[MSNoise] Fwd: MSNoise: filters

[Thomas Lecocq]

Hi Hélène,

The filters need to be defined in the configurator. Filter bounds define 
the bandpass that will whitened. Around this whitened part, the spectra 
will be tapered to 0.+0j.

Indeed, I've rechecked the doc, it's not very obvious. Thanks for 
pointing this out :) I'll try to spend some time on the doc soon.

Briefly: each filter is defined with:
low : lower frequency of the bandpass
high : upper ...

mwcs_low : slightly higher than low, it is the lower frequency for which 
MWCS' weighted linear regression will be estimated
mwcs_high : slightly lower than high,...

mwcs_wlen : window length for computing the relative delay in the MWCS
mwcs_step : step between two windows. if equal to mwcs_wlen: no overlap.

rms_threshold: not used any more, should disappear, set it to 0.0.

Hope this helps, let me know (via this mailing list, svp.)

Thom

-------- Message original --------
Sujet: 	MSNoise: filters
Date : 	Sat, 8 Mar 2014 14:14:33 +0100

	
Pour : 	Thomas Lecocq <thomas.lecocq at oma.be>



Thomas, we solved the problem with scikits_samplerate apprently related to
some export in my .bashrc.
I am now  at s03compute_cc.py and get an error because I did not define any
filters: apparently this is not implementted yet (?) , to what these filters
correspond?
Cheers
Helene

-----Message d'origine-----
From: Thomas Lecocq
Sent: Thursday, March 06, 2014 5:45 PM
To: Helene.Lyon-Caen at ens.fr
Subject: Re: scikits

Hélène,

You could remove the dependency to scikits by using only Decimation, if
all your data are in the same sampling rate !

Thom

Le 06/03/2014 16:18, Helene Lyon-Caen a écrit :
> Thomas,
> we have been trying to understand with the computer engineer here. The
> only differences with what is on your web site we could find is that we
> use a more recent version of anaconda (1.9) and of python (2.7.6)
> could that make a difference?