<?xml version="1.0" standalone="yes"?>
<Paper uid="W02-0706">
  <Title>Architectures for speech-to-speech translation using finite-state models</Title>
  <Section position="3" start_page="0" end_page="0" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"> Present finite-state technology allows us to build speech-to-speech translation (ST) systems using ideas very similar to those of automatic speech recognition (ASR). In ASR the acoustic hidden Markov models (HMMs) can be integrated into the language model, which is typically a finite-state grammar (e.g. a N-gram). In ST the same HMMs can be integrated in a translation model which consists in a stochastic finite-state transducer (SFST). Thanks to this integration, the translation process can be efficiently performed by searching for an optimal path of states through the integrated network by using well-known optimization procedures such as (beam-search accelerated) Viterbi search.</Paragraph>
    <Paragraph position="1"> This &amp;quot;integrated architecture&amp;quot; can be compared with the more conventional &amp;quot;serial architecture&amp;quot;, where the HMMs, along with a suitable source language model, are used as a front-end to recognize a sequence of source-language words which is then processed by the translation model. A related approach has been proposed in (Bangalore and Ricardi, 2000; Bangalore and Ricardi, 2001).</Paragraph>
    <Paragraph position="2"> In any case, a pure pattern-recognition approach can be followed to build the required systems.</Paragraph>
    <Paragraph position="3"> Acoustic models can be trained from a sufficiently large source-language speech training set, in the very same way as in speech recognition.</Paragraph>
    <Paragraph position="4"> On the other hand, using adequate learning algorithms (Casacuberta, 2000; Vilar, 2000), the translation model can also be learned from a sufficiently large training set consisting of source-target parallel text.</Paragraph>
    <Paragraph position="5"> In this paper, we comment the results obtained using this approach in EUTRANS, a five-year joint effort of four European institutions, partially funded by the European Union.</Paragraph>
    <Paragraph position="6"> Association for Computational Linguistics.</Paragraph>
    <Paragraph position="7"> Algorithms and Systems, Philadelphia, July 2002, pp. 39-44. Proceedings of the Workshop on Speech-to-Speech Translation:</Paragraph>
  </Section>
class="xml-element"></Paper>