[Discuss-gnuradio] introduction + question re: optimal demod for audio stream analysis

[m. cera]

hi all,
just wanted to say hello & offer a quick introduction...
i'm a sound / media artist, currently enrolled in the digital + media  
mfa program
at the rhode island school of design; for my thesis work, i'm  
developing a system
with which i will sonify / visualize data (primarily density levels)  
from cellular networks (i.e. gsm)
to create realtime performances and audiovisual installations using  
the usrp + gnu radio.

i'm running gnu radio (3.0.4) on fc7 w/ the planet ccrma kernel,
and have been successful receiving data using the usrp with AM / FM /  
WFM demodulation,
and sending it out via alsa to another machine running supercollider  
3 for analysis / resynthesis.
ultimately, i am planning on sending the data to supercollider via jack,
then onto max/msp/jitter on the other machine via open sound control
for 3d visualization using its implementation of openGL.

the audio stream generated from usrp_rx_nogui.py using AM demodulation
is fairly reliable in terms of spectral analysis: the cell traffic is  
generally pretty discernible,
and while the noise floor is decently high, it could be compensated  
for somehow.
but it seems like it might make more sense to be using GMSK  
demodulation...
(FM / WFM have lower noise floors, but seem to require too tight of a  
radius to be useful)

however, i'm hitting a brick wall when it comes to trying to rectify  
the differences
between the demodulation methods in usrp_rx_nogui.py and  
benchmark_rx.py / rx_voice.py...
i also noticed that, in gmsk.py, the comments for the GMSK  
demodulator block
say that the output is a stream of bits packed in the LSB of each  
byte it streams out.

but of course, a stream of bits in the LSBs of a bunch of bites does  
not a great audio signal make...
and while i'd originally thought i'd be streaming out data files,  
then using that data to drive a synthesis engine written in  
supercollider,
it seems to makes more sense to keep it all in an audio stream, as it  
is conveniently packaged for my needs,
and the latency is much lower than would be the case otherwise.

and since i'm not trying to transmit or receive specific data, just  
trying to analyze the level of activity
in specific ranges of the spectrum in a given location, maybe it  
doesn't make sense to bother with GMSK.
at any rate, any insight or advice you might have regarding optimal  
strategies for demodulation
in regards to this project would be greatly appreciated.

thanks in advance,
--mark

--
m. cera | c3ra.com