This is a sample LoRA model of VibeVoice trained on MrDragonFox's Elise dataset. It is intended to 1) demonstrate the capabilities of LoRA and 2) show the format in which LoRA fine-tuned models are saved.

Despite only be training on short audio clips (<10 seconds), the adapter still supports long-form audio generation.

This LoRA was trained without reference audio.

NOTE: This is not an official model from Microsoft or the authors of VibeVoice, it is simply a community-trained LoRA intended to showcase how to use the LoRA scripts.