An MCU (Multipoint Control Unit) is the older conferencing architecture in which a central server decodes every participant's audio and video, mixes them into a single combined stream, and re-encodes that for each recipient. Its advantage is the client side: each device sends and receives just one stream, so client bandwidth and CPU are minimal — valuable for weak devices, telephony bridges, and recording. The cost falls on the server, which must run a full decode-mix-encode pipeline per call, and on latency and quality, since transcoding adds delay and a generation of loss. For audio specifically, server-side mixing is still common (it is how a recording or a phone dial-in gets one clean blended track), even when video uses an SFU.
The book · Volume 5 of 7
Audio for Video: A Complete Guide to Sound in Video Products: From Jitter Buffers and Lip Sync to Atmos, Quality Metrics, and Neural Codecs
An MCU mixes every participant on the server, which saves client bandwidth but adds latency and server cost. Volume 5 covers the MCU in its chapter on bitrate and topology decisions: MCU against SFU and P2P for audio, mixing costs, recording and transcription, and how each topology fails. Written by Nikolay Sapunov, CEO at Fora Soft.
Weighing an MCU against an SFU for audio in your conferencing product?
Tell us your call sizes and devices and we will recommend MCU, SFU or a hybrid, with server cost and latency estimates. 250+ video products shipped since 2005.
