SoundvexAn engine separate from automation.
Audio processing, control and distribution have distinct responsibilities.
Pilot architecture
AzuraCast manages the library and schedule. Liquidsoap builds the continuous stream, Soundvex processes audio, then the encoder and Icecast distribute it to listeners. File normalization and cue points precede final processing.
C++20 + Go + web interface
The native engine processes PCM blocks. A Go service handles control and telemetry outside the audio path. The browser displays settings and measurements; it does not need to remain open for the radio to run.
What gets installed
The Linux pilot uses a native executable, a persistent control service, configuration files and a Liquidsoap integration adapter. A macOS environment is available for configuration and evaluation, while installation is adapted to the target Linux environment. Setting a password during installation and changing it later are product requirements.
Other automation systems and FM
The PCM engine is not conceptually tied to AzuraCast, but each automation system requires an adapter and validation. Programme processing can also be relevant to FM; a certified FM/MPX chain, pre-emphasis and deviation control are outside the streaming chain described here.
Resources and listeners
Processing runs once per processed stream, not per listener. More listeners mainly increase distribution-server traffic and connections. Additional encodes or independently processed streams can increase CPU usage. The pilot monitor shows actual resources, not a universal benchmark.