Merge pull request #164 from breakstring/main

Docker support

Docker/damo.sha256

@@ -0,0 +1,3 @@
+5bba782a5e9196166233b9ab12ba04cadff9ef9212b4ff6153ed9290ff679025 /workspace/tools/damo_asr/models/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/model.pb
+b3be75be477f0780277f3bae0fe489f48718f585f3a6e45d7dd1fbb1a4255fc5 /workspace/tools/damo_asr/models/speech_fsmn_vad_zh-cn-16k-common-pytorch/model.pb
+a5818bb9d933805a916eebe41eb41648f7f9caad30b4bd59d56f3ca135421916 /workspace/tools/damo_asr/models/punc_ct-transformer_zh-cn-common-vocab272727-pytorch/model.pb

Docker/download.sh

@@ -0,0 +1,11 @@
+#!/usr/bin/env bash
+
+set -Eeuo pipefail
+
+echo "Downloading models..."
+
+aria2c --disable-ipv6 --input-file /workspace/Docker/links.txt --dir /workspace --continue
+
+echo "Checking SHA256..."
+
+parallel --will-cite -a /workspace/Docker/links.sha256 "echo -n {} | sha256sum -c"

Docker/links.sha256

@@ -0,0 +1,12 @@
+b1c1e17e9c99547a89388f72048cd6e1b41b5a18b170e86a46dfde0324d63eb1 /workspace/GPT_SoVITS/pretrained_models/s1bert25hz-2kh-longer-epoch=68e-step=50232.ckpt
+fc579c1db3c1e21b721001cf99d7a584214280df19b002e200b630a34fa06eb8 /workspace/GPT_SoVITS/pretrained_models/s2D488k.pth
+020a014e1e01e550e510f2f61fae5e5f5b6aab40f15c22f1f12f724df507e835 /workspace/GPT_SoVITS/pretrained_models/s2G488k.pth
+24164f129c66499d1346e2aa55f183250c223161ec2770c0da3d3b08cf432d3c /workspace/GPT_SoVITS/pretrained_models/chinese-hubert-base/pytorch_model.bin
+e53a693acc59ace251d143d068096ae0d7b79e4b1b503fa84c9dcf576448c1d8 /workspace/GPT_SoVITS/pretrained_models/chinese-roberta-wwm-ext-large/pytorch_model.bin
+39796caa5db18d7f9382d8ac997ac967bfd85f7761014bb807d2543cc844ef05 /workspace/tools/uvr5/uvr5_weights/HP2_all_vocals.pth
+45e6b65199e781b4a6542002699be9f19cd3d1cb7d1558bc2bfbcd84674dfe28 /workspace/tools/uvr5/uvr5_weights/HP3_all_vocals.pth
+5908891829634926119720241e8573d97cbeb8277110a7512bdb0bd7563258ee /workspace/tools/uvr5/uvr5_weights/HP5_only_main_vocal.pth
+8c8fd1582f9aabc363e47af62ddb88df6cae7e064cae75bbf041a067a5e0aee2 /workspace/tools/uvr5/uvr5_weights/VR-DeEchoAggressive.pth
+01376dd2a571bf3cb9cced680732726d2d732609d09216a610b0d110f133febe /workspace/tools/uvr5/uvr5_weights/VR-DeEchoDeReverb.pth
+56aba59db3bcdd14a14464e62f3129698ecdea62eee0f003b9360923eb3ac79e /workspace/tools/uvr5/uvr5_weights/VR-DeEchoNormal.pth
+233bb5c6aaa365e568659a0a81211746fa881f8f47f82d9e864fce1f7692db80 /workspace/tools/uvr5/uvr5_weights/onnx_dereverb_By_FoxJoy/vocals.onnx

Docker/links.txt

@@ -0,0 +1,34 @@
+# GPT-SoVITS models
+https://huggingface.co/lj1995/GPT-SoVITS/resolve/main/s1bert25hz-2kh-longer-epoch%3D68e-step%3D50232.ckpt
+  out=GPT_SoVITS/pretrained_models/s1bert25hz-2kh-longer-epoch=68e-step=50232.ckpt
+https://huggingface.co/lj1995/GPT-SoVITS/resolve/main/s2D488k.pth
+  out=GPT_SoVITS/pretrained_models/s2D488k.pth
+https://huggingface.co/lj1995/GPT-SoVITS/resolve/main/s2G488k.pth
+  out=GPT_SoVITS/pretrained_models/s2G488k.pth
+https://huggingface.co/lj1995/GPT-SoVITS/resolve/main/chinese-hubert-base/config.json
+  out=GPT_SoVITS/pretrained_models/chinese-hubert-base/config.json
+https://huggingface.co/lj1995/GPT-SoVITS/resolve/main/chinese-hubert-base/preprocessor_config.json
+  out=GPT_SoVITS/pretrained_models/chinese-hubert-base/preprocessor_config.json
+https://huggingface.co/lj1995/GPT-SoVITS/resolve/main/chinese-hubert-base/pytorch_model.bin
+  out=GPT_SoVITS/pretrained_models/chinese-hubert-base/pytorch_model.bin
+https://huggingface.co/lj1995/GPT-SoVITS/resolve/main/chinese-roberta-wwm-ext-large/config.json
+  out=GPT_SoVITS/pretrained_models/chinese-roberta-wwm-ext-large/config.json
+https://huggingface.co/lj1995/GPT-SoVITS/resolve/main/chinese-roberta-wwm-ext-large/pytorch_model.bin
+  out=GPT_SoVITS/pretrained_models/chinese-roberta-wwm-ext-large/pytorch_model.bin
+https://huggingface.co/lj1995/GPT-SoVITS/resolve/main/chinese-roberta-wwm-ext-large/tokenizer.json
+  out=GPT_SoVITS/pretrained_models/chinese-roberta-wwm-ext-large/tokenizer.json
+# UVR5
+https://huggingface.co/lj1995/VoiceConversionWebUI/resolve/main/uvr5_weights/HP2_all_vocals.pth
+  out=tools/uvr5/uvr5_weights/HP2_all_vocals.pth
+https://huggingface.co/lj1995/VoiceConversionWebUI/resolve/main/uvr5_weights/HP3_all_vocals.pth
+  out=tools/uvr5/uvr5_weights/HP3_all_vocals.pth
+https://huggingface.co/lj1995/VoiceConversionWebUI/resolve/main/uvr5_weights/HP5_only_main_vocal.pth
+  out=tools/uvr5/uvr5_weights/HP5_only_main_vocal.pth
+https://huggingface.co/lj1995/VoiceConversionWebUI/resolve/main/uvr5_weights/VR-DeEchoAggressive.pth
+  out=tools/uvr5/uvr5_weights/VR-DeEchoAggressive.pth
+https://huggingface.co/lj1995/VoiceConversionWebUI/resolve/main/uvr5_weights/VR-DeEchoDeReverb.pth
+  out=tools/uvr5/uvr5_weights/VR-DeEchoDeReverb.pth
+https://huggingface.co/lj1995/VoiceConversionWebUI/resolve/main/uvr5_weights/VR-DeEchoNormal.pth
+  out=tools/uvr5/uvr5_weights/VR-DeEchoNormal.pth
+https://huggingface.co/lj1995/VoiceConversionWebUI/resolve/main/uvr5_weights/onnx_dereverb_By_FoxJoy/vocals.onnx
+  out=tools/uvr5/uvr5_weights/onnx_dereverb_By_FoxJoy/vocals.onnx

Dockerfile

@@ -0,0 +1,53 @@
+# Base CUDA image
+FROM cnstark/pytorch:2.0.1-py3.9.17-cuda11.8.0-ubuntu20.04
+
+LABEL maintainer="breakstring@hotmail.com"
+LABEL version="dev-20240123.03"
+LABEL description="Docker image for GPT-SoVITS"
+
+
+# Install 3rd party apps
+ENV DEBIAN_FRONTEND=noninteractive
+ENV TZ=Etc/UTC
+RUN apt-get update && \
+    apt-get install -y --no-install-recommends tzdata ffmpeg libsox-dev parallel aria2 git git-lfs && \
+    rm -rf /var/lib/apt/lists/* && \
+    git lfs install
+
+# Copy application
+WORKDIR /workspace
+COPY . /workspace
+
+# Download models
+RUN chmod +x /workspace/Docker/download.sh && /workspace/Docker/download.sh
+
+# 本应该从 requirements.txt 里面安装package，但是由于funasr和modelscope的问题，暂时先在后面手工安装依赖包吧
+RUN pip install --no-cache-dir torch numpy scipy tensorboard librosa==0.9.2 numba==0.56.4 pytorch-lightning gradio==3.14.0 ffmpeg-python onnxruntime tqdm cn2an pypinyin pyopenjtalk g2p_en chardet transformers jieba psutil PyYAML
+# 这里强制指定了modelscope和funasr的版本，后面damo_asr的模型让它们自己下载
+RUN pip install --no-cache-dir modelscope~=1.10.0 torchaudio sentencepiece funasr~=0.8.7
+
+# 先屏蔽掉，让容器里自己下载
+# Clone damo_asr
+#WORKDIR /workspace/tools/damo_asr/models
+#RUN git clone --depth 1 https://www.modelscope.cn/iic/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch.git speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch && \
+#    (cd speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch && git lfs pull)
+#RUN git clone --depth 1 https://www.modelscope.cn/iic/speech_fsmn_vad_zh-cn-16k-common-pytorch.git speech_fsmn_vad_zh-cn-16k-common-pytorch && \
+#    (cd speech_fsmn_vad_zh-cn-16k-common-pytorch && git lfs pull)
+#RUN git clone --depth 1 https://www.modelscope.cn/iic/punc_ct-transformer_zh-cn-common-vocab272727-pytorch.git punc_ct-transformer_zh-cn-common-vocab272727-pytorch && \
+#    (cd punc_ct-transformer_zh-cn-common-vocab272727-pytorch && git lfs pull)
+
+#RUN parallel --will-cite -a /workspace/Docker/damo.sha256 "echo -n {} | sha256sum -c"
+
+#WORKDIR /workspace
+
+EXPOSE 9870
+EXPOSE 9871
+EXPOSE 9872
+EXPOSE 9873
+EXPOSE 9874
+
+VOLUME /workspace/output
+VOLUME /workspace/logs
+VOLUME /workspace/SoVITS_weights
+
+CMD ["python", "webui.py"]

README.md

@@ -107,6 +107,31 @@ For Chinese ASR (additionally), download models from [Damo ASR Model](https://mo
 For UVR5 (Vocals/Accompaniment Separation & Reverberation Removal, additionally), download models from [UVR5 Weights](https://huggingface.co/lj1995/VoiceConversionWebUI/tree/main/uvr5_weights) and place them in `tools/uvr5/uvr5_weights`.
 
 
+### Using Docker
+
+#### docker-compose.yaml configuration
+
+1. Environment Variables：
+  - is_half: Controls half-precision/double-precision. This is typically the cause if the content under the directories 4-cnhubert/5-wav32k is not generated correctly during the "SSL extracting" step. Adjust to True or False based on your actual situation.
+
+2. Volumes Configuration，The application's root directory inside the container is set to /workspace. The default docker-compose.yaml lists some practical examples for uploading/downloading content.
+3. shm_size： The default available memory for Docker Desktop on Windows is too small, which can cause abnormal operations. Adjust according to your own situation.
+4. Under the deploy section, GPU-related settings should be adjusted cautiously according to your system and actual circumstances.
+
+
+#### Running with docker compose
+```
+docker compose -f "docker-compose.yaml" up -d
+```
+
+#### Running with docker command
+
+As above, modify the corresponding parameters based on your actual situation, then run the following command:
+```
+docker run --rm -it --gpus=all --env=is_half=False --volume=G:\GPT-SoVITS-DockerTest\output:/workspace/output --volume=G:\GPT-SoVITS-DockerTest\logs:/workspace/logs --volume=G:\GPT-SoVITS-DockerTest\SoVITS_weights:/workspace/SoVITS_weights --workdir=/workspace -p 9870:9870 -p 9871:9871 -p 9872:9872 -p 9873:9873 -p 9874:9874 --shm-size="16G" -d breakstring/gpt-sovits:dev-20240123.03
+```
+
+
 ## Dataset Format
 
 The TTS annotation .list file format:

config.py

@@ -1,10 +1,10 @@
-import sys
+import sys,os
 
 
 # 推理用的指定模型
 sovits_path = ""
 gpt_path = ""
-is_half = True
+is_half = eval(os.environ.get("is_half",True))
 is_share=False
 
 cnhubert_path = "GPT_SoVITS/pretrained_models/chinese-hubert-base"

docker-compose.yaml

@@ -0,0 +1,31 @@
+version: '3.8'
+
+services:
+  gpt-sovits:
+    image: breakstring/gpt-sovits:dev-20240123.03
+    container_name: gpt-sovits-container
+    environment:
+      - is_half=False
+    volumes:
+      - G:/GPT-SoVITS-DockerTest/output:/workspace/output
+      - G:/GPT-SoVITS-DockerTest/logs:/workspace/logs
+      - G:/GPT-SoVITS-DockerTest/SoVITS_weights:/workspace/SoVITS_weights
+      - G:/GPT-SoVITS-DockerTest/reference:/workspace/reference
+    working_dir: /workspace
+    ports:
+      - "9870:9870"
+      - "9871:9871"
+      - "9872:9872"
+      - "9873:9873"
+      - "9874:9874"
+    shm_size: 16G
+    deploy:
+      resources:
+        reservations:
+          devices:
+          - driver: nvidia
+            count: "all"
+            capabilities: [gpu]
+    stdin_open: true
+    tty: true
+    restart: unless-stopped

docs/cn/README.md

@@ -87,6 +87,32 @@ brew install ffmpeg
 
 下载并将 [ffmpeg.exe](https://huggingface.co/lj1995/VoiceConversionWebUI/blob/main/ffmpeg.exe) 和 [ffprobe.exe](https://huggingface.co/lj1995/VoiceConversionWebUI/blob/main/ffprobe.exe) 放置在 GPT-SoVITS 根目录下。
 
+### 在 Docker 中使用
+
+#### docker-compose.yaml 设置
+
+1. 环境变量：
+  - is_half: 半精度/双精度控制。在进行 "SSL extracting" 步骤时如果无法正确生成 4-cnhubert/5-wav32k 目录下的内容时，一般都是它引起的，可以根据实际情况来调整为True或者False。
+
+2. Volume设置，容器内的应用根目录设置为 /workspace。 默认的 docker-compose.yaml 中列出了一些实际的例子，便于上传/下载内容。
+3. shm_size：Windows下的Docker Desktop默认可用内存过小，会导致运行异常，根据自己情况酌情设置。
+4. deploy小节下的gpu相关内容，请根据您的系统和实际情况酌情设置。
+
+
+
+#### 通过 docker compose运行
+```
+docker compose -f "docker-compose.yaml" up -d
+```
+
+#### 通过 docker 命令运行
+
+同上，根据您自己的实际情况修改对应的参数，然后运行如下命令：
+```
+docker run --rm -it --gpus=all --env=is_half=False --volume=G:\GPT-SoVITS-DockerTest\output:/workspace/output --volume=G:\GPT-SoVITS-DockerTest\logs:/workspace/logs --volume=G:\GPT-SoVITS-DockerTest\SoVITS_weights:/workspace/SoVITS_weights --workdir=/workspace -p 9870:9870 -p 9871:9871 -p 9872:9872 -p 9873:9873 -p 9874:9874 --shm-size="16G" -d breakstring/gpt-sovits:dev-20240123.03
+```
+
+
 ### 预训练模型
 
 

docs/ja/README.md

@@ -93,6 +93,30 @@ brew install ffmpeg
 
 [ffmpeg.exe](https://huggingface.co/lj1995/VoiceConversionWebUI/blob/main/ffmpeg.exe) と [ffprobe.exe](https://huggingface.co/lj1995/VoiceConversionWebUI/blob/main/ffprobe.exe) をダウンロードし、GPT-SoVITS のルートディレクトリに置きます。
 
+### Dockerの使用
+
+#### docker-compose.yamlの設定
+
+1. 環境変数：
+    - `is_half`：半精度／倍精度の制御。"SSL抽出"ステップ中に`4-cnhubert/5-wav32k`ディレクトリ内の内容が正しく生成されない場合、通常これが原因です。実際の状況に応じてTrueまたはFalseに調整してください。
+
+2. ボリューム設定：コンテナ内のアプリケーションのルートディレクトリは`/workspace`に設定されます。デフォルトの`docker-compose.yaml`には、アップロード／ダウンロードの内容の実例がいくつか記載されています。
+3. `shm_size`：WindowsのDocker Desktopのデフォルトの利用可能メモリが小さすぎるため、異常な動作を引き起こす可能性があります。状況に応じて適宜設定してください。
+4. `deploy`セクションのGPUに関連する内容は、システムと実際の状況に応じて慎重に設定してください。
+
+#### docker composeで実行する
+```markdown
+docker compose -f "docker-compose.yaml" up -d
+```
+
+#### dockerコマンドで実行する
+
+上記と同様に、実際の状況に基づいて対応するパラメータを変更し、次のコマンドを実行します：
+```markdown
+docker run --rm -it --gpus=all --env=is_half=False --volume=G:\GPT-SoVITS-DockerTest\output:/workspace/output --volume=G:\GPT-SoVITS-DockerTest\logs:/workspace/logs --volume=G:\GPT-SoVITS-DockerTest\SoVITS_weights:/workspace/SoVITS_weights --workdir=/workspace -p 9870:9870 -p 9871:9871 -p 9872:9872 -p 9873:9873 -p 9874:9874 --shm-size="16G" -d breakstring/gpt-sovits:dev-20240123.03
+```
+
+
 ### 事前訓練済みモデル