soundfield mic for stereo application

[S.Blanchardon]

Originally Posted by Tom McC

I'm struggling to get my head round how Harpex manages to pull off this remarkable coincident -> ORTF trick.

In a thread some time ago, I pondered the notion of mapping a coincident pair to ORTF (assuming for now one would wish that) - by converting the interchannel intensity difference to a sound-source arrival direction and thence to an ORTF intercapsule arrival time difference. Petrus pointed out the futility of that, the flaw being that one needed to consider distance as well as direction, for there is an infinite number of sound source locations (lying along a hyperbola curve) that satisfy the condition of producing an identical arrival time difference at the ORTF capsules.

So what's then different with the (coincident) Soundfield? Does the extra dimension of spatial information allow the maths to be solved?. In abstract superficial terms, something akin to finding the intersection point of two hyperbolas - one derived, say, from info from the B-format X-Y pair and one from the X-Z pair? .

I talked about this matter with Svein Berge, the mastermind behind Harpex.

There is something like 17 different process steps inside the decoder.

If I understood correctly, it has to do with psychoacoustics and probability calculus to determine where the sound wave come from, at which frequency,etc... Nothing you can do with a regular decoder !

To understand a bit better the theory behind HARPEX (High Angular Resolution Planewave Expansion) :

http://harpex.net/harpex.pdf

http://harpex.net/aes40-free.pdf

And it doesn't even cover psychoacoustic or binaural subjects which are matters of thesis themselves...

Cheers,
Stéphane. (just a happy user btw ! )

PS : You can ask Svein directly, I'm sure he'll be pleased to answer you !

Harpex

[Petrus]

Originally Posted by Tom McC

I'm struggling to get my head round how Harpex manages to pull off this remarkable coincident -> ORTF trick.

In a thread some time ago, I pondered the notion of mapping a coincident pair to ORTF (assuming for now one would wish that) - by converting the interchannel intensity difference to a sound-source arrival direction and thence to an ORTF intercapsule arrival time difference. Petrus pointed out the futility of that, the flaw being that one needed to consider distance as well as direction, for there is an infinite number of sound source locations (lying along a hyperbola curve) that satisfy the condition of producing an identical arrival time difference at the ORTF capsules.

So what's then different with the (coincident) Soundfield? Does the extra dimension of spatial information allow the maths to be solved?. In abstract superficial terms, something akin to finding the intersection point of two hyperbolas - one derived, say, from info from the B-format X-Y pair and one from the X-Z pair? .

I suppose that the intensity information provided by a Soundfield microphone is so precise that Harpex can "vectorise" the signal and recalculate the "phantom" directions and phase differences for an ORTF or AB pair. Naturally Harpex can not know the exact sound source distance for a given signal, but if the distance is large enough like a few meters (it usually is), the hyperbola mentioned in the quote above starts to approach a straight line, which Harpex apparently can calculate. So the result Harpex gives can not be perfect, as it lacks the distance information, but it seems to be convincing enough to be usable.

This was not based on any knowledge about Harpex actually works, but spatial geometry in general.

The post above appeared while I was writing this.

[Lucas_G]

I understand that the Harpex plugin is better than the Soundfield plugin.

It is a pity that the price is 500 GBP. I am sure that price will be perfectly ok for the larger professional studios, but it makes all those amateurs and semi-pro people staying away from it.

There are now quite a few "budget" mics available: the Oktava (700 Euro), the Core Tetramix (1000 USD) and the Soundfield SPS200 (2000 Euro).

That is inviting a lot of budget oriented people to enter the world of Ambisonics. Most of them will either use the free Soundfield plugin that is available for the SPS 200, or one of the free other plugins like the excellent VVT.

Maybe Harpex could think about offering a "lite" version of the plugin for one or two hundred GBP, that offers no support and misses one or two non vital features...

I am sure they would benefit from it too, because they would likely sell a multitude of licenses..

Or am I wrong?

(Hope Harpex will read this suggestion one day...)

[David Spearritt]

Originally Posted by Lucas_G

I understand that the Harpex plugin is better than the Soundfield plugin.

How so? There are no alternative ways of decoding A or B format. Get the maths right and you're away.

[S.Blanchardon]

Originally Posted by David Spearritt

How so? There are no alternative ways of decoding A or B format. Get the maths right and you're away.

Having an SPS200 with both surroundzone and Harpex, I confirm that the later has a better soundspace, more extensive and refined controls and nicer features like binaural decoding.

So yes, one can say that Harpex is both subjectively and objectively "better" than surroundzone. Period.

It's a bit pricey but imho worth the investment plus you can authorize it on 2 different computers at the same time.

Cheers,
Stéphane.

[David Spearritt]

Originally Posted by S.Blanchardon

Having an SPS200 with both surroundzone and Harpex, I confirm that the later has a better soundspace ....

What does this mean? Where the better soundspace coming from?

[S.Blanchardon]

Originally Posted by David Spearritt

What does this mean? Where the better soundspace coming from?

Hi David,

Looks like you don't know how Harpex works !

To summarize it's a totally new way to decode B-Format that Svein Berge worked on and implemented in his plugin.

Do yourself a favor and read the whitepapers on Harpex website !

http://harpex.net/harpex.pdf

http://harpex.net/aes40-free.pdf

Cheers,
Stephane.

[Lucas_G]

Originally Posted by David Spearritt

How so? There are no alternative ways of decoding A or B format. Get the maths right and you're away.

Hi David,

The idea is to use the Surroundzone plugin to go from A-Format tot B-Format, and then use the Harpex plugin to go from B-Format to Stereo or Surround.

And the Harpex sounds really vastly superior, so it is a pity that it is financially a bit outside the reach of the small studios and amateurs.. What do others think about this?

Getting back to the Surroundzone plugin. I suspect that Soundfield omitted to do a proper eq within the conversion from A-Format onwards. I notice that the recordings done with the SPS200 have a profound lack of bass and low midrange which makes them sound thin and unpleasant. That may not be a problem for outside recordings, but for music the tonal balance is totally wrong! Maybe that is the reason why so many have expressed disappointment with their recording of classical music with this microphone.

Cardioids by nature have a problem to get the tonal balance right when used as main mics and a distance of several meters, but the capsules (good as they are) of the SPS200 have a tonal shift that makes them sound very thin.

If you look at the frequency response of the MBHO KA200 capsules you can see that from 300 Hz downwards the response goes down to 3 or 4 dB under the average level.

http://www.mbho.de/pdf/ka200n_ka400n.pdf

When these capsules were tested by the German PAM magazine their shift was even more profound.

For users of the MBHO mics this roll-off in the lows will often be compensated by the proximity effect of the cardioid when doing really close miking, but with the SPS200 you will usually be pretty far away from the soundsource...

Maybe this is also the reason why there are good sounding examples of recordings made with the cheaper Core Tetramic and not of the SPS200. Coresound uses cheap Sony capsules, but they also provide extensive eq. (even for each individual microphone).

Maybe individual eq is less of a necessity for the MBHO capsules, because they are made to better standards, but why did Soundfield omit eq in their plugin? Or am I plain wrong, and is my ear deceiving me? I think not…

I will do some tests on frequency linearity with the SPS200 and post my impressions. Maybe some of you can do the same, so that we can compare our findings…

I have one last question left: Is it important to do the eq before any Ambisonic processing is being done, or doesn’t it matter when we do it for instance only after processing with the resulting stereo file? It seems logical to me that the extensive Gerzon processing maths expect a flat frequency response right from the start of all the processing...

Professor Angelo Farina describes his methods to analyse a microphone for use with Ambisonics. That would be the most professional way to do it.

A-format to B-format conversion

I am afraid Soundfield did not go this road. Could some purist notion of avoiding extra processing be the culprit? But then, Ambisonics has processing all over the place.

Any comments by other Soundfield users?

[Kewl]

Originally Posted by Lucas_G

the Harpex sounds really vastly superior, so it is a pity that it is financially a bit outside the reach of the small studios and amateurs.. What do others think about this?

I'll respond by pointing out my recent post that talks about the new Double-MS decoder from Schoeps/Illusonic. It's not the Harpex-B, but for stereo or surround decoding, it's hard to beat as a free decoder plug-in. All you need to do is to transcode the B-Format to DMS beforehand.

Originally Posted by Lucas_G

Getting back to the Surroundzone plugin. I suspect that Soundfield omitted to do a proper eq within the conversion from A-Format onwards.

I've made some impulse responses out of the SPS200 SurroundZone and there is some EQ in the A to B conversion.

If, in post-production, I find that there's a lack of bass from my SPS200 recording, I'll do some EQ post-decode, on the speaker-ready signal.

[Lucas_G]

Originally Posted by Kewl

I've made some impulse responses out of the SPS200 SurroundZone and there is some EQ in the A to B conversion.

If, in post-production, I find that there's a lack of bass from my SPS200 recording, I'll do some EQ post-decode, on the speaker-ready signal.

Hi Kewl,

Thanks for your fast reaction. Can you say eanything about the kind of eq you could see (region, level) with Surroundzone?

If you use the SPS200 for music, what kind of eq do you use on the speaker ready product? Do you use it all the time?

Thanks for your insight and help!

Lucas

[David Spearritt]

Originally Posted by Lucas_G

And the Harpex sounds really vastly superior, so it is a pity that it is financially a bit outside the reach of the small studios and amateurs.

My comment still stands. The Harpex may be doing something "artificial" to "improve" the decoded sound in binaural or surround, but decoding A or B format to stereo is a fixed mathematical function and cannot vary between plugins. It's like MS decoding, one MS decoder does not sound different to any others, if they are doing it right.

EQ on the SPS200 pre or post decode is an interesting idea. I find that the recordings I have heard from it are somewhat thin and lacking in bass.

[Larry Elliott]

Interesting that this thread comes up again after quite a while.

My experience with the SPS200 has been mainly with orchestral and large choral works. I have used both Surroundzone and Harpex software.

I think that the results are on the "thin" side, but have put that down to it being coincident rather than a spaced pair. To counter this I tend to reduce the high mids a little around 2-3k. Certainly my best results have been when I have also used a pair of omnis as well. The omnis give the fullness and the Soundfield the detail.

Larry

[leddy]

I'd like to try to add to this discussion since I have been using the Coresound Tetramic for a year or two now. Most of this echos what Lucas said.

I've done some informal measurments, and the Tetramic is pretty darn flat accross a wide spectrum when configured in post to be a mono omni pattern. Though this is done through software EQ in the post process. Each mic somes with its own calibration files.

Adjusting the patterns causes some EQ changes as well. I'm assuming that, because we are dealing with 4 cardioid mics pointing different directions, the response has to change as you add or subtract different mics, where some are more on axis than others.

My conclusion is that a soundfield-type of microphone will just need a different approach to EQ than a traditional pair of mics. It took me a while to get used to the idea, and I almost sold off the Tetramic because I did not like having to EQ as much as I needed to to get the sound right*. I still have it, and I'm glad. I do like to use it.

So I don't know how it compares to the Soundfield in that regard, but that has been my experience.

*Just to add, most of the EQ-ing has to do with balancing the low and high end. The mid range is usually just fine.

[Kewl]

Originally Posted by Lucas_G

Can you say eanything about the kind of eq you could see (region, level) with Surroundzone?

If you use the SPS200 for music, what kind of eq do you use on the speaker ready product? Do you use it all the time?

This is the actual EQ and general gain applied by the SPS200 plug-in in the A to B conversion: 1st channel is the W and 2nd channel is the XYZ.

As we can see, no bass boost is present in neither the W and XYZ channels (but since the W channel is an omni, it would probably not need it anyway). This explains why some people find the sound of the SPS200 lacking in bass.

I would say that if you were to be using MBHO KA200 caps in an XY or ORTF stereo pick-up, you would probably characterize the resulting sound with the same adjectives that are given to the SPS200. I would also say that you would probably EQ that XY or ORTF to taste, depending on the source distance, the room response, etc.

So there's no reason why not to EQ the SPS200, but do it after B-Format decode so as not to mess with the spatial information.

[Kewl]

Originally Posted by David Spearritt

The Harpex may be doing something "artificial" to "improve" the decoded sound in binaural or surround, but decoding A or B format to stereo is a fixed mathematical function and cannot vary between plugins.

You're right about A to B conversion: for a given mic model, it is a fixed mathematical process. But there is no strict recipe to decode B-Format to speaker-ready signal, be it in stereo or in surround. R&D on the subject is still being done, Harpex being one of them.

The Harpex process is not inventing any spatial information, it is just better at extracting it. You could view it as a Fourier transform, but in the spatial domain.

[Kewl]

Originally Posted by Lucas_G

there are good sounding examples of recordings made with the cheaper Core Tetramic and not of the SPS200. Coresound uses cheap Sony capsules, but they also provide extensive eq. (even for each individual microphone).

The TetraMic caps are actually Chinese-made:




Below is the generic EQ and gain ("uncalibrated" setting) used for the TetraMic A to B conversion: 1st channel is W, 2nd is XYZ. No wonder there's more bass in the TetraMic compared to the SPS200...

[Lucas_G]

Originally Posted by Kewl

So there's no reason why not to EQ the SPS200, but do it after B-Format decode so as not to mess with the spatial information.

Thanks for all this valuable info, Kewl! I think this may in the end help a lot of SPS200 owners to improve their results when recording music.

My question remains: Is it really better to do eq only after the decoding? I cannot fully grasp the Gerzon maths, but it somehow seems logical to me that those maths expect to start from a point where flat mic frequency is presumed right at the start... Spatial info should depend on flat frequency (and therefore flat phase), or am I wrong?

Doesn't the Tetramic start with adding eq before doing the processing? The plugin seems to show this…

Why would Soundfield add so much treble to the Omni channel with the SPS200? Isn't there already quite a lot of treble surplus from the capsules themselves? Could this be the reason for the strident sound?

I would be very interested to know how Soundfield has come to their eq decisions...

I have mailed Soundfield, pointed them to this tread, and asked them to comment. They are known for being very open and user friendly, so maybe one of their tech people might inform us on the reason behind all this...
This is not an action aimed at discrediting the SPS200 or Soundfield. I like the product, the firm and all the brilliant post production features. Therefore I would desperately like to improve the results when recording acoustical music with the SPS200 as a main mic.

[Lucas_G]

If you look at the frequency response of the Tetramic capsules (as far as this manufacturer data is reliable), we see a rather similar response as with the SPS200 capsules.

Amazing that the eq and postprocessing is so extremely different. They actually go into opposite direction!

My ears tend to believe that Coresound is heading into the better direction!

[Lucas_G]

I just tried using the standalone “VVMIC for Tetramic” application on a chamber music recording done with the SPS200, instead of the Surroundzone plugin.

Of course this A-format to B-format conversion is not at all optimized for the SPS200, but it already sounds a lot better! If you tame the lowest bass (on the A format tab change LF boost from 150 to 60), you get so much more warmth and ambience. It is still not 100% tonally balanced, but it shows that the SPS200 CAN INDEED sound musical when we find the right settings.

In the Tetramic application one can configure various parameters like the distance between the capsules. Are the Tetramic and the SPS200 similar in this respect? (I don’t have a Tetramic at hand).

[Lucas_G]

Kewl,

Maybe we should try to test several modes for the SPS.

For instance compare the frequency response of a DPA 4006a (of which DPA always provide a plot) with the processed omni response of the SPS.

The same can be done with a well-known cordioid etc...

This way, we may come to an averaged eq that leads to a tonally balanced sound in the several modes.

[Lucas_G]

Originally Posted by Larry Elliott

Interesting that this thread comes up again after quite a while.

My experience with the SPS200 has been mainly with orchestral and large choral works. I have used both Surroundzone and Harpex software.

I think that the results are on the "thin" side, but have put that down to it being coincident rather than a spaced pair. To counter this I tend to reduce the high mids a little around 2-3k. Certainly my best results have been when I have also used a pair of omnis as well. The omnis give the fullness and the Soundfield the detail.

Larry

Hi Larry,

That was my basic idea too, and while adding a spaced pair of omnis may very well proof to be always a good idea, we should also be capable of improving the sound of the SPS itself.

Thanks for your impression concerning the sound. It seems that more users have a subjective impression that the SPS200 needs to sound bigger and warmer than it does in the current setup.

The first moment when I got the idea that something might be wrong with the tonal balance was when I tried the VVMic standalone application on a recording done with the Soundfield 442B which I had on loan. I accidently used it for a moment in the A-Format position and the Tetramic eq was applied. Suddenly the tonal balance went into the right direction. That gave me food for thought, and now with the SPS200 the need for better eq seems even greater than with the 422B.

[Larry Elliott]

Lucas,

Maybe you should contact the team at Soundfield. I visited them last November and was very impressed with the operation. I have requested a software change but they don't have the resources to fit this in.

It is certainly my impression that they want to have a very successful product with the SPS200

Larry

[Kewl]

Originally Posted by Lucas_G

My question remains: Is it really better to do eq only after the decoding? I cannot fully grasp the Gerzon maths, but it somehow seems logical to me that those maths expect to start from a point where flat mic frequency is presumed right at the start... Spatial info should depend on flat frequency (and therefore flat phase), or am I wrong?

Doesn't the Tetramic start with adding eq before doing the processing? The plugin seems to show this…

Why would Soundfield add so much treble to the Omni channel with the SPS200? Isn't there already quite a lot of treble surplus from the capsules themselves? Could this be the reason for the strident sound?

Historically, the EQ in a SoundField mic had various functions:

• Matching the caps to one another: AFAIK, the modern SoundFields (ST450, SPS200, etc) have done away with this by having tighter tolerances caps matching before assembly. The original SoundFields had processing boxes that were each matched to a specific mic unit. Not anymore. Since Core Sound uses chinese-made caps, with looser tolerances, they have to digitally match the caps like SoundField were doing back then in the analog domain.
• Flattening bass response: it's quite obvious from the EQ response from the VVTetraVST plug-in that they are doing just that for the TetraMic. It's no surprise that the bass boost is that much more important on the XYZ channels since they are figure-8 mics.
• Compensating for non-coincidence at higher frequencies: if the SPS200 EQ looks funky at high frequencies, it's not to better it's sound, it's to compensate for the fact that the caps are not really coincident and this "problem" becomes more real as the frequency goes up. The "uncalibrated" setting of the TetraMic has no high frequency compensation, but a calibrated one does.

So to compare again:

SPS200


TetraMic, uncalibrated


TetraMic, calibrated

[Kewl]

I'll add that Core Sound offers a calibration service for the SPS200. It's not stated as such, but one can presume that the outcome of this calibration service is a set of calibration files so as to use the VVTetraVST plug-in for the SPS200 A to B conversion. The best of both worlds? The better caps with the better conversion?

[S.Blanchardon]

Originally Posted by Kewl

I'll add that Core Sound offers a calibration service for the SPS200. It's not stated as such, but one can presume that the outcome of this calibration service is a set of calibration files so as to use the VVTetraVST plug-in for the SPS200 A to B conversion. The best of both worlds? The better caps with the better conversion?

Definitely interesting !

Hum, are we on the way of getting rid of surroundzone for good ?

[Lucas_G]

Originally Posted by S.Blanchardon

Definitely interesting !

Hum, are we on the way of getting rid of surroundzone for good ?

Very interesting indeed!

The Surroundzone plugin looks nice, but what counts is the sound that comes out of it.

Maybe we can test our SPS200 ourselves (compare with a known mic) and make a calibration file ourselves?

[Lucas_G]

Originally Posted by Kewl

I'll add that Core Sound offers a calibration service for the SPS200. It's not stated as such, but one can presume that the outcome of this calibration service is a set of calibration files so as to use the VVTetraVST plug-in for the SPS200 A to B conversion. The best of both worlds? The better caps with the better conversion?

Do you know anythong about pricing for calibration?

[Lucas_G]

Kewl,

What bothers me most is that de Omni channel of the SPS200 receives a 7 dB boost at 7.500 Hz.

It is not only that a low boost is missing with the SPS (in comparison with the Tetramic), but there is also a massive boost in the high frequencies compared to the rather small boost that the Tetramic receives in the treble region.

So, in fact these two sides of the spectrum both add up to the aggressive and thin sound of the SPS200.

[Kewl]

Originally Posted by Lucas_G

What bothers me most is that de Omni channel of the SPS200 receives a 7 dB boost at 7.500 Hz.

Yes, but this is done for the sake of spatial information integrity at high frequencies. Since the caps on the TetraMic are more coincident (because of their smaller size), the EQ correction at high frequencies is less pronounced for the TetraMic.

As I said before, once the SPS200 B-Format is decoded to stereo or surround, I don't refrain myself from boosting the bass, taming the mids (2 to 3 kHz usually) and rolling off above 10 or 12 kHz. Whatever sounds right to me, I do it (as I would also do on a discrete stereo or surround pickup.)

[David Spearritt]

Originally Posted by Kewl

You're right about A to B conversion: for a given mic model, it is a fixed mathematical process. But there is no strict recipe to decode B-Format to speaker-ready signal, be it in stereo or in surround.

But B-format to stereo is the same as MS decoding. You discard the vertical fig8 and the forward fig8+omni gives you the variable patterns for the mid. The side fig8 that is left is the S. There are no options here I see either.

The Harpex process is not inventing any spatial information, it is just better at extracting it.

How can you be better at doing MS decoding?