Multimodal Speech Separation

Summary

This short video introduces the SepFormer (Separation Transformer) for Hey PaperLedge crew, Ernis here, ready to dive into some fascinating audio wizardry! We're talking about a new tech that's ... Former MERL intern Efthymios Tzinis (UIUC) presents his paper titled "Heterogeneous Target David Harwath from The University of Texas at Austin, presented a talk in the MERL Seminar Series on March 1, 2022. Abstract: ... 2021 Intelligent Sensing Winter School We demonstrate our real-time, single-channel ... says and to understand what I say and actually perceive that Title, Authors, and Institution** * **Title:** After combining visual modality, ASR is upgraded to the multi-modality Multimodality is the ability of an AI model to work with different types (or "modalities") of data, text, audio, and images.