tag:blogger.com,1999:blog-316480898443887142024-03-08T05:24:10.021-08:00Speech PortalSophisticated spoken dialog systems platformprimetalk.ruhttp://www.blogger.com/profile/08997224460785903714noreply@blogger.comBlogger1125tag:blogger.com,1999:blog-31648089844388714.post-31565732543975475672013-09-19T05:28:00.001-07:002013-09-19T05:29:13.434-07:00Speech Portal<div dir="ltr" style="text-align: left;" trbidi="on">
<div>
<a href="http://www.primetalk.ru/en/technologies-en.html">Primetalk's Speech Portal</a> is a platform for constructing sophisticated spoken dialog systems. It uses <a href="https://github.com/Primetalk/SynapseGrid/">SynapseGrid</a> as an integration platform and contains all the necessary components for a dialog system to speak with a human being:<br />
<ol style="text-align: left;">
<li>Pipeline:</li>
<ol>
<li>Speech signal processing.</li>
<li>Recognizer.</li>
<li>Parser.</li>
<li>Dialog manager.</li>
<li>Templater.</li>
<li>Synthesizer</li>
</ol>
<li>Knowledge representation</li>
<ol>
<li>A priori </li>
<ol>
<li>Grammar expressions for input and output language.</li>
<li>Owl/rdf ontology import.</li>
</ol>
<li>A posteriori</li>
<ol>
<li>Strictly-typed frames.</li>
<li>Untyped frames. </li>
</ol>
<li>Partial (at runtime) and uncertain</li>
<ol>
<li>Fuzzy sets.</li>
<li>Incomplete frames.</li>
<li>Probability distributions.</li>
</ol>
</ol>
<li>Basic dialog logic building blocks</li>
<ol>
<li>User input comprehension.</li>
<li>Context representation.</li>
<li>Output utterance proposals.</li>
</ol>
<li>API for custom dialog logic</li>
<ol>
<li>Extended Petry nets.</li>
<li>Composable functional speaking strategy.</li>
</ol>
<li>API for real-time custom reactions.</li>
<ol>
<li>Selected set of <a href="https://github.com/Primetalk/SynapseGrid/">contacts</a> (hooks) for getting all interesting real-time events.</li>
<li>Carefully designed replaceable parts of pipeline.</li>
</ol>
</ol>
</div>
For recognizer part we support:<br />
<ol style="text-align: left;">
<li>sphinx4 + our acoustic model or any voxforge models.</li>
<li>google speech api (unofficial).</li>
</ol>
For speech synthesis:<br />
<ol style="text-align: left;">
<li>Festival + voice.</li>
<li>OpenMary + voice.</li>
<li>RhVoice + voice.</li>
<li>SpeechPro + voice.</li>
</ol>
<br /></div>
primetalk.ruhttp://www.blogger.com/profile/08997224460785903714noreply@blogger.com5