Views
No views yet
phei_watermark_bundle.tar.gz(2.8 GB),解压即可。1# aria2c 多线程(最快,推荐)
2apt-get install -y aria2 # 没装的话先装
3aria2c -x 16 -s 16 -o phei_watermark_bundle.tar.gz \
4 https://hf-mirror.com/taopeng/phei-watermark-fish-speech-1.5/resolve/main/phei_watermark_bundle.tar.gz
5
6# 或 wget 断点续传
7wget -c -O phei_watermark_bundle.tar.gz \
8 https://hf-mirror.com/taopeng/phei-watermark-fish-speech-1.5/resolve/main/phei_watermark_bundle.tar.gz
9
10# 或 hf cli + 镜像
11pip install -U "huggingface_hub[cli]"
12export HF_ENDPOINT=https://hf-mirror.com
13hf download taopeng/phei-watermark-fish-speech-1.5 phei_watermark_bundle.tar.gz --local-dir .1wget -O phei_watermark_bundle.tar.gz \
2 https://huggingface.co/taopeng/phei-watermark-fish-speech-1.5/resolve/main/phei_watermark_bundle.tar.gz1tar xzf phei_watermark_bundle.tar.gz
2cd phei_watermark_bundle
3bash RUN.sheinops / einx / ijson / frozendict / vector_quantize_pytorch。
你的 Python 环境还需要安装基础依赖:pip install torch torchaudio loguru hydra-core lightning soundfile librosa numpyoutputs/
├── clean/output_clean.wav # 不带水印
└── watermarked/output_wm.wav # 带水印 + 检测结果Detection: detected=True, id_match=True, fused_conf=0.844
vq: detected=False conf=0.523 thr=0.7
audio: detected=True conf=0.672 thr=0.5README.md。| 文件 | 大小 | 说明 |
|---|---|---|
phei_watermark_bundle.tar.gz | 2.8 GB | 一键运行包(含模型 + 推理代码 + 示例 prompt) |
step_000001500.ckpt | 3.6 GB | Stage 3 水印 ckpt(裸文件,需要自行配 fish-speech 环境) |
training_config.yaml | — | Stage 3 训练 Hydra 配置 |
stage2_config.yaml | — | Stage 2 域适应训练配置 |
base_model_config.json | — | 主干 (Slow+Fast Transformer) 结构 |
1git clone https://github.com/fishaudio/fish-speech # 用 v1.5 tag
2cd fish-speech
3# ... 加上本仓库 phei_watermark/ 下的水印模块 + Stage 3 ckpttools/synthesize_with_watermark.py,完全自包含。1@thesis{phei2026watermark,
2 title={面向大规模 VQ-TTS 模型的多层水印系统设计与实现},
3 school={北京电子科技学院},
4 year={2026}
5}