Hi Klaus,
You've seen the pics of my first prototype of the single speaker stereo.. with the FRS8 and all.. There is no bass to start with
I'm using separate dipole bass below 200 Hz.
I may not recommend using this single speaker concept at bass, since at most recordings the bass is mono so it will be emitted as single mono speaker, which do not perform well in a small room acoustic space. I've talked about that earlier
Elias Pekonen Home Page - Dipole Bass vs Monopole Bass
What comes to the ability to localise bass, I've seen studies indicating it is possible down to the lowest octave 20Hz, BUT it may require free field conditions since in a small room (smaller than or comparable to wavelength) there is no propagating wavefield and ITD is ambiguous.
Using more than one bass loudspeaker with decorrelated signal (e.g. reverb) may externalise bass, but I cannot say for sure if it is localisable. However, my hypothesis is even externalised bass can help in generating more believable sound image because not heavily contradicting cues take place.
- Elias
looks like a competition for SSS - as it stands - Audyssey stereolitic audio docks:
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Nowadays these things keep popping up like mushrooms after the rain ! General megatrend
Maybe we'll see such a trend for SSS as well, one day.- Elias
Elias, I seem to have missed how you achieve the 0.5x matrix-ing of the stereo signal. Are you using passive attenuating, or DSP, or active filtering?
I'm enjoying the thread, regardless.
It's purely passive, see post #11. Very easy !
Note, similar idea can be used in other spread speaker matrix reproduction systems, like Trinaural, or the one CLS is using.
I'm integrating all the 3 channel speakers as one.
Looking at your avatar, you have potential ! You'll already have one element at the top of the box
Now, but two more elements at the sides at do some rewiring - Elias
Looking at your avatar, you have potential ! You'll already have one element at the top of the box
- Elias
so You have flooded the center element?
At the moment I don't see a reason for that. I'm using the psychoacoustic shelving filter to adjust the balance between center and side elements. It seems to work well.
- Elias
basically it lowers the level of the center element in the HF? sort of electric pillow?
The filter is steering high freq energy from the direct sound (attenuation) to the side (boost) which will be reflected as indirect sound which adds decorrelation.
This will help in spatially homogenizing pinna cues ! The loudspeaker itself will not be localisable.
- Elias
quite like an electric equivalent of the pillow trick or is there something I don't understand?
While the pillow only attenuates the direct sound, filter also boost side signals.
Another difference is the amount of level shift which with pillow was up to 15dB attenuation at the top end, the filter only attenuates few dB. The side boost is about the same level as the direct sound attenuation.
However, I haven't fully optimised the filter yet. But it seems to be that less direct sound attenuation is enough if at the same time side signals are boosted.
Extra benefit of the filter is it looks more serious
- Elias
so a few dB here and a few dB there and then we have around 10 dB of difference in sum between the level of the direct and the reflected sound - not that far from 15 dB from the pillow trick after all
of course - clearly it is about relative levels of the direct sound vs reflected sound
Extra benefit of the filter is it looks more serious
- Elias
this I cannot dispute
still I don't understand the causes of the difference in performance between SSS and stereo bipole with pillow
can You do some comparative measurements with wavelets?
such measurement could reveal what exactly is going on
For the SSS a center panned signal produces real central sound source for full range spectrum, for example for a vocalist. But for The Cardboard with pillow the direct sound does not exist for the large portion of the spectrum for the same center panned signal. The perceptual difference is clear, night-and-day alike comparison. The SSS is better.
- Elias
I have taken a more thorough look at this diagram:
and perhaps now I understand it a bit better
it now looks to me that an alternative to SSS could be a bipole with direct sound blocker AND xL+xR channel in front of the blocker as the -xR and -xL are there instead of the direct sound blocker
OTOH it would be more like convetional 3-channel setup with which we are told (in the study You have linked to recently) not to expect any miracles vs conventional 2-channel
and yet here You claim that it is like
so perhaps what matters here is that with SSS ears pick up not only L and R and xL+XR signals but also -xR and -xL signals from the front 0 degrees direction as the part of the direct sound?
but the what psychoacoustical principle does operate here?
and perhaps now I understand it a bit better
it now looks to me that an alternative to SSS could be a bipole with direct sound blocker AND xL+xR channel in front of the blocker as the -xR and -xL are there instead of the direct sound blocker
OTOH it would be more like convetional 3-channel setup with which we are told (in the study You have linked to recently) not to expect any miracles vs conventional 2-channel
and yet here You claim that it is like
so perhaps what matters here is that with SSS ears pick up not only L and R and xL+XR signals but also -xR and -xL signals from the front 0 degrees direction as the part of the direct sound?
but the what psychoacoustical principle does operate here?
but the what psychoacoustical principle does operate here?
Apparently there are 2 main difference between x = 0 and x = 0.5. And especially if x = 0 is enhanced with pillow.
1)
For low freqs having wavelength longer that element spacing, both generate same output because (L - xR) + (xR + xL) + (R - xL) = R + L.
But for high freqs where directionality is achieved due to driver dimensions and/or box size, x = 0 have no direct sound with pillow and distorted off axis direct sound without pillow whereas x = 0.5 have defined direct sound 0.5R + 0.5L.
2)
For mid-high freqs x = 0 produces side signals from raw L and R signals to be reflected from side walls to listening position.
But x = 0.5 side signals can be written (for the Left side) as 0.5L + (0.5L - 0.5R) which can be seen as half of raw sound and half of 'ambiance' signal (L-R).
So I think x = 0.5 with more defined direct sound and with enhanced side ambiance produces 'better' reproduction.
At the moment I don't know which psychoacoustic literature to quote, but my listening tests confirm this.
- Elias
Lets rewind. I found some old impulse response measurements from The Cardboard, here reanalysed with gammatone wavelets.
This is The Cardboard, which is equivalent with SSS x = 0, full side panned i.e. only one side element is active and other is mute. Here is the direct sound at the listening position at 2m in 25 m2 room recorded with omni mic:
And this is the direct sound of The Cardboard with the pillow, otherwise same as above:
What kind of responses are these ?!? Is this hifi ??
The answer lies in the time-freq domain ! Here blinking the pillow effect:
Direct sound at 0 ms is attenuated by the pillow at high freqs, but otherwise the high freq energy remains untouched. Actually it seems that by removing the tilted direct sound the total response is more balanced as first reflections at 5 ms behave much better than the direct sound.
- Elias
This is The Cardboard, which is equivalent with SSS x = 0, full side panned i.e. only one side element is active and other is mute. Here is the direct sound at the listening position at 2m in 25 m2 room recorded with omni mic:
An externally hosted image should be here but it was not working when we last tested it.
And this is the direct sound of The Cardboard with the pillow, otherwise same as above:
An externally hosted image should be here but it was not working when we last tested it.
What kind of responses are these ?!? Is this hifi ??
The answer lies in the time-freq domain ! Here blinking the pillow effect:
Direct sound at 0 ms is attenuated by the pillow at high freqs, but otherwise the high freq energy remains untouched. Actually it seems that by removing the tilted direct sound the total response is more balanced as first reflections at 5 ms behave much better than the direct sound.
An externally hosted image should be here but it was not working when we last tested it.
- Elias
For comparison, here is SSS x = 0.5.
Direct sound measured with omni mic at 1 m distance (note! different distance than above) for full side panned signal (L = 1, R = 0):
And the corresponding time-freq domain data:
- Elias
Direct sound measured with omni mic at 1 m distance (note! different distance than above) for full side panned signal (L = 1, R = 0):
An externally hosted image should be here but it was not working when we last tested it.
And the corresponding time-freq domain data:
An externally hosted image should be here but it was not working when we last tested it.
- Elias
Newest incarnation of the Stereosphere:
Left channel goes to the left firing driver, right channel to the right driver and L+R is routed to the front driver.
The 25° side firing full range drivers were EQed for flat frequeny response on driver axis. The front firing driver is high passed at 1.6kHz so the listening axis frequency response is flat.
The speaker does what it is designed for, create strong side reflections:
The image is very centered with most sounds localized at the speaker. Only certain sounds become spacious. Is the side reflection not strong enough? Does there need to be more than just a single reflection to get a more spacious sound?
An externally hosted image should be here but it was not working when we last tested it.
Left channel goes to the left firing driver, right channel to the right driver and L+R is routed to the front driver.
The 25° side firing full range drivers were EQed for flat frequeny response on driver axis. The front firing driver is high passed at 1.6kHz so the listening axis frequency response is flat.
The speaker does what it is designed for, create strong side reflections:
An externally hosted image should be here but it was not working when we last tested it.
The image is very centered with most sounds localized at the speaker. Only certain sounds become spacious. Is the side reflection not strong enough? Does there need to be more than just a single reflection to get a more spacious sound?