Ming-derived model implementation
=================================

This directory contains code adapted from inclusionAI/Ming:
https://github.com/inclusionAI/Ming
Revision: e58533db227031990c5a6864dcf5f08fb53ed0d2
Copyright (c) 2025 inclusionAI

The upstream MIT license is reproduced in LICENSE.Ming. Retain that license
and the original copyright notices when redistributing the adapted code.
The repository's Apache-2.0 license applies to vLLM-Omni contributions; it
does not replace the notices applicable to incorporated third-party code.

Source references for the adapted components:
- native_text_segment_cut.py: front/text_segment_cut.py
- text_processing.py: front/ text segmentation and normalization utilities
- talker_module.py: talker_module/{dit,modules,cfm}.py
- ming_flash_omni_talker.py and voice_presets.py: modeling_bailing_talker.py

Additional upstream source and copyright notices are retained in the relevant
files, including the Meta notices in talker_module.py and the Apache-2.0
notice in talker_tn/token_parser.py. Model weights and reference audio are
separate artifacts and retain their own licensing terms.
