How QoS Shapes Video QoE: Key Findings from the New MPEI Whitepaper

In audio‑video scenarios, the exact impact of QoS (service quality) changes on QoE (experience quality) lacks a consensus in the industry. Practitioners often face three major questions: how to measure user experience gaps, how to define the relationship between experience and business growth with a metric system, and how far experience metrics can be optimized to balance cost and quality.

On December 14, Volcano Engine partnered with AMD to publish an Audio‑Video Experience Whitepaper , leveraging TikTok’s billions of daily active users and large‑scale deployment experience. The paper details evaluation metrics and models, shares Volcano Engine’s testing solutions, and presents typical strategies and cases for optimizing video experience to drive business growth.

Video functionality now permeates countless applications, making "video experience" a primary factor for user choice. Demand for higher resolution, stronger interaction, and immersive experiences drives longer playback, higher retention, and innovation, yet a comprehensive evaluation system remains absent, leading many companies to invest heavily with limited results.

Since 2021, Volcano Engine has conducted over 400 A/B tests on TikTok, quantifying business gains from each experience metric improvement. This effort produced a transparent, accurate, and comprehensive QoS metric system, centered on the Multimedia Playback Experience Index (MPEI) , enabling enterprises to pinpoint experience issues, optimize video quality, and boost growth.

The core audio‑visual quality and smoothness metrics cover the entire production‑consumption chain, from upload and storage to processing, distribution, and playback. Building on these, Volcano Engine further defines more than ten secondary metrics for detailed sub‑module modeling, capturing dynamic experience feedback.

Volcano Engine’s video lab lead Wang Fei explains that the metric system incorporates S‑type, J‑type, and D‑type model features to depict different user experience dynamics. After feature extraction, an "indicator fusion" process combines technical metrics, and statistical learning methods enhance overall prediction, ensuring each technical indicator accurately reflects experience improvements.

Using the MPEI model, customers can analyze various business scenarios, derive typical metric values, and formulate stage‑appropriate optimization strategies. Volcano Engine’s video cloud also conducts model analyses across different domains, selecting representative indicators to fit impact curves and typical values.

The released Audio‑Video Experience Whitepaper provides a detailed interpretation of these metrics and models, inviting readers to access the full document for deeper insights.