Speak, hear, talk: The long quest for technology that understands speech as well as a human
Speech technology is so close to the level of human performance,” he said. “I believe in the next three years we’ll reach parity.
Le problème c’est qu’en Français de nombreux chercheurs changent de domaine. On me disait lors d’un Meetup qu’il y avait un effet balancier entre DeepLearning et les méthodes Bayésiennes classiques.
The act of understanding what someone is saying is very different from comprehending the subtle nuances of a person’s speech.
Et vraiment très dépendant de la culture, de la langue…
Source : Speak, hear, talk: The long quest for technology that understands speech as well as a human
via @Nicolas Popy