VoxCrim

Last updated on Mar 21, 2023
Photo by Juan Montana on Unsplash

This project is funded by the French National Research Agency (ANR). It is the first time some French public research labs have teamed up with the forensics labs of Police Nationale and Gendarmerie Nationale. Our aim is to establish a reliable and officially certified workflow in the field of forensic voice comparison. I’m involved in 2 key aspects of the project:

  1. the acoustic-phonetic modelling of between-speaker variation
  2. the perceptual and neural correlates of voice comparison

Participants:

The projet website is here

Research Forensic Voice Phonetics Deep Learning
Emmanuel Ferragne
Emmanuel Ferragne
Professor of Phonetics

Publications

Analyse phonétique de la variation inter-locuteurs au moyen de réseaux de neurones convolutifs : voyelles seules et séquences courtes de parole
Des réseaux de neurones convolutifs ont été entraînés sur des spectrogrammes de voyelles /ɑ̃/ et de séquences aléatoires de 2 secondes …
Project
Analyse phonétique de la variation inter-locuteurs au moyen de réseaux de neurones convolutifs : voyelles seules et séquences courtes de parole
The N400 reveals implicit accent-induced prejudice
Our perception of someone’s accent influences our expectations about what they might say or do. In this experiment, EEG data were …
Cite Project DOI
The N400 reveals implicit accent-induced prejudice
Automatic classification of phonation types in spontaneous speech: towards a new workflow for the characterization of speakers' voice quality
Voice quality is known to be an important factor for the characterization of a speaker’s voice, both in terms of physiological features …
PDF Cite Project DOI
Automatic classification of phonation types in spontaneous speech: towards a new workflow for the characterization of speakers' voice quality
Deep learning and voice comparison: phonetically-motivated vs. automatically-learned features
Broadband spectrograms of French vowels /ɑ̃/, /a/, /ɛ/, /e/, /i/, /ə/, and /ɔ/ extracted from radio broadcast corpora were used to …
PDF Cite Project
Deep learning and voice comparison: phonetically-motivated vs. automatically-learned features
Towards phonetic interpretability in deep learning applied to voice comparison
A deep convolutional neural network was trained to classify 45 speakers based on spectrograms of their productions of the French vowel …
PDF Cite Project
Towards phonetic interpretability in deep learning applied to voice comparison

Events

Quelques éléments sur l'expertise vocale en criminalistique
La voix humaine ne peut pas être assimilée à des indices véritablement biométriques comme les traces papillaires ou l’ADN. …
May 26, 2023 2:00 PM Université Lumière Lyon 2
Project Project
Quelques éléments sur l'expertise vocale en criminalistique
Analyse phonétique de la variation inter-locuteurs au moyen de réseaux de neurones convolutifs : voyelles seules et séquences courtes de parole
Des réseaux de neurones convolutifs ont été entraînés sur des spectrogrammes de voyelles /ɑ̃/ et de séquences aléatoires de 2 secondes …
Jun 16, 2022 4:00 PM Noirmoutier
Project
Analyse phonétique de la variation inter-locuteurs au moyen de réseaux de neurones convolutifs : voyelles seules et séquences courtes de parole
Automatic classification of phonation types in spontaneous speech: towards a new workflow for the characterization of speakers' voice quality
Voice quality is known to be an important factor for the characterization of a speaker’s voice, both in terms of physiological features …
Aug 30, 2021 10:00 AM Brno, Czechia
Project
Automatic classification of phonation types in spontaneous speech: towards a new workflow for the characterization of speakers' voice quality
De la comparaison de voix en criminalistique à la N400 sociophonétique
En criminalistique, la comparaison de voix consiste à déterminer si deux enregistrements vocaux émanent d’un même individu ou non. …
Dec 1, 2020 10:00 AM Paris (Zoom actually...)
Project Project Video
De la comparaison de voix en criminalistique à la N400 sociophonétique
The N400 as a sociophonetic marker
Oct 1, 2020 1:00 PM Nice
Project
The N400 as a sociophonetic marker
Deep learning and voice comparison : phonetically-motivated vs. automatically-learned features
Aug 5, 2019 10:00 AM Melbourne
Project
Deep learning and voice comparison : phonetically-motivated vs. automatically-learned features
Towards phonetic interpretability in deep learning applied to voice comparison
Aug 5, 2019 10:00 AM Melbourne
Project
Towards phonetic interpretability in deep learning applied to voice comparison