Search results

We live in an era where more and more tasks, once thought to be impregnable bastions of human intelligence, have succumb to AI. Are we at the cusp where ASR systems have matched expert humans in conversational speech recognition? We try to answer this question with some experimental evidence on the Switchboard English conversational telephony corpus. On the human side, we describe some listening experiments that established a new human performance benchmark.

The workshop reviews the latest developments in two-dimensional and volumetric visual content coding, drawing on the collaborative exploration projects of the Joint Video Experts Team (JVET) and Moving Picture Experts Group (MPEG).

Talks:- On SMPTE-2094-50: A New Dynamic Metadata Standard for HDR, Chris Cameron, GoogleIntegrating high dynamic range (HDR) content into composited systems and dynamic viewing conditions has proven difficult because initial standards were focused on viewing a single content source in a theater-like environment. This talk discusses how SMPTE ST 2094-50 addresses these challenges by introducing metadata for an HDR reference white anchor and by characterizing HDR content and displays by their HDR headroom.

The proposed presentation will address a critical challenge in modern AI: adapting foundation models to individual user needs efficiently. As generative AI systems with large numbers of parameters become ubiquitous, personalization is essential for practical deployment across diverse applications. This talk will be highly relevant to ICASSP attendees, bridging signal processing, machine learning, and efficient algorithm design.

Triplane-based implicit representations offer a highly promising structured framework for modeling static and dynamic 3D scenes. To meet the rigorous bandwidth and latency demands of immersive real-time streaming with practical deployment considerations, pioneering methods leverage established video coding standards like HEVC and VVC to compress these high-dimensional implicit representations.

The digitization of information has profoundly transformed human lives. At the core of this process lies quantization—the rounding of time and amplitude by analog-to-digital converters, a concept taught widely in engineering education. The Shannon–Nyquist theorem (1949) proves that time quantization can be lossless, but amplitude quantization introduces quantization noise, exposing a fundamental trade-off between digital resolution and dynamic range. Can we recover a signal from quantization noise alone?