ASPARaGus Reading Group, Imperial College London, London, GB
Short presentation where I talked about EnCodec, a new neural codec that achieves state-of-the-art results on clean/noisy speech and music at 24/48 kHz (mono and stereo). It also has open-source code and outperforms Google’s approach - LyraV2
ASPARaGus Reading Group, Imperial College London, London, GB
Short presentation where I talked about neural vocoders, and how they are moving away from GANs and towards diffusion models. [slides]
Conference presentation, Interspeech 2022, Incheon, Korea
Oral presentation about our new scalable video-to-speech model (SVTS). [recording]. [slides] [paper]
OutSystems AI Reading Group, OutSystems, Lisbon, Portugal
Short talk where I presented our new scalable video-to-speech model (later published in Interspeech 2022). [recording] [slides]
Sight and Sound Workshop, CVPR 2022, New Orleans LA, USA
Short talk where I presented our new scalable video-to-speech model (later published in Interspeech 2022). [recording (presentation at 1:34:10)] [slides] [short paper]
Poster Presentation, Imperial College London, London, GB
I presented a summary of my PhD’s research contributions to the president of Portugal, Marcelo Rebelo de Sousa, as well as many other attendees at Imperial College London. [poster] [article]
ASPARaGus Reading Group, Imperial College London, London, GB
Short presentation where I talked about past, current and future trends in self-supervised learning, including BYOL. [slides]
Sight and Sound Workshop, CVPR 2021, Virtual Platform
Short talk where I presented our new end-to-end video-to-speech model (later published in IEEE Trans. on Cybernetics). [recording (presentation at 42:23, questions at 57:38)] [slides] [short paper]
Show & Tell Demo, ICASSP 2020, Barcelona, Spain
Short talk where I introduced our new video-to-speech model and showed its performance using an interactive demo. [recording] [slides]