# Third-party notices

This bundle is an unofficial community derivative. It is not affiliated with
or endorsed by Resemble AI or NVIDIA.

## Chatterbox-Flash

The model weights and vendored runtime are derived from
[ResembleAI/chatterbox-flash](https://github.com/resemble-ai/chatterbox-flash),
source commit `74e05baa8ce574bf2cc571702391a21f1b0d48c5`.

Copyright (c) 2026 Resemble AI. Licensed under the MIT License. The complete
license text is reproduced in [LICENSE](LICENSE) and in
`runtime/chatterbox-flash/LICENSE`.

The derivative changes the FlashInfer engine/provider integration and adds a
canonical, source-free NVFP4 loader. The unchanged S3Gen, voice-encoder, and
tokenizer assets are not redistributed; the loader fetches them from the
immutable Chatterbox-Flash Hub revision recorded in `nvfp4_config.json`.

## Chatterbox-TTS

Chatterbox-Flash uses architecture and runtime components from
[ResembleAI/chatterbox](https://github.com/resemble-ai/chatterbox).

Copyright (c) 2025 Resemble AI. Licensed under the MIT License.

## Runtime dependencies

[FlashInfer](https://github.com/flashinfer-ai/flashinfer) is an Apache-2.0
runtime dependency and is not vendored.

`nvidia-cutlass-dsl` and `nvidia-cutlass-dsl-libs-base` are installed from
their official Python packages and are not vendored. They are distributed
under the NVIDIA Software License Agreement included in those packages. Users
must review and accept that agreement when installing or using the dependency.

NVIDIA, CUDA, and Blackwell are trademarks and do not imply endorsement.