EtherSound

showstack / protocols / ethersound

Early low-latency audio-over-Ethernet system from Digigram, carrying 64 channels in a daisy-chained ring at fixed very low latency. Largely displaced by Dante and AVB but still found in installs from the 2000s.

Audio transportProprietary, documentedReportedDeprecated

Media Large, continuous, on time and in order

Continuous streams that are large enough to be sized properly, and that must arrive in order and on time.

What kills it. Anything that starves the link — usually a management transfer nobody scheduled — and jitter that outruns the receive buffer.

The four flows, and why the model is worth having →

What it carries

Sample rates: 48 kHz, 96 kHz
Audio bit depth: 24-bit
Audio channels: up to 64
Audio encoding: Linear PCM

Audio: 64 channels at 48 kHz on 100 Mbit for ES-100, which is what nearly every installed system in the field is. The later ES-Giga carries more on gigabit, but the two are not the same wire.

What goes wrong

EtherSound is layer 2 and topology-sensitive: it expects a daisy chain or ring, not an arbitrary switched network. Patching it through a general show switch does not work.

Encountering it now usually means an installed system nearing end of life. Plan the replacement path before a card fails, because spares are getting scarce.

Typical use

Legacy installed systems from the 2000s
Auvitran and Digigram based infrastructure
Fixed low-latency distribution in older venues

Sources

Digigram — Digigram primary

Understand the mechanism

Show networks Numbers that stand for signals

Something wrong, or missing?

Known gaps on this entry: default_ports, implementations, spec_url. If you can source one of them, that is a ten minute pull request.

Edit this entry on GitHub — one file, editable in your browser, and your handle goes on it permanently.