diff --git a/README.md b/README.md
index ab93cef4..ef7fec7f 100644
--- a/README.md
+++ b/README.md
@@ -12,7 +12,7 @@
 
 - 💗本项目的目标群体是：深度学习初学者，具备Python和PyTorch的基本操作是使用本项目的前置条件；
 - 💗本项目旨在帮助深度学习初学者，摆脱枯燥的纯理论学习，通过与实践结合，熟练掌握深度学习基本知识；
-- 💗本项目不支持实时变声；（也许以后会支持，但要替换掉whisper）
+- 💗本项目不支持实时变声；（支持需要换掉whisper）
 - 💗本项目不会开发用于其他用途的一键包。（不会指没学会）
 
 ![sovits_framework](https://github.com/PlayVoice/so-vits-svc-5.0/assets/16432329/402cf58d-6d03-4d0b-9d6a-94f079898672)
@@ -29,43 +29,37 @@
 
 本项目并不基于svc-develop-team/so-vits-svc，恰恰相反，见https://github.com/svc-develop-team/so-vits-svc/tree/2.0
 
-本项目将继续完成基于BIGVGAN的模型（32K），在此之后，有成果再更新项目
+## 模型和日志：
 
-## 模型和日志：https://github.com/PlayVoice/so-vits-svc-5.0/releases/tag/base_release_hifigan
+https://github.com/PlayVoice/so-vits-svc-5.0/releases/tag/base_release_hifigan
 
 - [5.0.epoch1200.full.pth](https://github.com/PlayVoice/so-vits-svc-5.0/releases/download/base_release_hifigan/5.0.epoch1200.full.pth)模型包括：生成器+判别器=176M，可用作预训练模型
 - 发音人（56个）文件在configs/singers目录中，可进行推理测试，尤其测试音色泄露
-- 发音人22，30，47，51辨识度较高，音频样本在configs/singers_sample目录中
+- 发音人22，30，47，51辨识度较高，训练音频样本在configs/singers_sample目录中
 
 | Feature | From | Status | Function | Remarks |
 | --- | --- | --- | --- | --- |
 | whisper | OpenAI | ✅ | 强大的抗噪能力 | 参数修改 |
-| bigvgan  | NVIDA | ✅ | 抗锯齿与蛇形激活 | GPU占用略多，主分支删除；新分支训练，共振峰更清晰，提升音质明显 |
+| bigvgan  | NVIDA | ✅ | 抗锯齿与蛇形激活 | GPU占用略多，主分支删除；新bigvgan分支训练，共振峰更清晰，提升音质明显 |
 | natural speech | Microsoft | ✅ | 减少发音错误 | - |
 | neural source-filter | NII | ✅ | 解决断音问题 | 参数优化 |
 | speaker encoder | Google | ✅ | 音色编码与聚类 | - |
 | GRL for speaker | Ubisoft |✅ | 防止编码器泄漏音色 | 原理类似判别器的对抗训练 |
 | one shot vits |  Samsung | ✅ | VITS 一句话克隆 | - |
 | SCLN |  Microsoft | ✅ | 改善克隆 | - |
-| band extention | Adobe | ✅ | 16K升48K采样 | 数据处理 |
 | PPG perturbation | 本项目 | ✅ | 提升抗噪性和去音色 | - |
 
-💗GRL去音色泄漏，更多的是理论上的价值；Hugging Face Demo推理模型无泄漏主要归因于PPG扰动；由于使用了数据扰动，相比其他项目需要更长的训练时间。
+💗Hugging Face Demo推理模型无泄漏主要归因于PPG扰动；由于使用了数据扰动，相比其他项目需要更长的训练时间。
 
 ## 数据集准备
-<div align="center">
-    
-![uvr5_config](https://github.com/PlayVoice/vits_chinese/assets/16432329/f72fd2fa-0f05-4da1-bb0b-f29d0c20ddbf)
-    
-</div>
 
-💗必要的前处理：
-- 1 降噪&去伴奏
+必要的前处理：
+- 1 伴奏分离
 - 2 频率提升
 - 3 音质提升
-- 4 将音频剪裁为小于30秒的音频段，whisper的要求
+- 4 剪切音频，whisper要求为小于30秒💗
 
-然后以下面文件结构将数据集放入dataset_raw目录
+然后按下面文件结构，将数据集放入dataset_raw目录
 ```shell
 dataset_raw
 ├───speaker0
@@ -137,52 +131,41 @@ dataset_raw
 
 ```shell
 data_svc/
-│
 └── waves-16k
-│    │
 │    └── speaker0
 │    │      ├── 000001.wav
 │    │      └── 000xxx.wav
 │    └── speaker1
 │           ├── 000001.wav
 │           └── 000xxx.wav
-│
 └── waves-32k
-│    │
 │    └── speaker0
 │    │      ├── 000001.wav
 │    │      └── 000xxx.wav
 │    └── speaker1
 │           ├── 000001.wav
 │           └── 000xxx.wav
-│
 └── pitch
-│    │
 │    └── speaker0
 │    │      ├── 000001.pit.npy
 │    │      └── 000xxx.pit.npy
 │    └── speaker1
 │           ├── 000001.pit.npy
 │           └── 000xxx.pit.npy
-│
 └── whisper
-│    │
 │    └── speaker0
 │    │      ├── 000001.ppg.npy
 │    │      └── 000xxx.ppg.npy
 │    └── speaker1
 │           ├── 000001.ppg.npy
 │           └── 000xxx.ppg.npy
-│
 └── speaker
-│    │
 │    └── speaker0
 │    │      ├── 000001.spk.npy
 │    │      └── 000xxx.spk.npy
 │    └── speaker1
 │           ├── 000001.spk.npy
 │           └── 000xxx.spk.npy
-|
 └── singer
     ├── speaker0.spk.npy
     └── speaker1.spk.npy
@@ -191,7 +174,7 @@ data_svc/
 ## 训练
 - 0， 如果基于预训练模型微调，需要下载预训练模型5.0.epoch1200.full.pth
 
-    > 指定configs/base.yaml参数pretrain: ""，并适当调小学习率
+    > 指定configs/base.yaml参数pretrain: "./5.0.epoch1200.full.pth"，并适当调小学习率
 
 - 1， 设置工作目录:heartpulse::heartpulse::heartpulse:不设置后面会报错
 
@@ -273,7 +256,7 @@ eva_conf = {
 
 生成的音色文件为：eva.spk.npy
 
-💗Flow和Decoder均需要输入，您甚至可以给两个模块输入不同的音色参数，捏出更独特的音色。
+💗Flow和Decoder均需要输入音色，您甚至可以给两个模块输入不同的音色参数，捏出更独特的音色。
 
 ## 数据集
 
diff --git a/customize/bandex/LICENSE b/customize/bandex/LICENSE
deleted file mode 100644
index 221561ab..00000000
--- a/customize/bandex/LICENSE
+++ /dev/null
@@ -1,21 +0,0 @@
-MIT License
-
-Copyright © 2022 Brent M. Spell
-
-Permission is hereby granted, free of charge, to any person obtaining a copy
-of this software and associated documentation files (the "Software"), to deal
-in the Software without restriction, including without limitation the rights
-to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
-copies of the Software, and to permit persons to whom the Software is
-furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice shall be included in all
-copies or substantial portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
-OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
-SOFTWARE.
diff --git a/customize/bandex/README.md b/customize/bandex/README.md
deleted file mode 100644
index 6134adef..00000000
--- a/customize/bandex/README.md
+++ /dev/null
@@ -1,3 +0,0 @@
-Model from:
-
-https://github.com/brentspell/hifi-gan-bwe
\ No newline at end of file
diff --git a/customize/bandex/hifi-gan-bwe-vctk-48kHz.pt b/customize/bandex/hifi-gan-bwe-vctk-48kHz.pt
deleted file mode 100644
index d4a24752..00000000
Binary files a/customize/bandex/hifi-gan-bwe-vctk-48kHz.pt and /dev/null differ
diff --git a/customize/bandex/inference.py b/customize/bandex/inference.py
deleted file mode 100644
index 35b9e89a..00000000
--- a/customize/bandex/inference.py
+++ /dev/null
@@ -1,71 +0,0 @@
-import os
-import argparse
-import librosa
-import torch
-from scipy.io.wavfile import write
-import numpy as np
-
-SCALE = 3
-
-
-def main(args):
-    device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
-    model = torch.jit.load(os.path.join(
-        "bandex", "hifi-gan-bwe-vctk-48kHz.pt")).to(device)
-    x, sr = librosa.load(args.wave, sr=16000)
-
-    hop_size = 320  # whisper hop size
-    hop_count = len(x) // hop_size
-    hop_frame = 10
-    bwe_chunk = 500  # 10S
-    bwe_index = 0
-    bwe_audio = []
-    has_audio = False
-
-    while (bwe_index + bwe_chunk < hop_count):
-        has_audio = True
-        if (bwe_index == 0):  # start frame
-            cut_s_16k = 0
-            cut_s_48k = 0
-        else:
-            cut_s_16k = (bwe_index - hop_frame) * hop_size
-            cut_s_48k = hop_frame * hop_size * SCALE
-
-        if (bwe_index + bwe_chunk + hop_frame > hop_count):  # end frame
-            cut_e_16k = (bwe_index + bwe_chunk) * hop_size
-            cut_e_48k = 0
-        else:
-            cut_e_16k = (bwe_index + bwe_chunk + hop_frame) * hop_size
-            cut_e_48k = -1 * hop_frame * hop_size * SCALE
-        x_chunk = x[cut_s_16k:cut_e_16k]
-
-        with torch.no_grad():
-            i_audio = torch.from_numpy(x_chunk).to(device)
-            o_audio = model(i_audio, sr).data.cpu().float().numpy()
-        o_audio = o_audio[cut_s_48k:cut_e_48k]
-        bwe_audio.extend(o_audio)
-        bwe_index = bwe_index + bwe_chunk
-
-    if (bwe_index < hop_count):
-        if (has_audio):
-            cut_s_16k = bwe_index - hop_frame
-            cut_s_48k = hop_frame * hop_size * SCALE
-        else:
-            cut_s_16k = 0
-            cut_s_48k = 0
-        x_chunk = x[cut_s_16k * hop_size:]
-        with torch.no_grad():
-            i_audio = torch.from_numpy(x_chunk).to(device)
-            o_audio = model(i_audio, sr).data.cpu().float().numpy()
-        o_audio = o_audio[cut_s_48k:]
-        bwe_audio.extend(o_audio)
-    bwe_audio = np.asarray(bwe_audio)
-    write("svc_out_48k.wav", 48000, bwe_audio)
-
-
-if __name__ == '__main__':
-    parser = argparse.ArgumentParser()
-    parser.add_argument('-w', '--wave', type=str, required=True,
-                        help="Path of raw audio.")
-    args = parser.parse_args()
-    main(args)