Layered Audio Delivery for Efficient and Resilient Multi-Station Internet Radio: A Case Study of RadioTEDU
Author: Arda Akgül
Broadcast Coordinator of RadioTEDU
TED University, Ankara, Türkiye
Date: 29 August 2026
Abstract & Summary
This paper presents a design-science case study of RadioTEDU, a multi-channel university internet radio platform operating 15 local public audio outputs across 7 distinct editorial station services (Classical, Lo-Fi, Pop, Jazz, Rock, Energize, and Situation Room). Rather than maintaining separate scheduling clocks or uncoordinated encoding pipelines, the system decouples its single authoritative editorial timeline and shared signed 16-bit 48 kHz stereo PCM processing from downstream codec-specific workers.
Key Architectural Highlights
- Multi-Tier Representation Matrix: Every station offers a 192 kbit/s AAC-LC primary stream for general-purpose high-fidelity listening; six music stations provide a 64 kbit/s HE-AAC v2 bandwidth-efficient stream for mobile networks; Classical and Jazz stations offer lossless Ogg-encapsulated FLAC streams for pristine studio reproduction.
- Shared PCM Production Pipeline: Editorial scheduling, 3.0-second quarter-sine music crossfades, and loudness processing take place once in uncompressed PCM before encoding, preventing generational transcoding losses across output branches.
- Normative Loudness & Headroom Policy: The production chain implements an EBU R 128 / ITU-R BS.1770-5 loudness normalization target (-23 LUFS integrated) and a -1 dBFS sample-limiter ceiling to maintain balanced listening consistency across genres and listening environments.
- Failure Isolation & Operational Resilience: Output workers operate under supervised bounded queues with staged exponential reconnection delays (1, 2, 4, 8, 15, 30 s). When an encoder or mount connection restarts, it automatically joins the live running programme PCM without interrupting editorial continuity or resetting the station clock.
