vllm_omni.model_executor.models.audex ¶

Modules:

Name	Description
`audex_code2wav`	Audex code2wav: streaming speech-codec-token → waveform stage.
`audex_omni`	Audex audio-understanding model (`checkpoint_folder_full`): S2T / audio QA.
`audex_omni_processing`	Multimodal processing for Audex audio understanding (S2T / audio QA).
`audex_thinker`	Audex thinker: the Nemotron dense decoder-only LM that autoregressively
`audex_xcodec`	Audex TTA stage 1: XCodec1 RVQ codes → waveform.
`cfg`	Classifier-free guidance for the Audex thinker stage.
`checkpoint`	Audex checkpoint layout preparation.
`pipeline`	Audex (Nemotron-Labs-Audex-2B) TTS pipeline topology.
`prompt`	Audex TTS/TTA prompt construction.
`speech_decoder`	Vendored Audex causal speech decoder (Apache-2.0).
`tta`	Audex text-to-audio (TTA) token space: RVQ phase masking and helpers.