Noniterative Content-Adaptive Distributed Encoding Through ML Techniques

Abstract

Distributed encoding is desirable for content preparation cloud workflows to reduce turnaround times. Content-adaptive bit allocation strategies have been proposed to achieve efficiencies in storage and delivery. Many of these methods tend to be iterative in nature and consume significant additional compute resources. There is a need to limit this increase in computational complexity. In this paper, we propose a noniterative codec-agnostic approach that employs machine learning techniques to achieve average bitrate savings and a target consistent quality by selecting a content-adaptive bitrate and resolution for each adaptive bitrate (ABR) segment within each ABR representation in a manner that makes it equally suitable for live and on-demand workflows. Test results are presented over a wide range of content types. Initial results indicate that the proposed approach can recover ~85% of the bit savings possible with more exhaustive techniques while its computational complexity is only 15%–20% of two-pass variable bitrate (VBR) encoding.