丛雨 TTS — Model Weights
Fine-tuned model weights for murasame-index-tts, a TTS voice pack based on IndexTTS 2.5 (bilibili indextts2).
Derivative Work Notice
Any modifications made to the original model in this Derivative Work are not endorsed, warranted, or guaranteed by the original right-holder of the original model, and the original right-holder disclaims all liability related to this Derivative Work.
Files
| 文件 | 说明 | 大小 |
|---|---|---|
gpt.pth |
微调主模型（LoRA step600） | 3.1 GB |
codec.pth |
编解码器（来自原项目） | 579 MB |
s2mel.pth |
Mel 生成模块（来自原项目） | 396 MB |
hf_cache/bigvgan/bigvgan_generator.pt |
BigVGAN 声码器 | 428 MB |
hf_cache/campplus_cn_common.bin |
说话人编码器 | 27 MB |
hf_cache/semantic_codec_model.safetensors |
语义编解码器 | 169 MB |
hf_cache/semantic_codec/model.safetensors |
语义编解码器（副本） | 169 MB |
小型辅助文件（feat1.pt, feat2.pt, config.yaml, reference.ogg 等）随代码仓库分发，无需从此处下载。
Usage
下载
pip install huggingface_hub
hf download YOUR_HF_USERNAME/murasame-index-tts-weights --local-dir ./checkpoints
推理
# 克隆代码并安装依赖
git clone https://github.com/YOUR_GITHUB_USERNAME/murasame-index-tts.git
cd murasame-index-tts
install.bat # Windows
# bash install.sh # Linux
# 命令行推理
python infer.py "ご主人、今日もよろしく。" -o out.wav
详细说明见 GitHub 仓库。
Training Details
- Base model: IndexTTS 2.5 (
bilibili indextts2) - Fine-tuning method: LoRA (rank=8, alpha=16, last 8 Transformer layers)
- Training data: Character voice lines from the game featuring Murasame
- Best checkpoint: step600, training loss 5.76
- Primary language: Japanese (also supports Chinese)
License
This model is a Derivative Work of bilibili indextts2 and is subject to the
bilibili Model Use License Agreement.
- Non-commercial use only.
- The character voice data used for fine-tuning is the property of its respective copyright holders.
- The authors of this repository are not responsible for any misuse.
- Downloads last month
- 8