[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

looking for a transformation to "extract" prosody from speech

Message en plusieurs parties et au format MIME.

------=_NextPart_000_0056_01BD79CB.D851BF80
Content-Type: text/plain;
        charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

Dear list,

I would like to know whether it is possible to identify one's language=20
(among others) on the basis of prosodic properties like intonation and =
rhythm.=20
Does anyone know if this issue has already been seriously investigated?=20

I am looking for transformations to apply to the speech signal=20
to keep its prosodic properties (i.e. intonation and rhythm) while=20
deleting (or replacing), as far as possible, the phones.

I have considered two ideas:=20
 1. use low pass filtering (under 300-400 Hz).=20
      =3D> but I would like stimuli that sound more like speech.=20
 2. use speech resynthesis, e.g. with a diphone synthesizer.
      =3D> but labeling the original utterances is too time-consuming.=20

Any suggestions?

Christophe Pallier,
LSCP, CNRS-EHESS
54 bd Raspail, Paris, France.


------=_NextPart_000_0056_01BD79CB.D851BF80
Content-Type: text/html;
        charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD W3 HTML//EN">
<HTML>
<HEAD>

<META content=3Dtext/html;charset=3Diso-8859-1 =
http-equiv=3DContent-Type>
<META content=3D'"MSHTML 4.72.2106.6"' name=3DGENERATOR>
</HEAD>
<BODY bgColor=3D#ffffff>
<DIV><FONT size=3D2>Dear list,</FONT></DIV>
<DIV><FONT size=3D2></FONT>&nbsp;</DIV>
<DIV><FONT color=3D#000000 size=3D2>I would like to know whether =
</FONT><FONT=20
size=3D2>it is possible </FONT><FONT size=3D2>to identify one's language =

</FONT></DIV>
<DIV><FONT size=3D2>(among others) </FONT><FONT size=3D2>on the basis of =

</FONT><FONT size=3D2>prosodic properties like intonation and rhythm.=20
</FONT></DIV>
<DIV><FONT size=3D2>Does anyone know </FONT><FONT size=3D2>if this issue =
has already=20
been seriously investigated? </FONT></DIV>
<DIV><FONT size=3D2></FONT>&nbsp;</DIV>
<DIV><FONT size=3D2>I am </FONT><FONT size=3D2>looking for =
transformations to apply=20
to the speech signal </FONT></DIV>
<DIV><FONT size=3D2>to keep its prosodic properties (i.e. intonation and =
rhythm)=20
while </FONT></DIV>
<DIV><FONT size=3D2>deleting (or replacing), as far as possible, the =
</FONT><FONT=20
size=3D2>phones.</FONT></DIV>
<DIV>&nbsp;</DIV>
<DIV><FONT color=3D#000000 size=3D2>I have considered two ideas: =
</FONT></DIV>
<DIV><FONT color=3D#000000 size=3D2>&nbsp;1. use low pass filtering =
(under 300-400=20
Hz). </FONT></DIV>
<DIV><FONT color=3D#000000 size=3D2></FONT><FONT=20
size=3D2>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; =3D&gt; but I would like stimuli =
that sound=20
more like speech. </FONT></DIV>
<DIV><FONT size=3D2>&nbsp;2. use speech resynthesis, e.g. with a diphone =

synthesizer.</FONT></DIV>
<DIV><FONT size=3D2>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; =3D&gt; but labeling =
the original=20
utterances is too time-consuming. </FONT></DIV>
<DIV><FONT size=3D2></FONT>&nbsp;</DIV>
<DIV><FONT color=3D#000000 size=3D2>Any suggestions?</FONT></DIV>
<DIV>&nbsp;</DIV>
<DIV><FONT size=3D2>Christophe Pallier,</FONT></DIV>
<DIV><FONT size=3D2>LSCP, CNRS-EHESS</FONT></DIV>
<DIV><FONT size=3D2>54 bd Raspail, Paris, France.</FONT></DIV>
<DIV><FONT size=3D2></FONT>&nbsp;</DIV></BODY></HTML>

------=_NextPart_000_0056_01BD79CB.D851BF80--