

Maybe people just don't like getting their Haas kicked.

 

[OCA]

Maybe people just don't like getting their Haas kicked.

 

[Tim Link]

Is this some Audacity matrix or are you talking about a RACE (Recursive Ambiophonic Crosstalk Elimination) filter? I am asking because I've been looking for a good RACE algorithm for some time.

No, I'm running the signal through my computer using Audio Hijack and just doing some simple channel mixing to create a 3 channel up-mix. I feed the center channel L+R, the left channel L-R, and the right channel R-L. This cancels any center panned sounds from the left or right speakers and sends them only to the center. Side panned sounds end up louder and/or phase shifted to the appropriate ear, so an appropriate stereo sound stage width is achieved. It's kind of like RACE but much simpler. It doesn't achieve the really high inter-aural difference you can get with RACE but it does create a great center with no inter-aural crosstalk to create comb filtering that you get with a standard phantom center.

 

[dasdoing]

No, I'm running the signal through my computer using Audio Hijack and just doing some simple channel mixing to create a 3 channel up-mix. I feed the center channel L+R, the left channel L-R, and the right channel R-L. This cancels any center panned sounds from the left or right speakers and sends them only to the center. Side panned sounds end up louder and/or phase shifted to the appropriate ear, so an appropriate stereo sound stage width is achieved. It's kind of like RACE but much simpler. It doesn't achieve the really high inter-aural difference you can get with RACE but it does create a great center with no inter-aural crosstalk to create comb filtering that you get with a standard phantom center.

it's a bad matrix though. think about hard panned stuff, it ends up in the sides plus the center. the result is loss in channel separation. not that it would make a diference with your somewhat strange setup lol. I get that you are experimenting, but I don't think you are on the right path here. why do you even want a center if there is single seat?

 

[Tim Link]

it's a bad matrix though. think about hard panned stuff, it ends up in the sides plus the center. the result is loss in channel separation. not that it would make a diference with your somewhat strange setup lol. I get that you are experimenting, but I don't think you are on the right path here. why do you even want a center if there is single seat?

I would have agreed with you except it sounds great! To understand the hard panned signal sounding so clean I had to do some wave simulations, which showed that the hard panned signals, like the center, remain free of comb filtering in the ears, witht he interference bands working to reduce sound in the wrong ear and increase it in the correct ear. The reason for a center is to avoid the phantom center, which is heavily comb filtered at each ear. Also, at other positions in the room the center stays in the center, which is the best way for a stereo sound field to gracefully fail as you move off center. This works especially well when watching movies.
More advanced matrixes cause their own problems, I have found. If you spread the speakers further apart, the side channels end up closer to side walls than the center, and that can create a tonal mismatch that I notice as a dryness to the center panned sound. Other people who use trinaural or other 2 to 3 channel upmixes have noted this effect. Having 3 speakers close together in the middle of the room creates a more coherent sound to my ears. It's the closest I've heard yet to using a physical barrier to reduce crosstalk.

 

[Tim Link]

it's a bad matrix though. think about hard panned stuff, it ends up in the sides plus the center. the result is loss in channel separation. not that it would make a diference with your somewhat strange setup lol. I get that you are experimenting, but I don't think you are on the right path here. why do you even want a center if there is single seat?

I made this video to show what the wave interference patterns look like with a hard panned signal. If it's panned left, you get left, left, and inverse left. The center and left channel end up building up in the left ear, and the center and right channel end up cancelling in the right ear. The result is a surprisingly solid, wide, and deep sound stage.
In the video I represented my head as a circle. Getting the spacing of the speakers correct for the listening distance is important. If you're sitting too close you can see a bit of a scrambling of interference patterns in high frequencies. If you get too far back everything just blends to mono. At lower frequencies below 500Hz there's not as much interaural difference happening but there's still a phase shift across the head, so even tones down below 400Hz still sound like they're coming from far off to the side.

In this rather dull video the signal is panned to the right, and I go through a bunch of different frequencies to show that the effect is stable up to very high frequencies. An interesting thing to note is that for panned signals, a null is always right in the middle of the listener's head, so each ear is never in a null. Instead there are alternating loud lobes and quiet lobes for each ear. With a normal 2 channel setup, hard panned sounds work perfectly as far as there being no interference, but center panned sounds end up with a lobe right in the middle of the listener's head. As the frequency goes up the lobe gets narrower until finally the listener's ears are both in a null. I think this typically happens somewhere around 2 kHz. You can get it to happen up higher if you push the speakers closer together or sit further back. This sounds better but there goes your stereo width.