Immersive audio systems can be difficult to configure when a performance or installation uses many speakers, audio interfaces and playback devices. Complex signal paths and centralized equipment can add setup and control work, limiting an engineer’s ability to focus on the sound. A system combining Ethernet Audio Video Bridging (AVB) with the ImmerGo spatial audio framework offers a different approach: transmit synchronized audio over a network, distribute rendering across endpoint processors and manage sound objects through a browser-based interface.
The approach was demonstrated in a choreographed installation spanning two rooms. Its technical ideas are relevant to anyone assessing networked media systems: define how devices communicate, separate sound positioning from speaker-specific mixing, and make the configuration manageable as the playback environment changes.
How Ethernet AVB supports networked audio
Ethernet AVB is an open IEEE-standard approach for transmitting time-sensitive audio and video over Ethernet. It uses networking mechanisms intended to support quality of service by accounting for transmission delays and buffering, alongside synchronization and device-connection management. AVB devices can be identified and connected through the network rather than relying only on a collection of manually configured point-to-point audio links.
Devices typically operate as talkers, which send audio streams, listeners, which receive them, or both. A device that can perform both roles may receive a stream and pass audio on to other devices. The 1722 transport protocol carries media, while the 1722.1 AVDECC protocol supports device discovery, connection management and control. Together, these functions allow compatible devices to be located and their stream connections managed over the network.
In the demonstrated setup, an AVB audio interface sent live microphone feeds to a computer, where the audio could be handled by a digital audio workstation (DAW). Audio from the DAW was then streamed to network-connected processors serving speakers in different parts of the installation. A connection matrix provided a way to establish the required stream paths.
Object-based audio separates position from output
Traditional channel-based mixing is built around a particular speaker layout: the mix contains channels assigned to that configuration. Object-based audio takes another route. Each track is treated as an audio object with associated metadata, such as its position, level and spread. A renderer uses that information to calculate how the sound should be distributed across the available speakers.
This distinction matters when the playback layout is irregular or may change. Rather than composing a separate fixed-channel mix for every speaker arrangement, the engineer positions audio objects and lets the renderer translate those positions into speaker levels. The audio’s spatial description is therefore separated from the final speaker feeds.
ImmerGo uses a client-server structure to manage this process alongside the DAW. The server receives object-position information from a browser-based client, calculates speaker mix levels and sends control data to the endpoint processors. It can also communicate with the DAW over MIDI, including transport commands and MIDI timecode. Timecode lets position changes be associated with the DAW timeline and updated in synchronization with playback.
Choosing a spatial rendering method
The rendering algorithm affects how an audio object is represented across a speaker array. The system described in the paper includes vector-base amplitude panning (VBAP) and distance-based amplitude panning (DBAP), each suited to different layouts.
VBAP uses a nearby group of speakers to position a virtual source. It works around a defined listening area, or “sweet spot,” so the intended source position is best perceived from a particular location. That constraint can be problematic in installations where visitors move through the space or where speakers cannot be arranged around one central listening point.
DBAP calculates speaker levels using the distances between the virtual source and the speakers. It does not depend on a single sweet spot, making it a better fit for the irregular speaker arrangements and moving audiences found in the installation described. Its normalization process also accounts for speakers being added to or removed from the array, supporting changes to the configuration without defining a fixed speaker geometry.
Distributing rendering and simplifying control
Instead of performing every stage of the rendering in the computer running the DAW, ImmerGo divides the work. The server calculates the mix levels for each audio object, while endpoint processors perform the mixing of audio samples and feed the speaker outputs. In the example, network DACs with processor-based mixing matrices received audio streams and control messages over AVB.
This arrangement lets processing be distributed among endpoints as the speaker system grows, rather than requiring all output mixing to happen in one central location. A speaker-configuration file records processor identifiers, speaker counts and physical positions. The server uses this information to manage connections and calculate the levels needed for the configured layout.
For the installation, a mobile device connected to the server through a web browser. The engineer could select tracks, adjust their position and level, record movement, and control DAW transport while moving around the space. This was useful in a performance distributed across two rooms, where some sources remained in fixed zones while others were moved in relation to the audience.
What the installation demonstrates
The sound design combined field recordings, prerecorded material and live microphone feeds. The installation included a tunnel made from plastic milk bottles and packaging, with performers stationed in three zones. Live audience sounds were also recorded and played back through the speakers. Within the DAW, audio tracks could be processed and treated as separate objects, allowing their positions to be controlled independently.
The practical lesson is architectural rather than tied to a particular creative project. Networked audio can bring multiple compatible devices into a shared signal and control framework; object-based mixing allows sound positions to be rendered for the speaker arrangement; and distributed endpoint processing can reduce dependence on a single centralized mixing stage. Mobile browser control gives the operator access to spatial and transport controls from within the performance space.
Conclusion
Ethernet AVB and ImmerGo illustrate how synchronized network streams, device discovery, object-based mixing and distributed processing can work together in an immersive audio installation. The approach does not remove the need to plan speaker positions, configure devices or choose an appropriate rendering method. It provides a structured way to handle those tasks, particularly when audio must move across an irregular, multi-room environment and be controlled during a live performance.