OpenMOSE

ASR for RWKV-7 G1 series. this is hard fork of mod-rwkv.

Unified ASR & TTS inference for Qwen3 — speech-to-text and text-to-speech in a single package.

No description.

The WorldRWKV project aims to implement training and inference across various modalities using the RWKV7 architecture. By leveraging different encoders, the project allows for flexible modality switching and aspires to achieve end-to-end cross-modal inference.