First European Conference on Speech Communication and Technology

Paris, France
September 27-29, 1989

A Massively Parallel Model of Speech-To-Speech Dialog Translation: A Step Toward Interpreting Telephony

Hiroaki Kitano, Hideto Tomabechi, Teruko Mitamura, Hitoshi Iida

Center for Machine Translation, Carnegie Mellon University, Pittsburgh, Pennsylvania, USA

This paper describes the overall picture of #DmDialog. #DmDialog is a real-time Japanese-English speech-to-speech dialog translation system that accepts speaker- independent continuous speech inputs. The scientific focus of the project is to model the cognitive process of simultaneous interpreters. As a result, the architecture of the system is very different from machine translation systems. Our model assumes hybridized parallelism as a basic computation mechanism, and the process of translation is highly interactive due to the dynamic participation of knowledge from morphophonetic-level to discourse-level. An almost concurrent parsing and generation scheme provides a simultaneous interpretation capability which is essential to interpreting telephony. #DmDialog has been publicly demonstrated at the Center for Machine Translation at Carnegie Mellon University.

Full Paper

Bibliographic reference.  Kitano, Hiroaki / Tomabechi, Hideto / Mitamura, Teruko / Iida, Hitoshi (1989): "A massively parallel model of speech-to-speech dialog translation: a step toward interpreting telephony", In EUROSPEECH-1989, 1198-1201.