A dependency-free (pure standard-library) Python tool that connects an objective PICTURE metric (VMAF) to a session QoE score. Given a session described segment by segment (the played rendition mapped to its measured VMAF) plus the rebuffering and switching that happened, it computes: (1) the DELIVERED-vs-ENCODED gap — top-rung VMAF vs time-weighted PLAYED VMAF; (2) the ABR QoE OBJECTIVE (a real published objective, MPC/Pensieve) QoE = Σq − μ·Σstall − Σ|Δq| in linear and log variants, mapping VM
Download free PDF