Speech synthesiser translates mouth movements into robot speech

Vocoders just got a serious upgrade. A new speech synthesiser can translate mouth movements directly into intelligible speech, completely bypassing a person's voicebox. Although the synthesiser might not be immediately useful, it's a first step towards building a brain-computer interface that could allow paralysed people to talk by monitoring their thought patterns. To create the speech synthesiser, scientists at INSERM and CNRS in Grenoble, France used nine sensors to capture the movements of the lips, tongue, jaw and soft palate. A neural network learned to translate the sensor data into vowels and consonants, which are emitted from a vocoder.