NeuroPsychologie Interventionnelle


29 rue d'Ulm
75005 Paris France


-                Pr. Anne-Catherine Bachoud-Lévi, MD PhD, INSERM/ENS (Paris),,

-                Dr. Emmanuel Dupoux, PhD ; INRIA/ENS (Paris),


-                Renaud Massart:

-                Rachid Riad:


Hosting laboratory 

INSERM U955, E01 'Neuropsychologie interventionnelle’, NPI


Our team is dedicated to the understanding of cognitive functions, and more specifically of language processing, as well as of related deficits in neurodegenerative diseases. It is composed of a mix of medical doctors, cognitive scientists, speech therapists and statisticians. The acquired knowledge is used to develop tools to improve medical care and to better assess the efficacy of treatments in clinical trials. It is connected to the Neurological service of the Henri Mondor Hospital (Créteil) and to the Cognitive Science Department of the ENS (Paris), and is a national and international leader in research on Huntington and Parkinson disease. 

This project is a collaboration with the Cognitive Machine Learning (CoML) team, hosted by INRIA and ENS, with a mix of expertise in cognitive science, machine learning and speech and language technology. 

The postdoc will be hosted at the ENS, in the middle of the quartier latin and Paris and will have access to the resources of both teams (patient data, computing resources, expertise and help from both teams). Candidates will also have the possibility to shadow clinical practitioners at the hospital.



Purpose of the project

The overall goal of this project is to explore how neurological deficits can  be predicted using machine learning applied to naturalistic/ecological speech data obtained from interviews with the patient, or from everyday interactions as measured by wearables. 


This project is important for three perspectives. From a practical and clinical point of view, the longitudinal assessment of cohorts of patients using standard clinical batteries is costly and cumbersome. Being able to assess patients from simple telephone interviews or from wearable data would  allow a more fine grained longitudinal follow-up to speed-up clinical trials.  From a theoretical and cognitive point of view, batteries of test and lab based experimental paradigms reveal patterns of impairment that may not necessarily correlate with how patients function in real life. Moving towards more implicit testing allows studying how cognitive functions are deployed in ecological settings. From a machine learning point of view, the extraction of useful information from realistic speech signals is still a challenge that fosters advances in robust, adaptable and inclusive AI.


In practice, the work of the postdoc/research engineer will center around two possible types of activities (or both), depending on the interest and skills of the candidate:

-                organization of a growing dataset of speech and clinical data, and analysis of this dataset for predicting current and future standard clinical scores (data science/statistics profile) 

-                development of novel machine learning tools to predict various aspects of language, emotion and cognition from raw audio recording and automatisation of such analyses (machine learning/computer science profile)



Field of research

Machine learning, cognition, neurosciences, language


Required skills

Machine learning, speech/language processing, statistical analyses


Required qualities

Capacity to work within multidisciplinary teams, interest in clinical applications. 


Female candidates are encouraged to apply.



Short Bibliography


1.             Titeux, H., Riad, R., Cao, X. N., Hamilakis, N., Madden, K., Cristia, A., Bachoud-Lévi AC & Dupoux, E. (2020). Seshat: A tool for managing and verifying annotation campaigns of audio data. arXiv preprint arXiv:2003.01472.

2.             Riad, R., Bachoud-Lévi, A. C., Rudzicz, F., & Dupoux, E. (2020, May). Identification of primary and collateral tracks in stuttered speech. In LREC 2020-12th Conference on Language Resources and Evaluation.

3.             Riad, R., Titeux, H., Lemoine, L., Montillot, J., Sliwinski, A., Bagnou, J. H., Cao,  X.N., Bachoud-Lévi, A. C. & Dupoux E. (2020). Comparison of Speaker Role Recognition and Speaker Enrollment Protocol for conversational Clinical Interviews. arXiv preprint arXiv:2010.16131.

4.             Riad, R., Titeux, H., Lemoine, L., Montillot, J., Bagnou, J. H., Cao, X. N., Dupoux E. & Bachoud-Lévi, A. C. (2020). Vocal markers from sustained phonation in Huntington's Disease. arXiv preprint arXiv:2006.05365.

5.             Giavazzi M, Sambin S, de Diego-Balaguer R, Le Stanc L, Bachoud-Lévi AC, Jacquemot C. Structural priming in sentence comprehension: A single prime is enough. PLoS One. 2018 Apr 2;13(4):e0194959. doi: 10.1371/journal.pone.0194959.

6.             Jacquemot C, Lalanne C, Sliwinski A, Piccinini P, Dupoux E, Bachoud-Lévi AC. Improving language evaluation in neurological disorders: The French Core Assessment of Language Processing (CALAP). Psychol Assess. 2019 Jan 10. doi: 10.1037/pas0000683. [Epub ahead of print]

7.             Jacquemot, Charlotte, and Anne-Catherine Bachoud-Lévi. "Striatum and language processing: Where do we stand?." Cognition (2021): 104785.