Another observation: the presentation becomes very spacious and high frequency sounds span from side wall to side wall when the center driver is switched off or a thick absorber in front of the speaker blocks direct high frequencies.
Here's the ETC (left channel):
Here's the ETC (left channel):
An externally hosted image should be here but it was not working when we last tested it.
This signalling scheme seems to be similar with Stereolith.
But the problem as:
Is partly solved by:
You have reinvented the pillow trick !
- Elias
Attachments
You have reinvented the pillow trick !
- Elias
No re-invention just a confirmation of what we had found weeks ago. Obviously precedence is very strong for high frequency localization cues. The problem with a single stereo speaker is that center sounds fall apart into two or more components: lower frequencies coming from the center and delayed high frequency parts from the side reflections.
Could you post the impulse responses of your matrixed speaker? I'm wondering what the ETC looks like.
That's probably the short and simple truth. Strong reflections simply act as virtual loudspeakers. No other magic involved.
The reflection is different than a loudspeaker - or at least than a normal speaker (..unless the loudspeaker itself is extremely directive - like 5 degrees). (..it is after all a multiplicity of reflections spread over a much larger area.)
Also, you have to overcome the listener's dominance to the direct sound - simply a "strong" reflection(s) is generally not going to be enough.
Still, I'd agree that it conforms to a short and simple truth.
You can also make a design that goes the "other way".
Diffusivity/expansion/envelopment can also be accomplished by the angle vs. freq. (..where the angle relates to the listener's head). In other words you can move higher freq. reproduction further apart so that it's moving closer to the listener's 90 degree positions. Of course we aren't talking about a single-box solution at that point.
Elias, how do you perceive my high pass filtered noise examples with your box?
http://dl.dropbox.com/u/21936387/pink_noise_localization_test.zip
http://dl.dropbox.com/u/21936387/pink_noise_localization_test.zip
Well, 10dB is strong
More side firing drivers might help. Something like this with a lower number of direct firing tweeters:An externally hosted image should be here but it was not working when we last tested it.
First directivity measurements !
I measured impulse responses of the SSS x = 0.5 to the angles shown here at 1 m distance using omni mic. Left and right side angles must be treated separately because the whole principle of the SSS is based on vector steered directional pattern and is different for left and right sides and it depends on the stereo signal panning.
These measurements are done for hard Left side stereo panning i.e. L = 1 and R = 0. According to the theory of SSS operation, in this case left side signal should be higher than direct signal and right side signal.
In the following wavelet plots time t = 0 is the arrival of the impulse.
Here is Bark wavelet spektrogram at 0 degrees angle:
Here is Bark wavelet spektrogram at 45 degrees angle Right:
Here is Bark wavelet spektrogram at 45 degrees angle Left:
And here is the frequency responses derived from the above wavelet plots at the time t = 0:
blue = 0 deg
green = 45 deg Left
red = 45 deg Right
Heureka ! It is seen that directivity can be steered over the full audio spectrum
This is the new revolution of home audio reproduction !
- Elias
I measured impulse responses of the SSS x = 0.5 to the angles shown here at 1 m distance using omni mic. Left and right side angles must be treated separately because the whole principle of the SSS is based on vector steered directional pattern and is different for left and right sides and it depends on the stereo signal panning.
An externally hosted image should be here but it was not working when we last tested it.
These measurements are done for hard Left side stereo panning i.e. L = 1 and R = 0. According to the theory of SSS operation, in this case left side signal should be higher than direct signal and right side signal.
In the following wavelet plots time t = 0 is the arrival of the impulse.
Here is Bark wavelet spektrogram at 0 degrees angle:
An externally hosted image should be here but it was not working when we last tested it.
Here is Bark wavelet spektrogram at 45 degrees angle Right:
An externally hosted image should be here but it was not working when we last tested it.
Here is Bark wavelet spektrogram at 45 degrees angle Left:
An externally hosted image should be here but it was not working when we last tested it.
And here is the frequency responses derived from the above wavelet plots at the time t = 0:
blue = 0 deg
green = 45 deg Left
red = 45 deg Right
An externally hosted image should be here but it was not working when we last tested it.
Heureka !
It is seen that directivity can be steered over the full audio spectrum This is the new revolution of home audio reproduction !
- Elias
A sort of evolutionary issue
Some may even perceive the location of the two tweeters Well, that is solely a problem of your Stereosphere which has no directivity at the low freqs
A true SSS should have steerable directionality, at full range spectrum I'm about to measure impulse responses at the listening position as a next step ! I will post the results. But I don't do ETC, that is sooo old skool
- Elias
I'm about to measure impulse responses at the listening position as a next step ! I will post the results. But I don't do ETC, that is sooo old skool
- Elias
Would be nice to have the impulse response as a sound file so I can do my old school stuff.
would be nice of Markus to check Elias' matrix filter with his stereosphere instead of testing dirty old school tricks
A sort of evolutionary issue
So how do you perceive your matrixed box? See #128
Well, that is solely a problem of your Stereosphere which has no directivity at the low freqs
The question is how loud are the reflections compared to the direct sound? An ETC at the listening position would help. Your graphs are somewhat hard to read (for me).
AirSOUND Founder Ted Fletcher explains their technology (What HiFi Sound and Vision Show 2007)
He's the same guy who designed E-Mu 1820M sound card mic preamps and Joemeek's.
http://www.orbitsound.com/default.asp
He's the same guy who designed E-Mu 1820M sound card mic preamps and Joemeek's.
http://www.orbitsound.com/default.asp
Last edited:
First of all I modified the samples to add stereo panning according to sine-cosine panning law. The sound is panned continously from left to right at the rate of 0.125Hz. Now it was possible to sit down and listen with eyes closed and concentrate
Here are the new noise files if anyone likes to try out:
Full range:
http://dl.dropbox.com/u/2400456/diyaudio/Mono_Pink_Noise_sin-cos-panning_0.125.wav
High-pass 1.5kHz:
http://dl.dropbox.com/u/2400456/diyaudio/Mono_Pink_Noise_HP_1.5kHz_sin-cos-panning_0.125.wav
High-pass 3kHz:
http://dl.dropbox.com/u/2400456/diyaudio/Mono_Pink_Noise_HP_3kHz_sin-cos-panning_0.125.wav
High-pass 6kHz:
http://dl.dropbox.com/u/2400456/diyaudio/Mono_Pink_Noise_HP_6kHz_sin-cos-panning_0.125.wav
With these samples I compared SSSx5 (single speaker stereo with x = 0.5) to speakers in conventional stereo triangle, using A/B switching from the listening chair.
With SSSx5 low frequency part of the full range noise is not clearly localisable, it's coming from somewhere in front but not clearly from the speaker either. High freq noise in all the samples slides very smoothly from side to side ! This is not bad !
Conventional stereo triangle produces ping-pong effect in all the sample at high freqs, I cannot hear anything coming from between the speakers at high freqs. It sounds like two fixed position sound sources having pumping intensity. On the other hand, some low freq part of the full range noise seems to come also from the center location according the panning, though.
So overall SSSx5 was better than conventional stereo triangle in that the speaker was not localisable, and panning produced smooth spatial slide from side to side.
- Elias
Niro Nakamichi's "1000" Surround bar looks more stereolitically with single it's tweeter"
During the listening test I found out it is beneficial if some conditions are fullfilled with SSSx5. So I'll create first rule of thumb for listening position set up for easy experimenting: There seems to be a minimum distance between the speaker and the listener which can be approximately determined from the circle coinciding the speaker, the two side walls and the listener, like in this pic:
If the distance is considerably too small, sound seems to come mainly from the speaker apparently because proportional level of side wall reflections is too weak.
In my 25 m2 room there seems to be a listening area about 2 m in diameter where the sound do not change much and enjoyable listening experience is expected.
- Elias
If the distance is considerably too small, sound seems to come mainly from the speaker apparently because proportional level of side wall reflections is too weak.
In my 25 m2 room there seems to be a listening area about 2 m in diameter where the sound do not change much and enjoyable listening experience is expected.
An externally hosted image should be here but it was not working when we last tested it.
- Elias
Also during the listening test I made some peculiar observations considering the conventional stereo triangle !
Observation: Using the panned pink noise sample, I noticed I could hear the high frequency sound panning smoothly from side to side according to the panning once or twice but after that immediately high freq sound is locked into speakers where it remains. This happened several times, but only if there was some time I did something else than listening to stereo triangle before I continued the test. Interesting !
It seems my brain can be fooled only for a short period of time to believe there should be a high freq phantom image between the conventional stereo speakers ! But very quickly I notice it's a trick and perception comes real and I can hear the speakers. I may have very well learned to reveal the artificiality of phantom images and the illusion cannot hold.
In comparison with SSSx5 I never noticed changes in perception no matter how long I listened panned noises.
- Elias
Observation: Using the panned pink noise sample, I noticed I could hear the high frequency sound panning smoothly from side to side according to the panning once or twice but after that immediately high freq sound is locked into speakers where it remains. This happened several times, but only if there was some time I did something else than listening to stereo triangle before I continued the test. Interesting !
It seems my brain can be fooled only for a short period of time to believe there should be a high freq phantom image between the conventional stereo speakers ! But very quickly I notice it's a trick and perception comes real and I can hear the speakers. I may have very well learned to reveal the artificiality of phantom images and the illusion cannot hold.
In comparison with SSSx5 I never noticed changes in perception no matter how long I listened panned noises.
- Elias