This page displays a few demonstration examples of the research conducted at MULTISPEECH. Don’t forget to check the publication page as well, since many papers come with sounds examples / a companion webpage.

Speech recognition

Below is a  robust speech recognition demo from the FUI VoiceHome project. The goal of this project was to conceive and implement a natural language voice interface for smart home and multimedia (set-top-box) appliances.

Speech separation and enhancement

Below are examples of  speech separation and enhancement obtained using our Asteroid framework. Check out the Asteroid demo webpage for more examples!

Speech separation

Here, a mixture  is separated into two single-speaker signals.



Original speakers


Estimated speakers


Speech enhancement

Here, the background noise is removed from a noisy speech signal.

Noisy signal


Original speech signal


Estimated speech signal