Speech recognition

Speech recognition appeals to a wide range of skills for enabling a computer to interpret the human voice and to correctly react to it. The problems to be solved range from the most basic ones, such as detecting human voice activity, to more challenging issues such as producing live closed-captioning of television broadcasts or determining the topic of a telephone conversation.

Related technologies: Keyword detection, audio content indexing, closed-captioning, text/audio alignment, recognition engine, pronunciation verification, finite-state transducers

CRIM has been taking part in competitions and technological evaluations in speech and speaker recognition since 1992. Our results in international evaluation campaigns such as ETAPE 2011 have established our position among the best research teams of our discipline. We actively pursue our research while keeping abreast with the latest developments of our scientific community, a keen example being our participation in the KALDI group that brings together  leading researchers of our field.

 

 

Teams

Releases

Recent news

  • Le CRIM en chiffres
    20/07/2018

    Indicateurs clés de l'année 2017-18.

    +

Upcoming event

  • Montréal A.I. Symposium 2018
    28 August 2018 8:00
    Pavillon Wirth de l'Université McGill
    CRIM will be part of the Montreal A.I. Symposium 2018 to be held on August 28th, 2018 at McGill University.
    +

Recent Publications

  • Towards Automatic Feature Extraction for Activity Recognition from Wearable Sensors: A Deep Learning Approach

    +
  • État des lieux des technologies web

    +