derf said:
With figure 8 capsule patterns, unless the capcule actually consists of two physical capsules, you will not be able to correctly determine which side of the capsule the sound is coming from. That means a total of 7 capsules. It really seems you can get better results with 4 capsules in a pyramid arrangement.
derf said:
So what is the minimu arrangement to be able to replay elevation ques with the Ambisonics. Do you also tell the system the elevation of each speaker?
Since different speakers also have different radiation patterns that vary with frequency. Is this also acounted for? or does Ambisonics recommend s specific type of radiation pattern?
The best demo that I've ever heard was 2+2+2 presented by the recording engineer Mr. Dabringhaus himself.
http://www.mdg.de/
The reproduction chain didn't consist of any magical high-end equipment but professional studio-monitors (Dynaudio Air Series) and a multichannel player and his own preamp. Can't remember if it was SACD od DVD-A (they offer both media AFAIK).
This was the most "real" reproduction that I have ever encountered.
EDIT: It was what one Dutch guy once called "cinema for the blind" !
Regards
Charles
http://www.mdg.de/
The reproduction chain didn't consist of any magical high-end equipment but professional studio-monitors (Dynaudio Air Series) and a multichannel player and his own preamp. Can't remember if it was SACD od DVD-A (they offer both media AFAIK).
This was the most "real" reproduction that I have ever encountered.
EDIT: It was what one Dutch guy once called "cinema for the blind" !
Regards
Charles
Hmm, you make a valid point. Although, if this was the case, there would be no point to the microphone whatsoever. I'll send Soundfield an email and see what they have to say on the matter(still saving for my Soundfield mic
)Well, firstly you need an Ambisonic decoder. For Ambisonic with height(elavation) information, you need a medium capable of carrying 4 channels.
Hmm, I said Ambisonics was an advanced technique, it's not magic though
Because it deals with phase and level, rather than just level as in pan-potted stereo, these differences/problems are less of a factor.
Here's the reply I got from Soundfield:
First of all there no figure-of-eight capsules in our microphones, the figure-of-eight and omni responses are generated from four coincident sub-cardioid capsules. Further, if we know the phase of a sound source on axis with a single figure-of-eight we can from the phase of the signal resulting from the figure-of-eight pick-up determine if the signal was in front (in phase) of or behind it (out of phase). Having two figure-of eight capsules at right angles to each other we can by means of a simple amplitude matrix create a figure-of-eight at any angle on that plane, adding a third dimension will allow us to create a figure-of-eight at any angle on the horizontal and vertical plane. If we then bring into play Blumlein’s theory we can mix a figure-of-eight pointing anywhere on a sphere with our omni-directional pattern to create a cardioid (or multiple) pointing anywhere on the sphere – or by altering the gain relationship any polar pattern ranging from omni to cardioid to figure-of-eight.
I hope that clarifies the situation somewhat
First of all there no figure-of-eight capsules in our microphones, the figure-of-eight and omni responses are generated from four coincident sub-cardioid capsules. Further, if we know the phase of a sound source on axis with a single figure-of-eight we can from the phase of the signal resulting from the figure-of-eight pick-up determine if the signal was in front (in phase) of or behind it (out of phase). Having two figure-of eight capsules at right angles to each other we can by means of a simple amplitude matrix create a figure-of-eight at any angle on that plane, adding a third dimension will allow us to create a figure-of-eight at any angle on the horizontal and vertical plane. If we then bring into play Blumlein’s theory we can mix a figure-of-eight pointing anywhere on a sphere with our omni-directional pattern to create a cardioid (or multiple) pointing anywhere on the sphere – or by altering the gain relationship any polar pattern ranging from omni to cardioid to figure-of-eight.
I hope that clarifies the situation somewhat
derf said:
With two figure 8 mics, is what is often used to record hall ambiences. I guess I need to look at some phase diagrams of figure 8 mics in order to further understand. Unless they custome designed their mics to have a specific phase response.
If they have a system that could use such technique to eliminate room effects, then I would really be impressed.
DIY ambisonic microphone and encoder
http://homepage.ntlworld.com/henry01/cheap_soundfield/cheap_soundfield.htm
DIY decoders:
http://www.geocities.com/ambinutter/AMBISONIC_HOME_PAGE.html
To remove hall effect: omnidirectional stereo microphone with both channels connected together.
http://homepage.ntlworld.com/henry01/cheap_soundfield/cheap_soundfield.htm
DIY decoders:
http://www.geocities.com/ambinutter/AMBISONIC_HOME_PAGE.html
To remove hall effect: omnidirectional stereo microphone with both channels connected together.
In regards to the original post...
Of course sound will be different when eminating from a paper cone than from a voice box. But we can come close.
That is until we can reproduce the human voice box using tissue engineering and pump air through it so as to make music. Which might not be too too far into the future...a few generations perhaps? (IMHO as a Biomedical Engineer) HMMM I wonder how hard it would be to veneer a reconstructed larynx?
-Wes
Of course sound will be different when eminating from a paper cone than from a voice box. But we can come close.
That is until we can reproduce the human voice box using tissue engineering and pump air through it so as to make music. Which might not be too too far into the future...a few generations perhaps? (IMHO as a Biomedical Engineer)
HMMM I wonder how hard it would be to veneer a reconstructed larynx?-Wes
DragonMaster said:
Thanks for posting these sites DragonMaster.
Looking at the mic setup and the circuit, while it might provide a sense of spacial environment, it does not seem possible that this setup will have good imaging. Having the mics too close together creates very little difference in what is recorded. Trying to use arithmetic methods to somehow cancel unwanted portions of data recorded from different mics requires precise matching of mic performances and circuits, otherwise you get smeared imaging caused by other speakers during playback.
If I had information in the decoder, I could probably explain more.
wrl said:In regards to the original post...
Of course sound will be different when eminating from a paper cone than from a voice box. But we can come close.
That is until we can reproduce the human voice box using tissue engineering and pump air through it so as to make music. Which might not be too too far into the future...a few generations perhaps? (IMHO as a Biomedical Engineer)
-Wes
Hearing the real thing (live performance) is always the most moving experience.
The FR of a voice box would be limited, I doubt it will sell. Well, unless maybe the recording industry implements a voice channel, then you probably would have to get a collection of voice boxes to play around with.
Maybe you can have Marilyn Monroe sing a Ray Charles song.http://aroundcny.com/technofile/texts/ambisonics.htm :
But Ambisonic recordings defy both the conventional multi-mike technique and the typical audiophile method. Often, the left and right microphones for Ambisonic recordings are built into the same housing, with their pickup elements only an inch or two apart.
Each one is aimed at an angle to the front and side. Because the mikes are so close, you'd think that the left and right channels would be barely different from each other, but a special signal technique makes sure that the sound heard at your speakers is exceptionally lifelike.
Some Ambisonic recordings use a fancier method involving four microphones, all nested together. The demo disc has tracks recorded by both methods, and each one seems to do a spectacular job.
####
Other microphones:
http://lalila.net/projects/ambisonicmicrophone/index.html
http://soundsorange.net/cgi-bin/blosxom.cgi/AmbisonicMicrophone/index.html
Different Ambisonic formats:
http://members.tripod.com/martin_leese/Ambisonic/reprint007.html
Software UHJ encoder:
http://www.cantares.on.ca/UHJenc.htm
Other info:
http://members.tripod.com/martin_leese/Ambisonic/why.html
http://www.soundonsound.com/sos/oct01/articles/surroundsound3.asp
I think she does still look nicer and sound more naturally as the most nicely veneered voice-box:
http://www.emiclassics.com/phpNewSite/news/news.php?story=139&mode=
Regards
Charles
http://www.emiclassics.com/phpNewSite/news/news.php?story=139&mode=
Regards
Charles
DragonMaster said:
I have no doubt the ambient sound will be better. Lots of old recording techniques sound good, but if there is not a fixed relationship between recording, decoding, and speaker location differences, then the recorded image is a hit or miss thing enven though the ambient sound seems good. This is an explanation I don't see in these sites.
With the 2+2+2, it is clear that there are 4 in the front and two in the rear, so at least you know that the 4 in the front will give you image height. One can adequately figure out the image vectors.
I totally agree that the reproduction of the ambience is very important and totally ignored part of the hi-fi experience. Also don't believe that reproducing what ( and only what ) is recorded on the 2 channel medium is a correct and complete representation of the acoustical event.
But I don't think that ambisonic alone is a solution. Ambisonic works perfectly to recreate the soundfiled around the listener's head in the LOWER frequencies, but fails to recreate the correct HF clues independently at both ears.
That's why I use ambiophonics, which is a kind of hybrid method using crosstalk cancelled front ( and rear) main speaker pair and an array of ambience speakers fed with either synthesized or ambisonically recorded (if available) ambience channels.
A big positive also that it is backward compatible with the existing library of recordings.
But I don't think that ambisonic alone is a solution. Ambisonic works perfectly to recreate the soundfiled around the listener's head in the LOWER frequencies, but fails to recreate the correct HF clues independently at both ears.
That's why I use ambiophonics, which is a kind of hybrid method using crosstalk cancelled front ( and rear) main speaker pair and an array of ambience speakers fed with either synthesized or ambisonically recorded (if available) ambience channels.
A big positive also that it is backward compatible with the existing library of recordings.
I agree that it's f***ing hard to understand, but once you find *the* text that tells you how it works, all becomes clear.
I've learned more than every other links I posted by looking here:
http://members.tripod.com/martin_leese/Ambisonic/faq_latest.html
http://www.york.ac.uk/inst/mustech/3d_audio/
I'm not really good into explanation, but it seems that all the information can be recorded using 3 figure-of-8 microphones placed differently and an omnidirectional microphone, all of them sending the signal to 4 channels; W,X,Y,Z. The signal was then encoded to stereo(for consumer apps) ommiting the Z(vert.pos) channel and decoded to W,X,Y.
Then what do you do with these W,X,Y and Z channels? You mix them using different resistor values and polarities for each channels. These resistor values match your speaker position. Better look at some decoder schemes and explanation to understand.
http://www.york.ac.uk/inst/mustech/3d_audio/bdecoder.htm
Also, the Integrex decoder documentation http://www.geocities.com/ambinutter/AMBISONIC_HOME_PAGE.html
sam9 : There is Nimbus records with over 450 titles and other companies with a total of about 120 records. All the records are old because the technology dates from 1970 and went obsolete for some reasons(Told in the ambisonic FAQ)
I've learned more than every other links I posted by looking here:
http://members.tripod.com/martin_leese/Ambisonic/faq_latest.html
http://www.york.ac.uk/inst/mustech/3d_audio/
I'm not really good into explanation, but it seems that all the information can be recorded using 3 figure-of-8 microphones placed differently and an omnidirectional microphone, all of them sending the signal to 4 channels; W,X,Y,Z. The signal was then encoded to stereo(for consumer apps) ommiting the Z(vert.pos) channel and decoded to W,X,Y.
Then what do you do with these W,X,Y and Z channels? You mix them using different resistor values and polarities for each channels. These resistor values match your speaker position. Better look at some decoder schemes and explanation to understand.
http://www.york.ac.uk/inst/mustech/3d_audio/bdecoder.htm
Also, the Integrex decoder documentation http://www.geocities.com/ambinutter/AMBISONIC_HOME_PAGE.html
sam9 : There is Nimbus records with over 450 titles and other companies with a total of about 120 records. All the records are old because the technology dates from 1970 and went obsolete for some reasons(Told in the ambisonic FAQ)
I think the arithmetic is there purely to counter the effects of spaced capsules(a few inches apart), so that they work as if the sound was recorded at the centre of all 4 capsules.
With regards to smearing, it's quite logical that there will be far less smearing than any conventional stereo mic setup, when using a spacing corrected Ambisonic mic. As what a spaced stereo pair is capturing is actually two different points in space, where as an Ambisonic microphone captures 3 dimensions of one point in space.
There is a relationship between decoding and speaker location. You effectively tell the decoder where your speakers are and it compensates, how does it do this?, I would assume by applying the correct phase/delay differences depending on the speaker position in the room, probably something to do with the omnidirectional capsule, but that's only an educated guess. I don't really know.
Hmm, is there 6 microphones in this equation?, if so how does one compensate for their spacing?
Also, how does one tell the recording(which ends up on how many channels?), where the speakers are?
Also, what pattern mics are used, if omni then the sound coming from the six speakers will be very similar and create very little difference with regard to directionality(sounds from the back of an omni, still get mapped to the front speakers), if cardoid, what happened to the rest of the sound?, is it not relevant?
It's a noble idea and I suspect it's got some merits, but it also seems to still retain a lot of the problems of spaced stereo pairs.
Hmm, how does it work at low frequencies, but fail at high ones?. Yes, there will be discrepancies at higher frequencies, due to the capsules not being in exactly the same place(physically impossible), but this is mathematically corrected for.
Also Ambisonics is as much/if not more about maintaining the correct phase/delay characteristics of the original event and mapping the sounds to the correct place, as much as it is about ambience.
Not to sound like I'm knocking you too much, but it sounds like a step backwards from Ambisonics to me. An Ambisonic mic captures the event in 3 dimensions, as it is. Synthesizing ambience channels with what can only be incorrect mapping(the sounds weren't there in the first place, even if you can put them in the space they *might* have been), seems a bit like the antithesis of Ambisonics to me.
Yep, there's one that mostly deals with Nimbus recordings, I'll try and find the url. It must be kept in mind that all cd recordings will be UHJ Ambisonics, i.e horizontal surround only, no height information. While this is great and certainly a step or two up from spaced stereo, the real magic starts to happen when height information is added(needs 4 channels of x recording medium)aka B-Format.
There is a few UHJ releases on the Organ of Corti label aswell, but I suspect it will probably be just a little too fringe for most people.
A lot of the best stuff is here:
Ambisonic.net
This next link is a good read for everyone interested, helps answer a lot of the questions posed so far/likely to be posed. Everyone should read this, prior to asking any further questions on the subject:
Ambisonics article
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.