There is no digital in/out miniambo yet and even if you get to the relevant pins on the board's chip, it uses a different sample rate than other minidsp boards. So I assume you will be using analog. I hope we will get to see a digital miniambio one day that will make all these implementations a lot more simple and elegant.
Your project is very original and exciting. Very promising. I do hope you document the building process with some photos.
Last edited:
Poldus, Yeah, I know, I've been emailing ambio4you about this digi-ambio stack conundrum for months. I think they are droping the ball on this!
I'm using the mini dsp, mini digi stack, so at least it will save 1 A/D conversion.
Norman, I would rather use smaller drivers for minimal over-shoot with the much lower mass compared to 12 inchers. Also, it would be difficult to curve the baffles with larger drivers in there.
Thanx for your interest guys
I will post pics of the entire build, promise!
I'm using the mini dsp, mini digi stack, so at least it will save 1 A/D conversion.
Norman, I would rather use smaller drivers for minimal over-shoot with the much lower mass compared to 12 inchers. Also, it would be difficult to curve the baffles with larger drivers in there.
Thanx for your interest guys
I will post pics of the entire build, promise!
poldus, one last question, how do you perceive sounds that are hard panned to the left or the right?
Markus, I will happily respond all your questions, so fire away.
Hard lateral sounds are very satisfying probably because of the amazement factor of seeing the speakers in front of you and hearing those sounds come from where there is no speaker.
As to how close it comes to the real thing, the lack of crosstalk makes them very solid and realistic-this quality is present no matter how far to the side they appear to be-but since the high frequencies are reluctant to go that far to the sides there is a certain quality of incompletness: the higher frequencies of those sounds are not really there.
I have wondered whether there is a way to steer the extreme high and left sounds from the stereo signal and feed them to a couple of tweeters placed at the sides of the listener to alleviate this problem.
So you're saying the higher frequencies stay at the speaker or don't they just move the same amount?
Do level panned signals without any interchannel time difference behave the same?
The highest frequencies-I'm guessing upwards of 5kh- spread on a narrower angle, they do not go as far laterally as the frequencies below that.
With electronic cancellation they tend to stay very close to where the speakers are and as you crank up the effect in order to achieve better lateral spread you begin to get artifacts and a general muddiness in the sound. This separation of where the highs are coming from and where the rest of the frequencies in the sound of a given instrument are is the biggest shortcoming of ambio. with electronic cancellation. The fact that it can deliver pretty extreme widening of the stage makes this even more obvious.
The physical barrier scores better in the highs but the rest of the frequencies do not expand as much to the sides as with electronic.
For me, this is the one aspect that needs to be addressed. I don't think it will be long before someone comes up with the right algorithms.
As for the problem we are referring to, there is no difference. Otherwise, there are recordings that fare better than others. Pan-potted recordings such as those of ECM which are excellent quality can be very, very good ambiophonically. The highs always remain a problem, though.
Last edited:
Check out "spirit chaser" by dead can dance.
Really good with ambiophonics.
A lot of left right information with smooth pans across the soundstage.
It really shows off the even timber across the horizontal range of an ambiophonics system.
Also, this is a very clean recording "no hiss" with great dynamics.
No compression, sounds like SACD or dts.
Great sense of space.
It's really great music too!
Really good with ambiophonics.
A lot of left right information with smooth pans across the soundstage.
It really shows off the even timber across the horizontal range of an ambiophonics system.
Also, this is a very clean recording "no hiss" with great dynamics.
No compression, sounds like SACD or dts.
Great sense of space.
It's really great music too!
Sorry, I don't understand exactly what speaker-listener position you are refferring to. The picture you posted shows a standard triangle.
Off-topic: I love your avatar. What's his name?
that´s my cat . his name is Becas (name of one of sesame´s street characters in the portuguese version).we was 1 or 2 years old when i picked him up from the streets. Quite a unique personality.
the picture is the std triangle but with speakers facing you and not tilted 60º like everyone does , also the idea is to have the speakers more separated to improve stereo perception
Hi Becas! Mine is Pelosy. She is purring in front of the laptop as I type this.
So you are talking about using electronic cancellation with those speaker
positions, or no cancellation?
With cancellation the speakers are too far apart for it to work, at least for my ears. Without, well it is nothing but one among the miriad minimal tweaks one can possibly make to a standard stereo triangle. I cannot comment because I have not listened to one in eight years or so.
Nice to have a member in this thread that is not so far from where I live.
So you are talking about using electronic cancellation with those speaker
positions, or no cancellation?
With cancellation the speakers are too far apart for it to work, at least for my ears. Without, well it is nothing but one among the miriad minimal tweaks one can possibly make to a standard stereo triangle. I cannot comment because I have not listened to one in eight years or so.
Nice to have a member in this thread that is not so far from where I live.
Iduarte1973, turning the speakers outwards does increase xtc, but not nearly enough with "normal" loudspeakers.
Sound engineers in recording studios mix for conventional stereophonics because most people do there listening in cars or on there tv's, I hope that one day they just make binaural recordings as more and more people use headphones and watch content on there tablets. Could you imagine a horror flick done binaurally?
In the position that you suggest, you would need very tightly controlled polar response. Something like very long horns for the high's "living voice" flagship speaker, coupled with a second order gradient system for the mids and lows.
The problem with this configuration is that there will be such a strong beam of sound coming off the back wall, it will ruin imaging anyway. I suppose you could use it outside, where there is no back wall.
All said and done, this system will still be unsatisfying compared to ambiophonics.
You should check out the stereolith thread.
Sound engineers in recording studios mix for conventional stereophonics because most people do there listening in cars or on there tv's, I hope that one day they just make binaural recordings as more and more people use headphones and watch content on there tablets. Could you imagine a horror flick done binaurally?
In the position that you suggest, you would need very tightly controlled polar response. Something like very long horns for the high's "living voice" flagship speaker, coupled with a second order gradient system for the mids and lows.
The problem with this configuration is that there will be such a strong beam of sound coming off the back wall, it will ruin imaging anyway. I suppose you could use it outside, where there is no back wall.
All said and done, this system will still be unsatisfying compared to ambiophonics.
You should check out the stereolith thread.
Last edited:
Hi Melo, your project is extremely interesting. The physical realisation will already be epic, but it's nothing when thinking of the research of a correct global setting.
If I dare to say something, I did some experimentations with adjustable curvature arrays, and indeed there is a measurable incidence on the time coherence. So, take care when choosing the curvature radius, it's once for all for a given specific distance.
I have an other suggestion : just in case this would not being working as expected, maybe you could save your work by separating the array in two, think of this possibility when designing it.
May the force be with you.
If I dare to say something, I did some experimentations with adjustable curvature arrays, and indeed there is a measurable incidence on the time coherence. So, take care when choosing the curvature radius, it's once for all for a given specific distance.
I have an other suggestion : just in case this would not being working as expected, maybe you could save your work by separating the array in two, think of this possibility when designing it.
May the force be with you.
Thanx Radugazon!
This is actually a project I was going to start last year, but I got caught up in building some omni prototypes.
The curvature is set for a 12 foot distance. I figure leaning back in a chair with the ears at twelve feet will be the reference position, but also it is possible to sit forward as well, having the ears at a 10 foot distance. I think even at 10 feet comb filtering will only be prevalent above 10k, not very noticeable.
As far as building as two separate arrays and attaching them is a good idea, but I have wacky idea for a straight omni line array if this project doesn't work out.
But I can't see why it won't work.
Especially if the force is with me
This is actually a project I was going to start last year, but I got caught up in building some omni prototypes.
The curvature is set for a 12 foot distance. I figure leaning back in a chair with the ears at twelve feet will be the reference position, but also it is possible to sit forward as well, having the ears at a 10 foot distance. I think even at 10 feet comb filtering will only be prevalent above 10k, not very noticeable.
As far as building as two separate arrays and attaching them is a good idea, but I have wacky idea for a straight omni line array if this project doesn't work out.
But I can't see why it won't work.
Especially if the force is with me
Hi poldus,
It has long been recognised that certain regions of the audio spectrum create problems for XTC and need to be treated differently. Various crossovers between bands have long been used, so in that sense BACCH is nothing new.
As far as high frequencies are concerned, there are three main problems:
It may be difficult for speakers to maintain matching of amplitude and phase where they are not acting as a simple piston.
The actual head attenuation between near and far speaker increases from near zero at low frequencies. I generally use a smooth curve, but RACE assumes a constant value which over-estimates at low frequencies and under-estimates at high frequencies.
At high frequencies, I would say particularly about +/ 2kHz around 10kHz, the pinna introduces a 'resonance' (part of the HRTF) which is significantly different for the two (near, far) speaker positions.
There are some way round the problems which could be used for DIY enthusiasts, but need a degree of dedication and tweaking that is not 'plug and play' that people seem to demand...
When there is a problem with XTC, such as the fact that is impractical to cancel at bass frequencies, or where a peak occurs, it is common to fold back to stereo or mono for that band.
There are other methods though. Even the very first filter I made (sdtest.wav?) although seemingly very simple, used EQ at low/mid frequencies to produce a flat frequency response at the ear Above there, the speaker angle (delay) was chosen to produce a notch at aout 11kHz where the (my) ear aready had one, and it was the speaker drive and hence diffuse field that was chosen to be flat. By the time I made some demo disks for Ralph, used at CES, I was already using a crossover to stereo/mono for the bass, and playing tricks with ITD and decorrelation.
One 'could' do a lot better than sdtest, RACE or even BACCH..IMO..
This is why I am high passing 6 db an octive with no high pass on R.A.C.E.
The under-estimation for high frequencies must be handled through the algorithm for xtc.
Which I cannot control.
I will just have to low pass R.A.C.E. I just wish it did so with a shallower slope.
Last edited:
In general, filters need to either be applied equally to both channels, or you must use constant group delay (symmetrical FIR) filters. If not the relative phase shifts upset timings, and can propagate a long way, from the corner frequency for example. I have seen this on some of the race apps when low/high pass filters are used.