<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859-1">
<META NAME="Generator" CONTENT="MS Exchange Server version 5.5.2654.45">
<TITLE>Integration with external ASR engines</TITLE>
</HEAD>
<BODY>

<P><FONT SIZE=2 FACE="Arial">Hello,</FONT>
</P>

<P><FONT SIZE=2 FACE="Arial">Question for developers: what is the asterisk way to integrate with ASR (speech recognition)?</FONT>
<BR><FONT SIZE=2 FACE="Arial">Question to the community: has someone done anything in this direction?</FONT>
</P>

<P><FONT SIZE=2 FACE="Arial">On the first glance that shouldn't be too hard. One part is delivering audio to the engine (for example,</FONT>
<BR><FONT SIZE=2 FACE="Arial">main ASR players Nuance and Speechworks will be happy with RTP) - can be done via RTP forking.</FONT>
<BR><FONT SIZE=2 FACE="Arial">The other part is stopping the prompt when speech is detected (barge-in implementation). One more</FONT>
<BR><FONT SIZE=2 FACE="Arial">thing is adding the related commands to agi.</FONT>
</P>

<P><FONT SIZE=2 FACE="Arial">Any comments are welcome.</FONT>
</P>

<P><FONT SIZE=2 FACE="Arial">Thank you.</FONT>
<BR><FONT SIZE=2 FACE="Arial">Alex Zarubin</FONT>
</P>
<BR>

</BODY>
</HTML>