demystify-voice-lingua 0.3.0

Low-latency multilingual extension for the Demystify voice pipeline: in-stream LID, hysteresis-damped mid-call language switching, per-language STT/TTS routing

demystify-voice-lingua

Low-latency multilingual extension for the Demystify voice pipeline (demystify-voice): in-stream language identification (LID), hysteresis-damped mid-call language switching, reply-language policy, and per-language STT/TTS routing over persistent connection pools.

This is the packaged plugin of the dmstfy-voice-lingua module. The full module docs (profiles, benchmark harness, runbooks, INTEGRATION.md) live in the repository at modules/dmstfy-voice-lingua/.

Install

pip install demystify-voice-lingua                 # standalone (uses the vendored protocol mirror)
pip install "demystify-voice-lingua[voicecore]"    # + the real demystify-voice runtime (>=0.1.0,<0.2)

The plugin works fully standalone (offline simulator + vendored protocol mirror) and, when demystify-voice is installed via the voicecore extra, wires into the real voice-core pipeline via install(pipeline, agent_config, ...).

Protocol-version handshake

install() performs a protocol-version handshake against the installed demystify-voice (PROTOCOL_VERSION) and raises a clear error on a MAJOR mismatch (a breaking frame/stage/pipeline change). Additive MINOR bumps are compatible. The voicecore extra is pinned to the compatible MAJOR (>=0.1.0,<0.2). See INTEGRATION.md.

MIT licensed.