Remains the question of _how_ to actually use the available
signals to produce anything surround. Linear beamforming is
limited to LF (due to aliasing). The only way I see is some
rather complicated non-linear 'logic' decoding. This kind of
thing is still a research topic.
They could apply some form of DirAC processing. If they really do this,
they probably would separate the 8-channel audio stream of the cubic
"hedgehog" mike into direct and indirect elements. (Probably) separating
the audio stream into time/frequency tiles, as "standard" (B format)
DirAC does.
Even if this might sound a bit far-fetched: There are direct
institutional and personal links between Nokia Research and Aalto
University. (They are kind of neighbours. The common areas are spatial
audio and VR research, in this case.)
There could be other forms of signal processing involved, not only
DirAC. In any case it seems to be very possible what I have "proposed".
It does. S/N ratio for microphones is by convention the noise
level referred to the output level for 1 Pa SPL or 94 dB SPL.
94 - 64 = 30. This figure is very typical for MEMS mics. Even
cheap consumer grade electrets will be 10 dB better on noise.
And 120 dB is a typical AOL (Acoustic Overload Point) for MEMS,
the level at which you can expect 10% harmonic distortion.
As 64dB S/N ratio is not good enough for music recordings - is it even
enough for ambient sound recording? I believe that Nokia could and
should improve this figure.
Or is 64dB S/N the value which remains < after signal processing >?
(Though DirAC techniques don't reduce S/N values at all....)
Does anybody of our Finnish colleagues on this list have some more
specific ideas or knowledge? If so, it would be quite nice if you could
share your wisdom with us. :-)
Best regards,
Stefan
Fons Adriaensen wrote:
On Sat, Dec 05, 2015 at 01:28:11AM +0000, Stefan Schreiber wrote:
I don't believe a S/N ratio of 64dB will imply "30 dB self-noise
caused by crappy MEMS capsules", at least not here.
It does. S/N ratio for microphones is by convention the noise
level referred to the output level for 1 Pa SPL or 94 dB SPL.
94 - 64 = 30. This figure is very typical for MEMS mics. Even
cheap consumer grade electrets will be 10 dB better on noise.
And 120 dB is a typical AOL (Acoustic Overload Point) for MEMS,
the level at which you can expect 10% harmonic distortion.
I guess the noise is introduced because of interference between
the capsules, which means sound from any direction will reach
always more than 2/3/4 capsules.
That does not add noise.
Anyway: This microphone is probably very able to record "location
sound", whereas the direct sound (and say voices) can be mixed into
this "background" at a later stage.
(A pretty normal procedure in the area of film sound, by the way?)
Exactly. Which is why the OZO's audio part can have these sort
of specs - it's not meant to be used on its own.
Remains the question of _how_ to actually use the available
signals to produce anything surround. Linear beamforming is
limited to LF (due to aliasing). The only way I see is some
rather complicated non-linear 'logic' decoding. This kind of
thing is still a research topic.
Ciao,
_______________________________________________
Sursound mailing list
Sursound@music.vt.edu
https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe here, edit
account or options, view archives and so on.
