Express Paper
Qualifying Timing Errors in Audio-over-Ethernet Networks for Live Sound
Express Paper · Paper 427 · May 2026
Abstract
Audio-over-Ethernet (AoE) protocols have become fundamental in modern live sound reinforcement systems, yet their real-world synchronization behavior under diverse stress conditions, both in terms of load and configuration, is not accurately characterized. Microsecond-scale timing mismatches between amplifier outputs can disrupt line-array interference patterns, reducing directivity control and spectral consistency. Ensuring robust timing accuracy across
large, mixed-traffic network topologies is therefore critical for predictable system performance. This paper presents a comprehensive, application-oriented evaluation of Dante, AES67 and Milan-AVB. A representative multi-hop architecture typical of touring deployments is considered. A controlled measurement campaign, combining eight daisy-chained switches, heavy concurrent data traffic approaching link saturation, and sub-sampled latency tracking, assesses each protocol under ideal conditions, typical field situations, and common misconfigurations. Results
reveal clear performance distinctions. Dante exhibits substantial timing variations, exceeding 100 ms under load. AES67 provides tighter synchronization but remains vulnerable to configuration errors, which can induce latency drift or even audio packet loss. Milan-AVB consistently maintains sub-microsecond accuracy across all scenarios.
