Topic: All About Ambisonics

[fotoralf.be]

Did you know that one manufacturer offers an A- to B-format encoder that has a very, very low (2- to 3- millisecond) latency?

Who ist it?

Ralf

[Len Moskowitz (Core Sound)]

Who ist it?

PM'ed

[Len Moskowitz (Core Sound)]

Yesterday I recorded a large community choir with two higher order ambisonic mics spaced by 17 cm. I decoded the recording in post to a large number of different microphone configurations including ORTF, Blumlein, XY, XCY, binaural, bilateral ambisonics, LR with a few different angles and polar patterns, and LCR again with a few different angles and polar patterns.

I settled on LCR with 130° spacing and first order hypercardioids. It took about an hour to go through all of the possible decodes. Each decode took about 2 minutes to set up.

I recorded 16 channels with a Millennia Media HV-316 and a JoeCo BBR64-Dante.

[Gutbucket]

Streaming Ambisonics For Headtracked Binaural Playback

Did you know that one manufacturer offers an A- to B-format encoder that has a very, very low (2- to 3- millisecond) latency?

That makes it possible to stream real-time headtracked binaural to your listeners.

Does it use your calibration files or does it work well enough for streaming purposes without them?

If it doesn't use them, how good are the generic A-format to stereo/binaural? transcoders built into the Zoom and SoundDevices recorders, which I presume are intended for monitoring use?  I've never tried using that functionality of the F8 with TetraMic, partly because of the lack of applying the calibration corrections.  If decent enough that's one way of achieving very low latency 2-channel output without additional hardware or software, although without the headtracking.

[Len Moskowitz (Core Sound)]

Does it use your calibration files or does it work well enough for streaming purposes without them?

Yes, it uses the calibration files.

[kuba e]

Thank you for this topic! I also have an ambisonic microphone. My goal is to try to use it with a spaced pair. I would like to use it for stereo playback and also for quadrophonic playback. Unfortunately, I don't have much time to record and haven't tried the ambisonic yet. Hopefully I'll get around to it around Christmas and be able to write more.

But what I can advise a little about is a quadrophonic playback for those who do not have a surround system. We can easily extend our good stereo playback to quadrophonic. Nowadays, it is very easy and affordable. It is possible to buy a second pair of speakers and an amplifier in the yard sale. I recommend passive speakers and an amplifier that has a remote control. You can then easily change the levels of the rear speakers while listening. As a player, I bought a raspberry pi (I installed moode player on it) and usb soundcard that have line out for surround (eg. Xonar U7 or Motu M4).

The quadrophonic system is interesting for us, the audience tapers. It is simple - we are adjusting only the volume ratio of the front and rear speakers. And it can expand our recording capabilities. There are surround recordings on LMA. Many recordings are made as a matrix (e.g. GD). Soundboard goes to the front speakers, audience goes to the rear speakers. It's an interesting listen and quadraphonic gives us the freedom to change the sbd/aud ratio. I also found audience recordings there that were made specifically for surround (John Mayer). I have to admit, I was very impressed with it. But beware, Gutbucket also mentioned this on TS before, surround recording requires completely different mics setup and configurations than for stereo.

I'm quite curious when I use an ambisonic mic to see what the quadrophonic playback results will be like. I suspect that the time difference is important. Maybe an additional huge spaced pair could help a little bit. I will let you know when I manage to make the first attempts.

[Gutbucket]

^In addition to trying a 4-channel quad decode from your ambisonic mic alone, try a 2-channel stereo decode from the ambisonic mic routed to the front speaker pair and route the spaced pair to the rear speakers.  Spaced omnis work well for that and they doesn't have to be positioned super wide.  If using a directional spaced pair, point them to the rear or sides rather than forward, else you'll get too much front image blurring into the rear and in order to avoid that you won't be able to bring up the rear level as much as would otherwise be desirable.  Even omnis will often pickup a bit too much front content, so best to orient them in the same way when the intent is feeding surround channels.  A bit of delay to the rears can help with this too, not enough to be audible as such, just enough to keep the front image in front while allowing a bit more level in back.

[kuba e]

Thank you for your tips, I'll definitely give it a try. I also thought about the delay. I understand when the delay is small (no echo yet), the delayed source will disturb less the main pair. Then I can amplify the delayed source more in the rear speakers.

[Gutbucket]

Yes, a relatively short delay applied to the rear channels leverages the Hass effect to help perceptually keep the front image in front even though there is significant acoustic cross-talk from the front into the rear microphone channels.  Use just enough delay to help keep the front image in front without causing a perceivable echo, as short as possible to be effective, no more than 20-30ms.  EQing the rear channel signals carefully can also help.

That works for music playback in surround because other than specific audience noises and other specific random noises, there is no so much significant direct imaging content happening outside of the front quadrant, its mostly reverberant content, non-specifically distributed applause and general audience reaction around the sides and back which a touch of delay won't upset.  That stuff doesn't have to image in specific positions so much as just being in the correct general location.

The other thing you might try is starting with a quad decode from the ambisonic microphone to the four playback speakers, and mixing in a bit of the spaced microphone pair to both the front and rear speaker pairs to improve immersiveness.  Delay the portion sent to the rears a bit to keep the front image in front.

Better yet, if you have an addition pair of speakers and amp channels, place them to either side of the listening position and route the spaced pair just to them, with the quad decode from the ambisonic microphone feeding the front and back pairs.  You then get the good immersive quality of the spaced pair from the side speakers which are best positioned to convey that, and direct imaging of the ambisonic quad arrangement with increased front/back differentiation, all with less potential conflict as each signal is routed to its own playback speaker.  Can delay the spaced pair if necessary but probably won't need to as the image shift toward the side-positioned speakers will mostly just make the image wider rather than pulling it too far into the rears.

Fun stuff.

[Gutbucket]

There are surround recordings on LMA. Many recordings are made as a matrix (e.g. GD). Soundboard goes to the front speakers, audience goes to the rear speakers. It's an interesting listen and quadraphonic gives us the freedom to change the sbd/aud ratio. I also found audience recordings there that were made specifically for surround (John Mayer). I have to admit, I was very impressed with it.

I've not head the recordings of John Mayer you mention that were specifically recorded for surround playback and should seek them out, but I don't currently have my surround playback system setup.  I did go listen to a bunch of matrix encoded (mostly DTS) GD, Dave Mathews and some Panic taper recordings years ago when I was working up my surround recording methods.  Most of them were produced as you mention with dry SBD routed to the front channels and AUD microphone recording routed to the rear channels.  Honestly I wasn't overly impressed playing those back in quad the intended way.  Too dry in front, AND too much front leakage in back from the forward facing microphones.  They are better used as you mention as a way to dial in your own SBD/AUD ratio (effecting the direct/reverberant-sound ratio) into 2-channel playback, or by physically repositioning the rear quad speakers with the AUD content in them up front, pointed away from the listener so as to flood the front of the room with that content while the front speakers project the dry SBD content back toward you.  That physically mixes them "in the room" so to speak and avoids the front/back differentiation problem.

[kuba e]

Thank you Gutbucket! This is great inspiration. I am very happy to try it. At the end of the year I should go record a children's concert, I will try the ambisonic mic there. I will probably use it with the omni since they will ask me for a stereo recording. But then I'll play with the quad. I completely agree with you about sbd/aud surround recordings. Perhaps, apart from the choice of ratio, it had one more small advantage for me when compared with stereo matrix. It seemed to me that the optimal listening position has become larger. I'm a bit of a slob and I'm not always in the optimal place when listening. But it's just my subjective feeling, I haven't pay much attention to it.

John Mayer's surround recordings made a big impression on me. I think they show how much more options we have in audience recording. If anyone listens to it, please write, I will be glad if you confirm or deny it. I have to go back to those recordings and listen to it again. Unfortunately, I didn't give it much time back then.

Gd: https://archive.org/search?query=collection%3AGratefulDead+AND+NOT+collection%3Astream_only+surround+dts+
non Gd: https://archive.org/details/etree?query=%28%28surround+5.1%29+OR+%28surround+dts%29%29+AND+NOT+%28collection%3AGratefulDead%29
John Mayer: https://archive.org/details/etree?query=surround+John+Mayer

I'm looking forward to trying out the ambisonic mic for the quad. Hopefully it won't take me long time.
And what convinced me one hundred percent for the quad (unfortunately, it is no related to audience recording) was listening Miles Davis' Bitches Brew album.

[wforwumbo]

To follow up here, I want to encourage tapers to try coincident mics in between a spaced AB pair. I’ve done this with subcards as the spaced pair and a mid side (which is a simplification of ambisonics in some ways) pair between them with great results. Both pairs need massaging but with patience and a good ear it can produce great results.

For proof of validation of the concept you can hear what the above method sounds like on stereo: phish.in/2019-08-31

I think ambisonics between a spaced pair has the potential to solve many issues i have with either format on their own. If another taper wants to give this a go, shoot me a PM and I’ll offer some insights into how I think the tapes can be made slicker.

[Len Moskowitz (Core Sound)]

To follow up here, I want to encourage tapers to try coincident mics in between a spaced AB pair.
...
I think ambisonics between a spaced pair has the potential to solve many issues i have with either format on their own. If another taper wants to give this a go, shoot me a PM and I’ll offer some insights into how I think the tapes can be made slicker.

Ambisonic mics decoded to virtual microphones can replace - with better performance - traditional mono mics in AB pairs and other spaced arrays. And that's in addition to them being able to perform as any coincident array you can imagine.

(They do require more recording channels, but these days channels are inexpensive.)

[Gutbucket]

Many TS readers will be familiar with my common refrain here at TS of recommending a coincident pair in the center between a spaced pair (generally spaced around twice the typical distance one would use for just 2 spaced channels on their own) to tapers interested in trying a 4 channel microphone array with stereo playback as the end goal.  I feel that represents something of a sweet spot for the use of four microphone channels in keeping complexity and channel count reasonable while reaping a good share of the benefits that a multichannel array can provide for stereo concert playback.

To help tapers with that, I'm still working on a 4 channel "Improved PAS table", similar to the 2 channel Improved PAS table linked in my signature, which will suggest specific spacings based on the PAS angle. (FYI, the link to the OMT PDF in my signature is currently well out of date and is also being worked on, a new version of which includes this arrangement of a coincident pair between a spaced pair as the primary 4 channel OMT4 arrangement).  This is based more on "Stereo Zoom" methodology and the needs of "taping from various distances", and may vary somewhat from wforwumbo's primary concerns - although we have discussed and agree on the generalized underpinnings of both approaches.

An further extension of that arrangement to 6 recording channels, if 6 are available and acceptable, is to replace the center coincident pair with a 1st order ambisonic microphone that will use 4 channels of the 6, as wforwumbo mentions.  This arrangement retains the same coincident center plus spaced pair 3-point in space, while providing dramatically increased control of angle and pattern over the virtual forward-facing center coincident stereo-pair derived from the ambisonic microphone, and additionally allows for deriving a rear facing virtual microphone or coincident pair as well.  Its not overly complicated to add a pair of spaced omnis with the ambisonic microphone in the center.

Because its been many years since I explored this option (and quickly moved on to using multichannel near-spaced arrays in the center at the time, driven primarily by optimizing for surround playback specifically) any recordings you might make using this method that you wish share should help provide welcome data points, Kuba e.

[Gutbucket]

Len raises a good point about the possibility of using ambisonic microphones in the spaced positions, although in doing that the channel count requirement grows quickly.  Extending the center coincident + spaced pair arrangement by using 1st order ambisonic microphone in all three positions would require 12 channels.  Used in the spaced positions to either side of a standard 2ch coincident pair in the center, 10 channels.  It is a very attractive proposition to be able to vary angle and pattern of the wide spaced positions on mixdown however, with full control over angle and pattern, and the ability to derive both front and rear-facing spaced pairs.

A more productive use for 8 channels might be a single ambisonic microphone in the center between a spaced pair of omnis with each omni configured as a Strauss-packet (a coincident 2ch arrangement of an omni and a forward facing fig-8).  That provides full ambisonic coincident center, along with choice of pattern and of pointing the spaced pair forward or backward (or both), lacking only fine angle adjustment over the spaced pair.  Maximal flexibility within a practical limit of 8 channels, while still retaining the basic 3-point arrangement of coincident center + spaced pair.

The all-ambisonic approach when limited to 8 channels would either be a spaced pair of 1st order ambisonic microphones, sampling 2-points in space (perhaps optimal for head-tracked binaural if limited to 8ch?), or a single higher spatial resolution 2nd order ambisonic microphone such as the OctoMic, which would be limited to sampling a single-point in space.

I currently sort of go the opposite way, using an 8ch OMT8 array primarily based on spacing, sampling 7 separate in space with the center position being a coincident Mid/Side pair.  I do have plans to try replacing 4 of those channels - the center Mid/Side and rear-facing near-spaced supercardioid pair - with the TetraMic in the center position at some point. 

Not sure about harder, but that should be smaller and faster.. will it also be better and stronger, Daftpunk?