Seminar from Julien Hauret, PhD students at LMSSC (CNAM de Paris)


Date: 10/11/2022
Time: 11h00
Localization: IC2 auditorium, and online
Speakers: Julien Hauret

Extension of the bandwidth for voice signals captured by noise-proof microphones


Speech recording is traditionally performed using microphones located near the speaker’s lips. However, this sound capture is sensitive to ambient noise, which reduces the intelligibility of communications in high noise levels. The use of unconventional microphones such as in-ear transducers seems appropriate. Indeed, this type of equipment eliminates the ambient noise but this is done at the expense of the bandwidth due to the low-pass nature of human tissue. An improvement of the captured signal (regeneration of the broad band content) is then necessary to make it more intelligible.

To meet these requirements, we will present our method compatible with real-time inference combining deep learning and signal processing.