Prerequisites for an automatic speech recognition technology for the Romanian language used in the court of law
Journal Title: Romanian Journal of Human - Computer Interaction - Year 2015, Vol 8, Issue 1
Abstract
In response to an IT-based, strategic goal of the Romanian Ministry of Justice, the running JustASR project is responsible for developing the architecture and technical elements to be used in implementing a system which mediates the man-computer interaction via the automatic detection of the Romanian language. The project seeks to develop the core of a technology based on the automatic speech recognition of the discourse used in the Romanian court of law. The paper briefly presents the premises and some of the phonetic elements which are paramount to the project’s success. After introducing the Speech Recognition technology, the paper presents the advances of Romanian research in the field. The paper then focuses on the linguistic peculiarities of the Romanian language, the first and foremost element of the technology underlying the JustASR project. The most significant part of the paper deals with the applied aspect of the phonetic theory for the Romanian language. As far as the creation of the dictionary is concerned, the general criteria for constructing the dictionary are illustrated with a special focus on Phonetics and the word categories involved.
Authors and Affiliations
Melania Duma , Corina Giurgea , Mihaela Ordean , Paula Zălhan
Improving parsing using morpho-syntactic and semantic information
In this paper we present the efforts of creating a syntactic parser with a very good performance on Romanian sentences. Instead of creating a parser from scratch, we decided to test the freely available existing ones and...
Visual Analytics on Multidimensional Massive Data
The paper proposes visual analysis techniques as a solution to the scientific community concerned with developing performant algorithms for exploring and analyzing massive and multidimensional data. The actually high per...
Applications for automatic transliteration between English and Romanian
Transliteration has been previously introduced for translating proper names between languages, when the two languages are incompatible at the phonetic level and they use very different alphabets. We propose an data-drive...
Diagramatic description of distributed spatial data processing
Most of the times, modeling and simulating natural phenomena involves a large number of operations, based on different compatibility constraints and rules. The solution proposed in this paper optimizes the entire descrip...
A Multidimensional Model of the Usefulness of Facebook for University Students
The popularity of social networking websites among university students stimulated the interest for studying the potential of use for educational purposes. The objective of this study is to test and validate a multidimens...