Stereo2surround using Ambisonics: an important question

[nugi]

My question applies to the various stereo-to-surround methods proposed in this forum. It also applies to general Ambisonics encoding for output on a 5.1 system.

When one uses for instance Plogue with the Ambisonics encoder vst-plugins, one can choose to output the signals to for example a 5-channel speaker arrangement system, but: if the goal is to play the signal (music etc) over a 5.1 system (in general DD, DTS) then there should have been a plugin that adapted the signal to a 5.1 system. An Ambisonics encoder assumes that the speakers be(eqully spaced) in a circle. A 5.1 system is not!

Does anyone know how to do this?

I would say that that this is a really important issue. Soundfield has released a plugin for this: http://www.soundfield.co.uk/news/news6.php . This is an expensive plugin (and not for Plogue). Anyone who has an idea how to adapt the Ambisonics signals to 5.1 (or a nice plugin etc.) please share your thoughts!

Thanks.

[ursamtl]

Hi Nugi,

Strictly speaking, Ambisonic speaker layouts are designed to be equidistant from a central listening point. However, you may get good results with a speaker layout that breaks this "rule." If you want to develop a bidule decoding approach, you could use the formula found at Ambisonics - Wikipedia (http://en.wikipedia.org/wiki/Ambisonics). Simply use the formula five times (ONCE for each of the five main speakers) and use the following angles for the speaker placement (based on the ITU specs for 5.1): L - 30°, R - 330°, C - 0°, sL - 110°, sR - 250°. Your trusty Windows calculator can be switched to Scientific view to get the sin and cos functions.

If you need LFE, use a crossover to remove the frequencies below say 80Hz from each of the five channels, then send them combined through the LFE output.

As for VSTs, you can play around with the free Emigrator VST (http://www.gerzonic.net/?prod=5&rub=2). There's also another one I found several months ago called B-Format to Surround, but for some reason I cannot find the link to it anymore. It was from the York University music department and was mentioned in the Sursound mailing list this summer, but it's not listed with other york plugins. If I find it again, I'll add the link to the guide list. As I remember, the plugin wasn't bad but I was in the middle of developing V.I at the time so I kind of forgot about it.

The more I dig into the whole field of Ambisonics, the more schizophrenia I find! If one reads a solid introductory article such as Richard G. Elen's Ambisonic Surround-Sound in the Age of DVD (http://www.ambisonic.net/ambidvd.html), it becomes clear that Ambisonics is designed for material that's specially recorded with Ambisonic decoding in mind.The primary method of capturing the original performance "Ambisonically" was to be a special new kind of microphone. Called the Soundfield Microphone, the device included a tetrahedral array of capsules in a single enclosure. This would suggest that simply encoding and decoding regular stereo files does not produce the desired result. Yet, there are VST encoding plugins available with inputs marked L & R! After reading lots about this subject over the past months and also corresponding with Angelo Farina, a leading expert in the field, I still don't believe that feeding a stereo file into an Ambisonic encoder/decoder will result in a truly ambisonic soundfield. Farina himself confirmed this in email to me. He did say that good results can be obtained by convoluting the file with impulse responses that basically "graft" the missing ambisonic information onto the stereo file. This is in fact where the idea for the very first guide to upmixing on this forum came from, Farina's web site for Aurora (http://www.ramsete.com/aurora/).

My only problem with this approach is that it's like running the file through an artificial reverb unit. One permanantly imposes one specific sound environment on the recording. I'd rather find a way that extracts the ambience information of the original recording. I felt I accomplished that somewhat with the first version of the V.I bidule and the next version will take this even further.

Anyway, the point is to have fun with this stuff, don't worry about the "rules." Since what we're doing is for amusement, go ahead and try anything that works for you!

Regards,
Steve.

[nugi]

Thanks alot for your post. I really appreciate it. I do think it would be a bit difficult to implement the equation, but I'll try some way or another. Now I'm going to do a recording at the concert hall with the soundfield mic and afterwards I'll try to "adapt" the Ambisonics signals to 5.1 output.

Thanks again!

If anyone has more thoughts upon this subject - please share them.

[ursamtl]

Originally posted by nugi
Thanks alot for your post. I really appreciate it. I do think it would be a bit difficult to implement the equation, but I'll try some way or another. Now I'm going to do a recording at the concert hall with the soundfield mic and afterwards I'll try to "adapt" the Ambisonics signals to 5.1 output.

Thanks again!

If anyone has more thoughts upon this subject - please share them.

Glad I could help. The equation is actually quite simple. I threw together a 5.1 Ambisonic Decoder bidule group after coming home from work today. You can download it HERE (http://stevethomson.ca/audio/guides/5.1_Ambisonic_Decoder.zip). It seems to map sounds quite well. You can test it feeding a single channel to one side of the bpan or Panoram VST encoding plugin and then running the first three outputs through my decoder. As you adjust the panning or azimuth control, the sound travels evenly around the 5 speakers. I also implemented an LFE crossover at 60Hz, which is well down out of the way of bass management crossovers.

And hey, if you've got a soundfield mic then you're really ready to go! Just feed the W X and Y signals through this decoder and tell us all what happens! Obviously there's no height channel Z since 5.1 doesn't allow for height. Still, with properly recorded material, you should be able to achieve really good surround.

I look forward to hearing about your results!

Enjoy!

[Umma]

Nugi,

Is this (http://www.ramsete.com/aurora/conversion_between_uhj_and_b.htm)what you're looking for? Go about 3/4 of the way down the page.

Regards,

Umma

[nugi]

Thanks ursamtl and Umma. I don't think your 5.1 Ambisonic Decoder bidule group will work good enough ursamtl, because there is a lot more computation when going from "equidistant" to 5.1 arrangement. I talked to a professor in electroacoustics and he told me so. I also talked to the author of CATT (www.catt.se) and he helped me with a plugin to his program that does adapt the signals to a 5.1 speaker arrangement. I also think the "surround" speaker arrangement in the gerzonic plugin might work (the link in Umma's post).

[ursamtl]

No problem. However, the actual ITU 5.1 layout I used for the computations is, in fact, equidistant from the listening point. The speakers are placed at the following angles (counterclockwise from front center): C: 0°, L: 30°, sL: 110°, sR: 250°, R: 330°. Mind you, many 5.1 systems are not set up exactly this way, but this is the recommended ideal. For example, a lot of home theater systems place the C speaker on the same plane as the L and R, thus causing a slight distortion of the theoretical soundstage. If you calculate the speed of sound, this could be compensated with a delay of 2-4ms, depending on the size of the listening environment. So, it's not a big deal.

Did you share the formula at Wikipedia (http://en.wikipedia.org/wiki/Ambisonics) with the professor? It's a sound mathematical equation I've seen in quite a few Ambisonics-related documents written by engineers.

Emigrator is a good plugin and, as mentioned on the gerzonic.net web site, it is based on the equations found at First and Second Order Ambisonic Decoding Equations (http://www.muse.demon.co.uk/ref/speakers.html). As is pointed out on the site, the "Surround" layout can result in an unstable sound image. Moreover, if you plot the speaker positions given for the Surround layout, you'll find that they located the surrounds at 120° and 240°, which deviates from the ITU 5.1 spec. However, their positions are still equidistant from the listening point.

Anyway, good luck and be sure to let us know how your recordings come out.