Skip to content

vllm_omni.model_executor.models.personaplex.duplex.config ¶

Constants of the PersonaPlex full-duplex integration.

PersonaPlex (nvidia/personaplex-7b-v1) is a Moshi finetune: a pure-lockstep full-duplex speech-to-speech model running at the Mimi codec frame rate (12.5 Hz / 80 ms). The values are fixed by the pretrained checkpoint; they live here so the rest of the package never hard-codes them.

CHUNK_PERIOD_MS `module-attribute` ¶

CHUNK_PERIOD_MS = int(round(1000 / FRAME_RATE))

DEFAULT_PERSONA `module-attribute` ¶

DEFAULT_PERSONA = "You are a wise and friendly teacher. Answer questions or provide advice in a clear and engaging way."

DEFAULT_VOICE `module-attribute` ¶

DEFAULT_VOICE = 'NATF2.pt'

FRAME_RATE `module-attribute` ¶

FRAME_RATE = 12.5

FRAME_SIZE `module-attribute` ¶

FRAME_SIZE = int(SAMPLE_RATE / FRAME_RATE)

SAMPLE_RATE `module-attribute` ¶

SAMPLE_RATE = 24000