Featured Projects

Imagine being in a crowded room where several people are speaking at the same time. Yet somehow, your brain can focus on a single voice you care about and filter…

Motivation Ultra-low-bitrate neural speech codecs have made remarkable progress in recent years. Modern systems can operate at a few hundred bits per second while maintaining surprisingly good perceptual quality. However,…

Overview In everyday life, our devices run many speech/audio applications that can benefit from the target speaker extraction (TSE) concept: the ability to pull out your voice from a noisy mixture of sounds.…

Machine-Learned Latent Features Are Codes for That Machine! When we think about compressing sound, we usually imagine MP3s or AACs, or even neural codecs with a higher compression ratio these…

[latexpage] As we've proposed in the BLOOM-Net project, scalability matters. Just to reiterate the argument here once again, the main issue with the current deep learning-based models for speech enhancement…

[latexpage] Motivation We bring the cool generative power of a diffusion model to speech coding. We call our codec LaDiffCodec as it is actively using the concept of latent diffusion((R.…

TLDR: In this project, we developed an end-to-end neural network system that takes a music mixture as input and produces its remixed version per the user's intended volume changes of…

The spatial image of a music source is an essential feature in the stereophonic music listening experience. An extreme case would be "Yellow Submarine" by The Beatles, where all the…

[latexpage] Scalability is a big deal when it comes to video coding. When you watch a movie via a streaming service on Friday night, the video quality fluctuates—it's the video…

(Download Interspeech 2022 Tutorial Slides) The outstanding development in modern AI has relied greatly on the improved modeling capacity. The deep learning models, for example, are effective in scaling up…