HiveMind Voice Relay Documentation

July 31, 2026 ยท View on GitHub

Local wakeword detection. STT and TTS run remotely on hivemind-core with the hivemind-audio-binary-protocol plugin.


What is HiveMind Voice Relay?

HiveMind Voice Relay is a satellite client for the HiveMind mesh. It runs the microphone, voice-activity detection (VAD), and wakeword engine on-device, but forwards audio to hivemind-core running the hivemind-audio-binary-protocol plugin for speech-to-text (STT). The server synthesises speech (TTS) and sends audio back for local playback.

The point is architectural, not just resource savings: the hive owns STT/TTS. They run inside hivemind-core (via the hivemind-audio-binary-protocol plugin), behind the same access-key authentication as the rest of the mesh. A relay does not, and cannot, choose its own STT/TTS engine, model, or voice. The hive operator decides, centrally, for every relay. (Contrast a voice-sat, which can point at any plugin, including a public ovos-stt-plugin-server.) Relay also demonstrates the base64 speech API (recognizer_loop:b64_transcribe / speak:b64_audio), the same job mic-satellite does over the binary protocol.


Satellite spectrum

SatelliteMicVADWake wordSTTTTSConnects to
HiveMind-clin/an/an/an/an/ahivemind-core
hivemind-mic-satellitelocallocalserverserverservercore + audio-binary-protocol
HiveMind-voice-relaylocallocallocalserverservercore + audio-binary-protocol
HiveMind-voice-satlocallocallocallocallocalhivemind-core

Choose voice-relay when:

  • You want HiveMind to own and govern STT/TTS as an authenticated service, uniform engine/model/voice across all relays, decided by the hive operator, not the device.
  • You want wakeword detection on-device (latency, privacy, audio is not streamed until activation).
  • You do not want to run STT or TTS models on the device (a welcome consequence, not the main reason).
  • Your hivemind-core server has the hivemind-audio-binary-protocol plugin installed.

PageAudience
Getting startedFirst-time users
ConfigurationPlugin setup, all CLI flags
ArchitectureHow the relay pipeline works internally
Deploymentsystemd, Raspberry Pi, autostart
Development & testsInstall from source, run + understand the E2E suite
TroubleshootingCommon problems and fixes