Hi, On 01/07/2011 12:33 AM, fons(a)kokkinizita.net wrote: [...] in ambdec, is the distance used for anything else then delay and gain compensation? Is it correct that for regular setups this does not matter? - Giso

On 01/07/2011 10:01 AM, Arnold Krille wrote: i guess you are hinting at jerome daniel's thoughts on near-field compensation? i guess his point was that you have to assume a "standard" distance when encoding, and the decoding would be wrong if your diameter is different. however, as you say, since the decoder knows the assumed standard, it can correct accordingly. for those non-ambiheads wishing to follow this discussion: the near-field effect in ambisonics results in a bass boost. it's caused by the speakers. ambisonic theory (without NFC) assumes plane waves, i.e. speakers that are very far away, so that the wave fronts are not curved. since they are curved in practice, you get the bass boost. the same thing is responsible for the well-known proximity effect on directional microphones. well, yes and no. the failure modes are different, too. depending on how 5.1 and 7.1 are mixed, you can usually place the speakers any old way and still get something out of it, only the source positions will be displaced. heck, ever seen those setups with all 5 speakers below the screen, sitting on top of the vcr? not much surround, but you still get the idea of the mix. in ambi, it tends to work well for minor displacements (which you can of course also correct in the decoder, but i'm assuming user errror here), but for large uncompensated placement errors, the whole reconstruction will fail. in this latter case, ambi will behave far less predictable than any discrete speaker technique. and people should not expect wonders from ambi rigs. the ITU 5.1 setup is the most irregular you should try - anything worse than that, and ambisonics won't be much fun. maybe when franz zotter and his friends from graz get their partial spheres decoder into a usable shape (right now it's pd magic only). it basically enables you to build only parts of a loudspeaker sphere and discard all direct sound coming from the wrong directions.

On 01/07/2011 12:44 PM, Arnold Krille wrote: you can get away with such things, especially in higher orders. but as you noted, it works only when the missing speaker is in particular places wrt the sound stage. since ambisonics is about isotropic reproduction, it does not work in the general case. and for bob's sake don't tell anybody you got away with it, because then everybody will want to do the same :) it's a dirty hack. i've done it in the past, sometimes it works, sometimes it doesn't. it depends on the source material. in a recent experiment with 8 line arrays, we played a stereo file over 3rd order, spreading the stereo base to 120 degrees. sounded great, fantastic envelopment, yet very clear, and we blew the paint off the wall with the systems. then we figured, hey, the opposite 4 arrays are about 30 dB down, why don't we just switch them off entirely? and this we did. still sounded almost as good as before. morale: maybe you can use ambisonic panning to drive a line of line arrays in order to cover a wide area, if and only if you only want to reproduce sources from one direction. but _THIS_IS_NOT_A_RECIPE_. it sounded cool when we did it, but in some other place with other source material, it may not work at all. maybe it only worked because we were in a reverberant space... and more importantly: this is no longer ambisonics. yes you do. you just need a reference, which means listening experience with proper setups. i feel i'm getting more confident with these cheats, because i jump at every opportunity to build or listen to a new rig and have heard some very good ones, and i always use the same demonstrably correct software, which helps. ambi rigs are a pain to debug, much more so if you don't have a precise idea of what they can do and what they can't do. therefore, soapbox preacher mode on: for your first ambi rig, don't mix and match speakers, don't build your own decoder, don't use source material recorded with your DIY tetrahedral mike. go the textbook way. only then start doing funny stuff, but with one component at a time, just like you would debug software. and then, maybe, when you come into a facility where nobody knows which channel order and normalisation they're using (hi guys, you know who you are), you can fix it, with some help from major ambisonic gurus and 20 hours of trial-and-error :)

On 01/07/2011 12:58 PM, fons(a)kokkinizita.net wrote: need to double check, but iirc i did hear a very pronounced (read: unbearable) boominess on my organ recordings when listening to them on my <4m array in second order without NFC... this i understand. this is news to me. where does this phase delay come from? interesting hypothesis. i guess the myth is more correctly formulated as "you can't localize anything except some select and very obvious mid-frequency cues in AB stereo". of course, bose and friends would jump on this to sell ridiculously small satellites with woofers going up to 300 hz :)

On Sat, Jan 08, 2011 at 02:14:58AM +0100, Arnold Krille wrote: It doesn't. Ciao, -- FA There are three of them, and Alleline.

On 01/08/2011 03:30 AM, fons(a)kokkinizita.net wrote: hmm. i need to think about this some more. you mean this one?: daniel/moreau, "Further Study of Sound Field Coding with Higher Order Ambisonics" i'm wading through it... :) enlightening. i always found it utterly counter-intuitive that LF boost was all there is to it, but i could never quite grasp why...

On 01/07/2011 09:10 AM, Giso Grimm wrote: it is also used for near-field compensation. in first order, you can get away without, since the effect (a noticeable bass boost) only kicks in below 50hz or so, but as soon as you move to 2nd or higher orders, you really want NFC.