Merge pull request Ikaros-521#130 from Ikaros-521/owner

GUI适配DDSP-SVC；补充说明文档

README.md

@@ -800,11 +800,36 @@ ChatterBot 的核心思想是：基于历史对话数据，使用机器学习和
 个人提供的已训练好的模型：[点我跳转](https://github.com/Ikaros-521/VITS-fast-fine-tuning/releases)  
 视频教程：[【AI语音合成-数据集篇#00】模型使用简单讲解，手把手演示](https://www.bilibili.com/video/BV1Lm4y1r7Pi)  
 
+### DDSP-SVC
+官方仓库：[DDSP-SVC](https://github.com/yxlllc/DDSP-SVC)  
+个人提供的已训练好的模型：[点我跳转](https://github.com/Ikaros-521/DDSP-SVC/releases)  
+羽毛佬视频教程：[【AI翻唱/变声/整合包】有张N卡就能跑！媲美So-VITS却不吃配置，全新的DDSP-SVC 3.0训练/推理教程](https://www.bilibili.com/video/BV1rs4y1Q7BQ)  
+修改`flask_api.py`最后几行中的模型路径，其他配置自行调整，另外需要注意！`enable_spk_id_cover`改为`False`，即可接入。  
+```
+# config和模型得同一目录。 这是我自己训练的DDSP模型，新鲜出炉
+checkpoint_path = "exp/combsub-test/model_68000.pt"
+# 是否使用预训练的基于声码器的增强器增强输出，但对硬件要求更高。
+use_vocoder_based_enhancer = True
+# 结合增强器使用，0为正常音域范围（最高G5)内的高音频质量，大于0则可以防止超高音破音
+enhancer_adaptive_key = 0
+# f0提取器，有parselmouth, dio, harvest, crepe
+select_pitch_extractor = 'crepe'
+# f0范围限制(Hz)
+limit_f0_min = 50
+limit_f0_max = 1100
+# 音量响应阈值(dB)
+threhold = -60
+# 默认说话人。
+spk_id = 1
+# 是否优先使用默认说话人覆盖vst传入的参数，改为False，不要覆盖！！！
+enable_spk_id_cover = False
+```
+
 ### so-vits-svc
 官方仓库：[so-vits-svc](https://github.com/svc-develop-team/so-vits-svc)  
 个人提供的已训练好的模型：[点我跳转](https://github.com/Ikaros-521/so-vits-svc/releases)  
 视频教程：[【AI歌姬】so-vits-svc 简单使用讲解 + 伊卡洛斯 模型开源](https://www.bilibili.com/video/BV1k24y1F7Us)  
-修改`flask_api_full_song.py`最后几行中的模型和配置路径，运行，配置相关信息即可接入。  
+修改`flask_api_full_song.py`最后几行中的模型路径，运行，配置相关信息即可接入。  
 ```
 if __name__ == '__main__':
     model_name = "logs/44k/ikaros_G_54600.pth" # 你的模型地址
@@ -1062,6 +1087,9 @@ cmd输入命令即可：`doctoc /path/to/file`
 - 优化抖音日志信息，只输出info信息，不在输出debug日志，更加清爽。
 - 接入DDSP-SVC，简单测试了下，效果好像不太对。优化了变声部分的代码
 
+### 2023-07-19
+- GUI适配DDSP-SVC；补充说明文档
+
 </details>
 
 

UI_main.py

@@ -86,7 +86,7 @@ def setupUi(self, MainWindow):
         self.scrollArea.setWidgetResizable(True)
         self.scrollArea.setObjectName("scrollArea")
         self.scrollAreaWidgetContents = QtWidgets.QWidget()
-        self.scrollAreaWidgetContents.setGeometry(QtCore.QRect(0, 0, 980, 5794))
+        self.scrollAreaWidgetContents.setGeometry(QtCore.QRect(0, -4345, 980, 6162))
         self.scrollAreaWidgetContents.setObjectName("scrollAreaWidgetContents")
         self.verticalLayout = QtWidgets.QVBoxLayout(self.scrollAreaWidgetContents)
         self.verticalLayout.setContentsMargins(35, 20, 35, 20)
@@ -746,6 +746,56 @@ def setupUi(self, MainWindow):
         self.gridLayout_33.addWidget(self.lineEdit_choose_song_match_fail_copy, 4, 1, 1, 1)
         self.gridLayout_34.addLayout(self.gridLayout_33, 0, 0, 1, 1)
         self.verticalLayout.addWidget(self.groupBox_choose_song)
+        self.groupBox_ddsp_svc = QtWidgets.QGroupBox(self.scrollAreaWidgetContents)
+        self.groupBox_ddsp_svc.setObjectName("groupBox_ddsp_svc")
+        self.gridLayout_49 = QtWidgets.QGridLayout(self.groupBox_ddsp_svc)
+        self.gridLayout_49.setObjectName("gridLayout_49")
+        self.gridLayout_48 = QtWidgets.QGridLayout()
+        self.gridLayout_48.setObjectName("gridLayout_48")
+        self.label_ddsp_svc_enable = QtWidgets.QLabel(self.groupBox_ddsp_svc)
+        self.label_ddsp_svc_enable.setObjectName("label_ddsp_svc_enable")
+        self.gridLayout_48.addWidget(self.label_ddsp_svc_enable, 0, 0, 1, 1)
+        self.label_ddsp_svc_fSafePrefixPadLength = QtWidgets.QLabel(self.groupBox_ddsp_svc)
+        self.label_ddsp_svc_fSafePrefixPadLength.setObjectName("label_ddsp_svc_fSafePrefixPadLength")
+        self.gridLayout_48.addWidget(self.label_ddsp_svc_fSafePrefixPadLength, 3, 0, 1, 1)
+        self.lineEdit_ddsp_svc_sSpeakId = QtWidgets.QLineEdit(self.groupBox_ddsp_svc)
+        self.lineEdit_ddsp_svc_sSpeakId.setObjectName("lineEdit_ddsp_svc_sSpeakId")
+        self.gridLayout_48.addWidget(self.lineEdit_ddsp_svc_sSpeakId, 5, 1, 1, 1)
+        self.checkBox_ddsp_svc_enable = QtWidgets.QCheckBox(self.groupBox_ddsp_svc)
+        self.checkBox_ddsp_svc_enable.setObjectName("checkBox_ddsp_svc_enable")
+        self.gridLayout_48.addWidget(self.checkBox_ddsp_svc_enable, 0, 1, 1, 1)
+        self.lineEdit_ddsp_svc_api_ip_port = QtWidgets.QLineEdit(self.groupBox_ddsp_svc)
+        self.lineEdit_ddsp_svc_api_ip_port.setObjectName("lineEdit_ddsp_svc_api_ip_port")
+        self.gridLayout_48.addWidget(self.lineEdit_ddsp_svc_api_ip_port, 2, 1, 1, 1)
+        self.label_ddsp_svc_api_ip_port = QtWidgets.QLabel(self.groupBox_ddsp_svc)
+        self.label_ddsp_svc_api_ip_port.setObjectName("label_ddsp_svc_api_ip_port")
+        self.gridLayout_48.addWidget(self.label_ddsp_svc_api_ip_port, 2, 0, 1, 1)
+        self.label_ddsp_svc_config_path = QtWidgets.QLabel(self.groupBox_ddsp_svc)
+        self.label_ddsp_svc_config_path.setObjectName("label_ddsp_svc_config_path")
+        self.gridLayout_48.addWidget(self.label_ddsp_svc_config_path, 1, 0, 1, 1)
+        self.lineEdit_ddsp_svc_fSafePrefixPadLength = QtWidgets.QLineEdit(self.groupBox_ddsp_svc)
+        self.lineEdit_ddsp_svc_fSafePrefixPadLength.setObjectName("lineEdit_ddsp_svc_fSafePrefixPadLength")
+        self.gridLayout_48.addWidget(self.lineEdit_ddsp_svc_fSafePrefixPadLength, 3, 1, 1, 1)
+        self.lineEdit_ddsp_svc_fPitchChange = QtWidgets.QLineEdit(self.groupBox_ddsp_svc)
+        self.lineEdit_ddsp_svc_fPitchChange.setObjectName("lineEdit_ddsp_svc_fPitchChange")
+        self.gridLayout_48.addWidget(self.lineEdit_ddsp_svc_fPitchChange, 4, 1, 1, 1)
+        self.label_ddsp_svc_sSpeakId = QtWidgets.QLabel(self.groupBox_ddsp_svc)
+        self.label_ddsp_svc_sSpeakId.setObjectName("label_ddsp_svc_sSpeakId")
+        self.gridLayout_48.addWidget(self.label_ddsp_svc_sSpeakId, 5, 0, 1, 1)
+        self.lineEdit_ddsp_svc_config_path = QtWidgets.QLineEdit(self.groupBox_ddsp_svc)
+        self.lineEdit_ddsp_svc_config_path.setObjectName("lineEdit_ddsp_svc_config_path")
+        self.gridLayout_48.addWidget(self.lineEdit_ddsp_svc_config_path, 1, 1, 1, 1)
+        self.label_ddsp_svc_fPitchChange = QtWidgets.QLabel(self.groupBox_ddsp_svc)
+        self.label_ddsp_svc_fPitchChange.setObjectName("label_ddsp_svc_fPitchChange")
+        self.gridLayout_48.addWidget(self.label_ddsp_svc_fPitchChange, 4, 0, 1, 1)
+        self.label_ddsp_svc_sampleRate = QtWidgets.QLabel(self.groupBox_ddsp_svc)
+        self.label_ddsp_svc_sampleRate.setObjectName("label_ddsp_svc_sampleRate")
+        self.gridLayout_48.addWidget(self.label_ddsp_svc_sampleRate, 6, 0, 1, 1)
+        self.lineEdit_ddsp_svc_sampleRate = QtWidgets.QLineEdit(self.groupBox_ddsp_svc)
+        self.lineEdit_ddsp_svc_sampleRate.setObjectName("lineEdit_ddsp_svc_sampleRate")
+        self.gridLayout_48.addWidget(self.lineEdit_ddsp_svc_sampleRate, 6, 1, 1, 1)
+        self.gridLayout_49.addLayout(self.gridLayout_48, 0, 0, 1, 1)
+        self.verticalLayout.addWidget(self.groupBox_ddsp_svc)
         self.groupBox_so_vits_svc = QtWidgets.QGroupBox(self.scrollAreaWidgetContents)
         self.groupBox_so_vits_svc.setObjectName("groupBox_so_vits_svc")
         self.gridLayout_36 = QtWidgets.QGridLayout(self.groupBox_so_vits_svc)
@@ -1440,6 +1490,15 @@ def retranslateUi(self, MainWindow):
         self.label_choose_song_song_path.setText(_translate("MainWindow", "歌曲路径"))
         self.checkBox_choose_song_enable.setText(_translate("MainWindow", "是"))
         self.label_choose_song_match_fail_copy.setText(_translate("MainWindow", "匹配失败文案"))
+        self.groupBox_ddsp_svc.setTitle(_translate("MainWindow", "DDSP_SVC"))
+        self.label_ddsp_svc_enable.setText(_translate("MainWindow", "启用"))
+        self.label_ddsp_svc_fSafePrefixPadLength.setText(_translate("MainWindow", "安全前缀填充长度"))
+        self.checkBox_ddsp_svc_enable.setText(_translate("MainWindow", "是"))
+        self.label_ddsp_svc_api_ip_port.setText(_translate("MainWindow", "API地址"))
+        self.label_ddsp_svc_config_path.setText(_translate("MainWindow", "配置文件路径"))
+        self.label_ddsp_svc_sSpeakId.setText(_translate("MainWindow", "说话人ID"))
+        self.label_ddsp_svc_fPitchChange.setText(_translate("MainWindow", "变调"))
+        self.label_ddsp_svc_sampleRate.setText(_translate("MainWindow", "采样率"))
         self.groupBox_so_vits_svc.setTitle(_translate("MainWindow", "so-vits-svc"))
         self.checkBox_so_vits_svc_enable.setText(_translate("MainWindow", "是"))
         self.label_so_vits_svc_config_path.setText(_translate("MainWindow", "配置文件路径"))

config.json

@@ -67,9 +67,9 @@
     "enable": false,
     "config_path": "E:\\DDSP-SVC-3.0\\exp\\combsub-test\\config.yaml",
     "api_ip_port": "http://127.0.0.1:6844",
-    "fSafePrefixPadLength": 0,
-    "fPitchChange": 0,
-    "sSpeakId": 3,
+    "fSafePrefixPadLength": 0.0,
+    "fPitchChange": 0.0,
+    "sSpeakId": 0,
     "sampleRate": 44100
   },
   "so_vits_svc": {

config.json.bak

@@ -67,9 +67,9 @@
     "enable": false,
     "config_path": "E:\\DDSP-SVC-3.0\\exp\\combsub-test\\config.yaml",
     "api_ip_port": "http://127.0.0.1:6844",
-    "fSafePrefixPadLength": 0,
-    "fPitchChange": 0,
-    "sSpeakId": 3,
+    "fSafePrefixPadLength": 0.0,
+    "fPitchChange": 0.0,
+    "sSpeakId": 0,
     "sampleRate": 44100
   },
   "so_vits_svc": {

main.py

@@ -341,6 +341,15 @@ def init_config(self):
             self.ui.label_choose_song_song_path.setToolTip("歌曲音频路径（默认为本项目的song文件夹）")
             self.ui.label_choose_song_match_fail_copy.setToolTip("匹配失败返回的音频文案 注意 {content} 这个是用于替换用户发送的歌名的，请务必不要乱删！影响使用！")
 
+            # DDSP-SVC
+            self.ui.label_ddsp_svc_enable.setToolTip("是否启用DDSP-SVC进行音频的变声")
+            self.ui.label_ddsp_svc_config_path.setToolTip("模型配置文件config.yaml的路径(此处可以不配置，暂时没有用到)")
+            self.ui.label_ddsp_svc_api_ip_port.setToolTip("flask_api服务运行的ip端口，例如：http://127.0.0.1:6844")
+            self.ui.label_ddsp_svc_fSafePrefixPadLength.setToolTip("安全前缀填充长度，不知道干啥用，默认为0")
+            self.ui.label_ddsp_svc_fPitchChange.setToolTip("音调设置，默认为0")
+            self.ui.label_ddsp_svc_sSpeakId.setToolTip("说话人ID，需要和模型数据对应，默认为0")
+            self.ui.label_ddsp_svc_sampleRate.setToolTip("DAW所需的采样率，默认为44100")
+
             # so-vits-svc
             self.ui.label_so_vits_svc_enable.setToolTip("是否启用so-vits-svc进行音频的变声")
             self.ui.label_so_vits_svc_config_path.setToolTip("模型配置文件config.json的路径")
@@ -648,6 +657,16 @@ def init_config(self):
             self.ui.lineEdit_choose_song_song_path.setText(self.choose_song_config['song_path'])
             self.ui.lineEdit_choose_song_match_fail_copy.setText(self.choose_song_config['match_fail_copy'])
 
+            # ddsp-svc
+            if config.get("ddsp_svc", "enable"):
+                self.ui.checkBox_ddsp_svc_enable.setChecked(True)
+            self.ui.lineEdit_ddsp_svc_config_path.setText(config.get("ddsp_svc", "config_path"))
+            self.ui.lineEdit_ddsp_svc_api_ip_port.setText(config.get("ddsp_svc", "api_ip_port"))
+            self.ui.lineEdit_ddsp_svc_fSafePrefixPadLength.setText(str(config.get("ddsp_svc", "fSafePrefixPadLength")))
+            self.ui.lineEdit_ddsp_svc_fPitchChange.setText(str(config.get("ddsp_svc", "fPitchChange")))
+            self.ui.lineEdit_ddsp_svc_sSpeakId.setText(str(config.get("ddsp_svc", "sSpeakId")))
+            self.ui.lineEdit_ddsp_svc_sampleRate.setText(str(config.get("ddsp_svc", "sampleRate")))
+
             if self.so_vits_svc_config['enable']:
                 self.ui.checkBox_so_vits_svc_enable.setChecked(True)
             self.ui.lineEdit_so_vits_svc_config_path.setText(self.so_vits_svc_config['config_path'])
@@ -1084,6 +1103,16 @@ def save(self):
             config_data["choose_song"]["song_path"] = self.ui.lineEdit_choose_song_song_path.text()
             config_data["choose_song"]["match_fail_copy"] = self.ui.lineEdit_choose_song_match_fail_copy.text()
 
+            # DDSP-SVC
+            config_data["ddsp_svc"]["enable"] = self.ui.checkBox_ddsp_svc_enable.isChecked()
+            config_data["ddsp_svc"]["config_path"] = self.ui.lineEdit_ddsp_svc_config_path.text()
+            config_data["ddsp_svc"]["api_ip_port"] = self.ui.lineEdit_ddsp_svc_api_ip_port.text()
+            config_data["ddsp_svc"]["fSafePrefixPadLength"] = round(float(self.ui.lineEdit_ddsp_svc_fSafePrefixPadLength.text()), 1)
+            config_data["ddsp_svc"]["fPitchChange"] = round(float(self.ui.lineEdit_ddsp_svc_fPitchChange.text()), 1)
+            config_data["ddsp_svc"]["sSpeakId"] = int(self.ui.lineEdit_ddsp_svc_sSpeakId.text())
+            config_data["ddsp_svc"]["sampleRate"] = int(self.ui.lineEdit_ddsp_svc_sampleRate.text())
+
+            # so-vits-svc
             config_data["so_vits_svc"]["enable"] = self.ui.checkBox_so_vits_svc_enable.isChecked()
             config_data["so_vits_svc"]["config_path"] = self.ui.lineEdit_so_vits_svc_config_path.text()
             config_data["so_vits_svc"]["api_ip_port"] = self.ui.lineEdit_so_vits_svc_api_ip_port.text()

ui/main.ui

@@ -134,9 +134,9 @@ font: 75 12pt &quot;微软雅黑&quot;;</string>
            <property name="geometry">
             <rect>
              <x>0</x>
-             <y>0</y>
+             <y>-4345</y>
              <width>980</width>
-             <height>5794</height>
+             <height>6162</height>
             </rect>
            </property>
            <layout class="QVBoxLayout" name="verticalLayout">
@@ -1448,6 +1448,93 @@ font: 75 12pt &quot;微软雅黑&quot;;</string>
               </layout>
              </widget>
             </item>
+            <item>
+             <widget class="QGroupBox" name="groupBox_ddsp_svc">
+              <property name="title">
+               <string>DDSP_SVC</string>
+              </property>
+              <layout class="QGridLayout" name="gridLayout_49">
+               <item row="0" column="0">
+                <layout class="QGridLayout" name="gridLayout_48">
+                 <item row="0" column="0">
+                  <widget class="QLabel" name="label_ddsp_svc_enable">
+                   <property name="text">
+                    <string>启用</string>
+                   </property>
+                  </widget>
+                 </item>
+                 <item row="3" column="0">
+                  <widget class="QLabel" name="label_ddsp_svc_fSafePrefixPadLength">
+                   <property name="text">
+                    <string>安全前缀填充长度</string>
+                   </property>
+                  </widget>
+                 </item>
+                 <item row="5" column="1">
+                  <widget class="QLineEdit" name="lineEdit_ddsp_svc_sSpeakId"/>
+                 </item>
+                 <item row="0" column="1">
+                  <widget class="QCheckBox" name="checkBox_ddsp_svc_enable">
+                   <property name="text">
+                    <string>是</string>
+                   </property>
+                  </widget>
+                 </item>
+                 <item row="2" column="1">
+                  <widget class="QLineEdit" name="lineEdit_ddsp_svc_api_ip_port"/>
+                 </item>
+                 <item row="2" column="0">
+                  <widget class="QLabel" name="label_ddsp_svc_api_ip_port">
+                   <property name="text">
+                    <string>API地址</string>
+                   </property>
+                  </widget>
+                 </item>
+                 <item row="1" column="0">
+                  <widget class="QLabel" name="label_ddsp_svc_config_path">
+                   <property name="text">
+                    <string>配置文件路径</string>
+                   </property>
+                  </widget>
+                 </item>
+                 <item row="3" column="1">
+                  <widget class="QLineEdit" name="lineEdit_ddsp_svc_fSafePrefixPadLength"/>
+                 </item>
+                 <item row="4" column="1">
+                  <widget class="QLineEdit" name="lineEdit_ddsp_svc_fPitchChange"/>
+                 </item>
+                 <item row="5" column="0">
+                  <widget class="QLabel" name="label_ddsp_svc_sSpeakId">
+                   <property name="text">
+                    <string>说话人ID</string>
+                   </property>
+                  </widget>
+                 </item>
+                 <item row="1" column="1">
+                  <widget class="QLineEdit" name="lineEdit_ddsp_svc_config_path"/>
+                 </item>
+                 <item row="4" column="0">
+                  <widget class="QLabel" name="label_ddsp_svc_fPitchChange">
+                   <property name="text">
+                    <string>变调</string>
+                   </property>
+                  </widget>
+                 </item>
+                 <item row="6" column="0">
+                  <widget class="QLabel" name="label_ddsp_svc_sampleRate">
+                   <property name="text">
+                    <string>采样率</string>
+                   </property>
+                  </widget>
+                 </item>
+                 <item row="6" column="1">
+                  <widget class="QLineEdit" name="lineEdit_ddsp_svc_sampleRate"/>
+                 </item>
+                </layout>
+               </item>
+              </layout>
+             </widget>
+            </item>
             <item>
              <widget class="QGroupBox" name="groupBox_so_vits_svc">
               <property name="title">