SMarT is one of Loria’s teams, it was created on December 2013
The acronym SMART comes in two forms:
Speech Modelisation and Text
Statistical Machine Translation
The objective of SMarT is to model the written or the spoken language (nature or altered) by using Machine Learning techniques.
The main objective of the SMarT team is to develop language representation models for machine translation and speech recognition systems. This modeling involves the use of mathematical methods to identify, extract and propose associations between two or more languages for translation and speech recognition. Languages are studied through monolingual, parallel or comparable corpora for low resourced languages(Arab dialects), Arabic, French and English.
Another axis of research in SMART concerns our contributions to the continuous speech recognition of the pathological voice. To do so, we develop models for enhancing esophageal speech using voice conversion techniques. These techniques of conversion are considered as a kind of translating one signal to another.
- Machine Translation
- Speech recognition
- Recognition of pathological voices
- Language identification
Keywords: Language Modelling, Machine Translation, unresourced languages, Estimation quality, Mining comparable corpora, Code-switching, Deep Learning,
voice conversion, enhancement of esophageal speech, pathological speech recognition.
- Delphine HUBERT (Université de Lorraine)
- Anne-Marie MESSAOUDI (CNRS)
- Chiraz Latiri – Professor (University of Tunis – Tunisia)
- Karima Meftouh – Assistant Professor (University of Badji Mokhtar, Annaba – Algeria)
- Salima Harrat – Assistant Professor (ENS – Algiers, Algeria)
- 2016-2020: Amine Menacer.
- 2019-2022: Fadi ghawanmeh co-supervision with university de Oslo.
- 2019-2023: Youness Moukafih with International University of Rabat
- 2019-2023: Maysoon Alkheir (University of Khartoum)
- 2019-2023: Donia Mohammed (University of Khartoum)