[linux-audio-dev] 2-channel stereo compatible ambisonics...

[Steve Harris]

On 14 Mar 2007, at 10:10, Joern Nettingsmeier wrote:

> Steve Harris wrote:
>> On 13 Mar 2007, at 09:15, Valent Turkovic wrote:
>>> Hi,
>>> is it possible to record multi channel speech audio and down  
>>> sample it
>>> 2 stereo channels BUT with horizontal audio positioning of each  
>>> audio
>>> channel ie. like dolby surround effect.
>> Sure, but you might be disappointed by the results, Pro Logic is  
>> not very sophisticated (eg. there's no separate rear left and  
>> right channel, and the bandwidth is limited). Also the only free  
>> software implementation that I know of is not state of the art,  
>> partly due to patent problems, and partly due to lack of interest.
>> There's a LADSPA plugin: http://plugin.org.uk/ladspa-swh/docs/ 
>> ladspa-swh.html#id1401
>> which does Pro Logic matrix encoding, you will need some other  
>> software to mix your independent streams into LCR channels (you  
>> can do it by ear with a 3+ variable send bus mixer), but once you  
>> have that you can feed it to the plugin and get a pro logic  
>> compatible stream out.
>> PS ignore the documentation, it has been tested, and it does work,  
>> just not brilliantly. You will get better results if you EQ the  
>> mics a bit to cut the level of the highs and lows a bit,  
>> especially in the rear channel.
>> You would get much better results with an ambisonic stream, but I  
>> don't /think/ you can make stereo-compatible ambisonic streams,  
>> and not many people have ambisonic decoders. Other people on this  
>> list know a lot more about ambisonics though.
>
> i'm very new to ambisonics, so i may be wrong, but i think there  
> are indeed stereo compatible first-order ambisonics streams. iirc,  
> the format is called UHJ, and it matrixes WXY (lossily) into L/R  
> (i.e. no height information).
>
> i've changed the subject to bait fons :-D
> if we're lucky he knows an implementation of 2-channel UHJ.
>
> i found this article: http://www.ambisonic.net/ambi_AM91.html
> but it does not give details as to how the matrix encoding is  
> accomplished.

Hi Joern,

Hmm, interesting, though it sounds like you'd need a specialised UHJ  
ambisonic decoder to play it back. That probably narrows the audience  
even further.

- Steve