Séminaire R-D: Learning Speaker Embeddings with Deep Networks

Séminaire R-D: Learning Speaker Embeddings with Deep Networks
15/11/16 11h00
CRIM (405, avenue Ogilvy, bureau 101, Montréal)

Learning Speaker Embeddings with Deep Networks

Présentation en anglais.

Conférencier

Gautam Bhattcharya, Ph. D., étudiant à l'université McGill et au sein de l'équipe Parole et Texte du CRIM.

Résumé

Speaker verification is a binary classification problem. Given two recordings, the goal is to determine if the two recordings came from the same speaker or not. Most successful verification systems break the task into two parts. The first part involves producing a representation of the speaker that is easy to work with, i.e. a low-dimensional representation. The second part involves making the actual verification decision. 

This talk will focus on the first part of the process - speaker representation. Specifically we show how neural networks can be used for speaker representation learning. We also present results that highlight the robustness of this type of representation, when there is a gross miss-match between the training and test data distributions.

Biographie

Gautam Bhattcharya is currently a Ph.D. student at McGill University. He is jointly advised by Prof. James Clark,and Prof. Patrick Kenny, Lead Researcher at CRIM. His research at CRIM is focused on the use of neural networks for speaker verification. He is broadly interested in machine learning, natural language understanding and speech processing.

 


Les séminaires scientifiques du CRIM, gratuits et ouverts à tous, sont donnés par des experts de renommée internationale, des collaborateurs universitaires, le personnel de R-D et les boursiers de 2e et 3e cycles du CRIM. Au programme, des présentations conviviales sur les dernières avancées scientifiques et technologiques.

Eventbrite - Journée Techno du CRIM : 30 ans d'innovation collaborative en TI

Conférence gratuite. Inscription requise.
Mardi, le 15 novembre 2016de 11 h à 12 h. Salle 11. Au CRIM, 405, avenue Ogilvy, bureau 101, Montréal. 

  • #Demain ???? Détection de signes de maladie mentale et de troubles du comportement à partir de la communication en lig… https://t.co/y8ZEQDBYsc
  • CNRC RT @CNRC_NRC: Le Gala des Prix Innovation 2019 de l’@ADRIQ_RCTi aura lieu demain! Des scientifiques d’#AutomobileCNRC et #TransportsCNRC so…