SEGMENTS:2024 Sponsors:

Speaker Profile: Vignesh V Menon

Postdoctoral Researcher



Online Bitrate ladder prediction for Adaptive VVC Streaming

9:35 AM to 9:50 AM

Traditional per-title encoding methods yield lower storage and delivery costs and improved Quality of Experience (QoE), with an additional cost of convex-hull computation. However, the streaming industry can reduce its carbon footprint and energy consumption by minimizing the processing time for per-title bitrate ladder estimation, encoding, and decoding. This is particularly important as environmental consciousness grows and companies seek to adopt greener practices. This talk highlights energy-aware, perceptually-aware per-title encoding approaches adapted for adaptive versatile video coding (VVC) streaming applications. Content- and JND-aware bitrate ladders are estimated using low-complexity features based on Discrete Cosine Transform (DCT) energy, set of encoding parameters (e.g., resolutions, framerates, presets, etc.) supported by the streaming service provider. This talk further introduces the open-source implementations of the proposed methods, and discusses the objective results, in terms of improved compression and energy efficiency.