π I'm a Ph.D. student specializing in Natural Language Processing (NLP) and Speech Processing. My research focuses on the evaluation of Automatic Speech Recognition (ASR) systems, the analysis of their errors and their particularities! I work between Nantes and Avignon Universities π at the LS2N and the LIA. My supervisors are Richard Dufour, Jane Wottawa and Mickael Rouvier. Previously, I studied Cross-Lingual Embeddings and their evaluation methods at Aix-Marseille University with Carlos Ramisch and Manon Scholivet.
- π‘ One of my objectives is to propose a metric that correspond to human expectations but is as interpretable as Word Error Rate.
- π I developed a data set using 7,150 human choices which were annotated by 143 great people! This β¨open-accessβ¨ data set can be used to evaluate rigorously ASR metrics so we can finally get rid of Word Error Rate. You are more than welcome to develop your own metric and to evaluate it on the HATS dataset using our toolkit π€
- π One of the major implications of my work is the analysis of the impact of ASR hyperparameters on performances using diverse metrics. My research tends to show that metrics are not necessarily correlated at the transcription and system level.
- π Mathematics
- πΈ Guitar and Sing
- π² Game Theory
- π± Ethics theory
Feel free to connect with me on my LinkedIn or Twitter/X π