[Ardour-Users] H2N Google Spatial sound support

Jörn Nettingsmeier nettings at stackingdwarves.net
Thu Jun 9 11:47:10 PDT 2016


On 06/09/2016 08:10 PM, Gunter Königsmann wrote:
> Dear all,
>
> I have found a audio format of which I don't know if it will gain
> relevance and which currently is only partially supported by ardour:
>
> The long part of the story:
> My H2n field recorder contains 5 microphones providing MS Stereo (at the
> front of the device) and XY-Stereo (at the back of the device) and was
> always able to record both things simultaneously into two separate
> stereo files. As Ardour can merge them during import this is no big
> problem (that means: If any of you knows how to teach faac to put
> information about the right speaker positions into the resulting .m4a
> file...). After a firmware update it was able to record surround audio
> into a single file. I automatically assumed that it would just put all 4
> channels into a single file.
>
> Turns out that the device instead saves surround audio in a somehow
> peculiar format instead:
>  - The file contains 4 tracks
>  - One track is the center Channel
>  - One track the difference between Left and Right
>  - One track the difference between Back and Front
>  - and the waveform of one of the four tracks looks like it is filled
> with zeros.

This is called Ambisonics B-format, or rather, the first channels of an 
AmbiX channel sequence. In traditional nomenclature, they are W, Y, Z, 
and X. Z would be the height component, which the H2 cannot record, so 
it is left empty. The choice of file format is for compatibility with 
Google's 360° format for YouTube.

This format also defines the necessary metadata. The H2 firmware update 
comes with an addendum to the manual in PDF format that describes the 
details.

> I assume the file will be equipped with special metadata similar to what
> can be seen here:
> https://github.com/google/spatial-media/blob/master/docs/spatial-audio-rfc.md
>
> As the new format reduces 5 channels to 3 I am convinced using it looses
> more information than the 4-track-in-two-files-approach would => I
> personally won't use it. But I wanted to inform you that it exists in
> case it turns out relevant enough (surround might be part of the future)
> that you might want to support it anyway.

It loses information in the sense that the mic array in the H2 is a 
rather crude approximation of a B-format microphone in the sense that it 
is nowhere near co-incident. But for a truly coincident array, three 
channels carry all the relevant information for horizontal surround.

By using the native four-channel format, you retain a tiny bit of 
run-time difference between the microphones, but it is too small to be 
practically relevant.


All best,


Jörn



-- 
Jörn Nettingsmeier
Lortzingstr. 11, 45128 Essen, Tel. +49 177 7937487

Meister für Veranstaltungstechnik (Bühne/Studio)
Tonmeister VDT

http://stackingdwarves.net




More information about the Ardour-Users mailing list