No subject

Use a few low level features, such as energy of low and high frequencies ba=
nds, spectral spread. In a second step, these approaches are often refined =
using machine learning techniques bayesian networks or support vector machi=
nes.<br>


<br>
See for instance these papers:<br>
<a href=3D"http://cobweb.ecn.purdue.edu/~malcolm/interval/1996-085/" target=
=3D"_blank">http://cobweb.ecn.purdue.edu/~malcolm/interval/1996-085/</a><br=
>
<a href=3D"http://www.aclweb.org/anthology/O/O08/O08-1015.pdf" target=3D"_b=
lank">http://www.aclweb.org/anthology/O/O08/O08-1015.pdf</a><br>
<a href=3D"http://www.hindawi.com/journals/asp/2009/628570.html" target=3D"=
_blank">http://www.hindawi.com/journals/asp/2009/628570.html</a><br>
<br>
i would like to achieve &gt; 90% of accuracy if possible. any suggestions a=
re welcome!<br>
<br>
_______________________________________________<br>
<a href=3D"mailto:Pd-list at iem.at" target=3D"_blank">Pd-list at iem.at</a> mail=
ing list<br>
UNSUBSCRIBE and account-management -&gt; <a href=3D"http://lists.puredata.i=
nfo/listinfo/pd-list" target=3D"_blank">http://lists.puredata.info/listinfo=
/pd-list</a><br>
</blockquote></div><br><br clear=3D"all"><br>-- <br>Pedro Lopes (MSc)<br>co=
ntact: <a href=3D"mailto:pedro.lopes at ist.utl.pt" target=3D"_blank">pedro.lo=
pes at ist.utl.pt</a><br>website:=A0<a href=3D"http://web.ist.utl.pt/Pedro.Lop=
es" target=3D"_blank">http://web.ist.utl.pt/Pedro.Lopes</a>=A0/=A0<a href=
=3D"http://pedrolopesresearch.wordpress.com/" target=3D"_blank">http://pedr=
olopesresearch.wordpress.com/</a>=A0|=A0<a href=3D"http://twitter.com/plope=
sresearch" target=3D"_blank">http://twitter.com/plopesresearch</a><br>


</div>

--0015174c3ca4c91f31049bb58cf8--