Team flowers

Overall Objectives
Scientific Foundations
Application Domains
New Results
Contracts and Grants with Industry

Section: New Results

A bag-of-features framework for incremental learning of speech invariants in unsegmented audio

Participants : Olivier Mangin, David Filliat, Pierre-Yves Oudeyer.

We have introduced a computational framework that allows a machine to bootstrap flexible autonomous learning of speech recognition skills. Technically, this framework enables a robot to incrementally learn to recognize speech invariants from unsegmented audio streams and with no prior knowledge of phonetics. To achieve this, we imported the bag-of-words/bag-of-features approach from recent research in computer vision, and adapt it to incremental developmental speech processing. We evaluated positively an implementation of this framework on a complex speech database reused from the ACORNS FP7 european FET project. This work was described in [25] .


Logo Inria