The INSPIRATON project is part of the Knet COAST, and in some aspects, it is a continuation of OFAI's work in the SPARC project (funded in the FIT-IT program "Semantic Systems"). As automatic speech recognition (ASR) systems so far do not exhaust the potential of automatic text processing and semantic technology, the INSPIRATION project aims at modelling and supporting document creation processes aided by speech recognition. The first application area is the production of medical reports.

In today's speech recognition applications, particularly in the medical domain, a manual correction of the created reports is required. Correcting the speech recognition errors is less time-consuming than reformulating and formating the dictated text. The reason for this is that documents have to conform to specific standards or user preferences, which in most cases cannot be gleaned directly from the dictated, raw text. This leads to the situation that only for users who are very proficient in dictating, the system is able to take full advantage of automatic speech recognition and document creation. Only these practiced users are able to exactly adhere to the standards in formulation and formating and to exactly specify individual preferences in content and format whenever it is needed.

By using semantic technology, the speech recognition system is being expanded to move from producing only a word-by-word transcription of the dictation toward an automatic transformation, from the verbal to the written and formatted level. The language model which is used in the recognition process is adapted to incorporate the typical transformations between the dictation and the final written document. Adding this intelligent natural language understanding component to the speech recognition system lays a foundation for successfully using ASR technology in document creation and data-mining tasks.

Duration: 2006 - 2010
Sponsor: Austrian Ministry of Economics and Labour (K-Net Programme)
Researchers: Alexandra Klein, Johannes Matiasek, Martin Huber, Jeremy Jancsary, Hannes Pirker, Stephanie Schreitter, Katalin Lejtovicz
Partners: Philips Speech Recognition Systems, Fachhochschule Technikum Wien