[Fwd: Re: [Ffmpeg-devel] Ambisonic decoder]

[Alex Beregszaszi]

Hi,

> <alex at rtfs.hu>:
> > Actually that means you only need to write two filters:
> > ambisonic_encoder and ambsonic_decoder. Every sane compressor should
> > work on the preprocessed data.
> 
> It is indeed a good idea to implement the Ambisonic transcoder as a
> filter. But when a codec detects the stream as Ambisonically encoded,
> the filter (which is always on) should change its behaviour. We can
> have 5.1 ITU sources as well as 6ch Ambisonic B-Format. Thus,
> interaction between codec and filter is essential.

Most sane codecs implement a data type field, like pixel format /
colorspace for video codecs. Sampling rate and channel number could be
used to detect this in older codecs.

> > Is there a standard ambisonic audio format + codec which is preferred by
> > the authors of it? Or it just uses pcm+wav?
> 
> Currently they use Microsoft(r) WAVE-EX. See
> http://www.ambisonicbootlegs.net for examples. There are also
> discussions regarding the incoming Ambisonic Vorbis standard on
> vorbis-dev.

Can you upload such .wav and vorbis file to mplayerhq samples archive?

--
Alex Beregszaszi