Computer Audition The Era of Large Models
Speaker: Björn Schuller – Munich, GermanyTopic(s): Artificial Intelligence, Machine Learning, Computer Vision, Natural language processing
Abstract
In this lecture, we delve into the fascinating field of Computer Audition Ð that is, hearing, understanding, and generating audio by computers powered by the latest advancements in Artificial Intelligence. By adopting the current trend of Foundation or Large Models, we explore how this recent brute force methodology utilizes extensive data to transform the realm of artificial audio intelligence. Combining advanced deep learning techniques with inquisitive self-learning, the large model approach surpasses traditional learning paradigms, ushering in a new era of machine hearing. In this era, audio can be meticulously analyzed dissected into its components, with rich attributions to their states and characteristics. Techniques like Zero-shot and Few-shot learning leverage the emergent capabilities of large audio models. Additionally, we present a novel approach to using language as a medium for AI to comprehend and generate audio. Join, as we dissect the complex architecture of today's Computer Audition, integrating deep prompting, in-context learning, model merging, and adaptive in-situ methods, revealing the potential of contemporary advancements in deep and large-scale modeling. Applications span across speech, music, and sound and emphasize real-world problems as diverse as health, fault-detection, or wild-life monitoring. Nature language processing will play a crucial role, as language will be used to describe audio, and speech analysis and generation will include language processing. Further, potential for multimodal exploitation across video and further data is touched upon.About this Lecture
Number of Slides: 60Duration: 60 minutes
Languages Available: English, German
Last Updated:
Request this Lecture
To request this particular lecture, please complete this online form.
Request a Tour
To request a tour with this speaker, please complete this online form.
All requests will be sent to ACM headquarters for review.