Benjamin O'Brien

- Nov 2023 - Present : ANR-project in collaboration with Orange Telecom and IRCAM, with the purpose of explaining vectors generated by speech recognition models (deep learning). How ? Identifying and correlating dimensions with acoustic features associated with the voice and setting them as parameters for voice synthesis. - Apr 2022 - Nov 2023 : Worked with startup Semaxone (aerospace sector) on analyzing and applying machine learning to acoustics and biometric signals (ECG, fNIRS) for the purposes of modeling cognitive and behavioral changes (stress, hypoxia) in pilots.

- Project VoicePersonae, with Eurecom / Sorbonne and NII National Institute of Informatics (Japan) : Perceptual evaluation of anonymized speech for the purposes of developing cyber-security tools and evaluation methods. - ANR-Project VOXCRIM, in collaboration with Service National de Police Scientifique : Evaluate the effects of continuous modifications of speech characteristics and task design on automatic (deep learning models) and perceptual performance for the purposes of forensics speech.

Project PEPS (Pédalage Par La Sonification) : Develop augmented-reality protocols to provide real-time auditory feedback based on the pedal stroke and study the effects on expert cyclist performance (efficiency, kinematic movement, muscular activations).

ANR-Project in collaboration with Stellantis (PSA) : Develop augmented-reality protocols for providing real-time auditory feedback based on golf putting and swing strokes, and study the effects of sound on novice and expert golfers.