Peter Dirix Email & Phone Number

Leuven, Flanders, Be

Leuven, Belgium

Managing a team in-house linguistic engineers, interns and external contractors, working on ASR for the Automotive industry.The teams' tasks are:- creating large pronunciation lexica for large-vocabulary continuous speech recognition (LVCSR) and morphological generation/analysis & unifying these resources with the speech generation ones- creating grammars for morphological analysis of phrases- creating tokenization/normalization and formatting grammars for LVCSR and unifying tokenization over ASR, NLU, NLG and TTS- creating generation grammars for training data for ASR and NLU language model training- researching pronunciation guessing for ASR(/TTS), in particular for foreign named entities- linguistic support for engineers- researching methods to improve recognition by changes in lexica and grammars- analysis of field dataWe currently support 43 languages: Arabic, Bengali, Bulgarian, Cantonese, Catalan, Croatian, Czech, Danish, Dutch, English, Estonian, Finnish, French, German, Greek, Hindi, Hebrew, Hungarian, Indonesian, Italian, Japanese, Kannada, Korean, Latvian, Lithuanian, Malay, Mandarin (Putonghua and Sichuanese), Marathi, Norwegian, Polish, Portuguese, Romanian, Russian, Shanghainese (Wu), Slovak, Spanish, Swedish, Tamil, Telugu, Thai, Turkish, Ukrainian, VietnameseParticipated in the organisation of the Cerence Technology Conference 2022 (Ulm).Member of the Industrial Advisory Committee of the Erasmus Mundus European Masters Program in Language and Communication Technologies (2020-2023) and the UPSKILLS project (2020-2023).Also gave seminars on speech recognition at the National Library of Norway and the Université catholique de Louvain.

Leuven

- METIS projects on hybrid machine translation (2002-2007)- ATRANOS project on summarization for subtitling (2004)- reviewing annotation of the Corpus Spoken Dutch (CGN) (2004)- writing project proposals- member of the local organizing committee of conferences (EMSSL 2002, HPSG 2004, Metis II workshop 2007, CLIN 17, CLIF 2008, EAMT 2011, CLIN 27)- editor of CLIN 17 conference proceedings- developed a multilingual ontology of job titles for Selor (the Selection Office of the Belgian federal government)- AfriTools project (2012-): creating tools and resources for linguistic research in Afrikaans, including a tokenizer, a lemmatizer, a POS tagger, a 250K manually verified lexicon and a UD version of the AfriBooms treebank- research on syntax in Afrikaans (2012-): verb clusters (IPP, preteritive assimilation/replacement, cluster creepers, quirky V2, word order), tag questions, insubordination, reduplication- independent expert for the European Commission's CORDIS programme- member of the scientific committee of CLIN 17, the Afrikaans colloquium in Ghent 2015, the 3rd Afrikaans Grammar Workshop (2023), the UniDive 2023 General Meeting, COLING (since 2018), EMNLP 2021, CLIN Journal 2021 and LREC (since 2007)- co-chair of the special section on Language Technologies of the International Conference on Information and Software Technologies (ICIST) (2016-)- member of European COST action working group "Universality, diversity and idiosyncrasy in language technology" (UniDive) (2022-)

Leuven

Managing a team of up to 14 in-house linguistic engineers, 2 interns and 35 external contractors.The teams' tasks are:- creating large pronunciation lexica for large-vocabulary continuous speech recognition (LVCSR) and morphological generation/analysis- creating grammars for morphological analysis of phrases- creating tokenization/normalization and formatting grammars for LVCSR- creating recognition grammars for finite-state language model speech recognition- creating generation grammars for training data for ASR and NLU language model training- linguistic support for engineers- researching methods to improve recognition by changes in lexica and grammars- analysis of field dataProduct lines we support are:- desktop dictation (Dragon Naturally Speaking, Dragon Professional, Dragon AnyWhere)- healthcare (Dragon Network Medical Edition, Dragon Medical Practice Edition, PowerScribe 360)- smart television (Dragon TV)- mobile (Dragon Dictation, Dragon Search, Dragon Mobile Assistant, Voicemail-to-Text, Nuance Mix, Nuance Transcription Engine)- cars (Dragon Drive)- enterprise (Nina)Support for 37 languages: Arabic, Cantonese, Catalan, Croatian, Czech, Danish, Dutch, English, Estonian, Finnish, French, German, Greek, Hindi, Hebrew, Hungarian, Indonesian, Italian, Japanese, Korean, Latvian, Lithuanian, Malay, Mandarin (Putonghua and Sichuanese), Norwegian, Polish, Portuguese, Romanian, Russian, Shanghainese (Wu), Slovak, Spanish, Swedish, Thai, Turkish, Ukrainian, Vietnamese- Member of the Industrial Advisory Committee of the SCATE project (2014-2018).

Diegem / Leuven

- building medical language models for Dutch and German- maintaining lexical framework, developing lexical and validation tools- phonetic transcriptions for English, Dutch, French, Spanish, German, Italian, and Russian- lexicon development, tokenizer & postprocessor grammars for English, Dutch, French, Spanish, German, Italian, and Russian- teaching seminar on speech recognition at Provinciale Hogeschool Limburg

Diegem

- building medical language models for Dutch, German, Spanish, and French- developing corpus cleaning & lexical tools- phonetic transcriptions for Dutch, German, and Spanish- managing linguistic contractors- teaching seminar on speech recognition at Provinciale Hogeschool Limburg

Wemmel

- developing tools for corpus cleaning and annotation- setting up a text corpus server- specialist on character encodings- giving internal training on basic computer skills, Linux, character encodings, corpus annotation, Perl

M.Sc., Artificial Intelligence, Cum Laude

Activities and Societies: Plutonica, KSC, Endivia, KVHV Also European Master in Speech and Language. Cum laude. Internship report.

Dipl., Applied Computer Science

Activities and Societies: Plutonica, Endivia, KVHV Specialisation: Computational Linguistics Thesis: Comparison of parser techniques.

Dipl., Computer Science

Activities and Societies: Plutonica, Endivia, KVHV Specialisation: Computational Linguistics

Lic. (M.Sc.), Physics

Activities and Societies: Plutonica, Endivia, KVHV Specialisation: astrophysics Thesis: Stationary state of linear magnetohydrodynamic.

Cand., Physics

Activities and Societies: Filii Lamberti

Latin, Greek, Maths