refactor: 配置TTS参数支持试听功能

apps/application/serializers/application_serializers.py

@@ -973,6 +973,16 @@ class ApplicationSerializer(serializers.Serializer):
                                                             **application.tts_model_params_setting)
                 return model.text_to_speech(text)
 
+        def play_demo_text(self, form_data, with_valid=True):
+            text = '你好，这里是语音播放测试'
+            if with_valid:
+                self.is_valid(raise_exception=True)
+            application_id = self.data.get('application_id')
+            application = QuerySet(Application).filter(id=application_id).first()
+            if application.tts_model_enable:
+                model = get_model_instance_by_model_user_id(application.tts_model_id, application.user_id, **form_data)
+                return model.text_to_speech(text)
+
     class ApplicationKeySerializerModel(serializers.ModelSerializer):
         class Meta:
             model = ApplicationApiKey

apps/application/urls.py

@@ -70,5 +70,7 @@ urlpatterns = [
          name='application/audio'),
     path('application/<str:application_id>/text_to_speech', views.Application.TextToSpeech.as_view(),
          name='application/audio'),
+    path('application/<str:application_id>/play_demo_text', views.Application.PlayDemoText.as_view(),
+         name='application/audio'),
 
 ]

apps/application/views/application_views.py

@@ -566,3 +566,19 @@ class Application(APIView):
                 request.data.get('text'))
             return HttpResponse(byte_data, status=200, headers={'Content-Type': 'audio/mp3',
                                                               'Content-Disposition': 'attachment; filename="abc.mp3"'})
+
+    class PlayDemoText(APIView):
+        authentication_classes = [TokenAuth]
+
+        @action(methods=['POST'], detail=False)
+        @has_permissions(ViewPermission([RoleConstants.ADMIN, RoleConstants.USER, RoleConstants.APPLICATION_ACCESS_TOKEN],
+                                        [lambda r, keywords: Permission(group=Group.APPLICATION,
+                                                                        operate=Operate.USE,
+                                                                        dynamic_tag=keywords.get(
+                                                                            'application_id'))],
+                                        compare=CompareConstants.AND))
+        def post(self, request: Request, application_id: str):
+            byte_data = ApplicationSerializer.Operate(
+                data={'application_id': application_id, 'user_id': request.user.id}).play_demo_text(request.data)
+            return HttpResponse(byte_data, status=200, headers={'Content-Type': 'audio/mp3',
+                                                              'Content-Disposition': 'attachment; filename="abc.mp3"'})

apps/setting/models_provider/impl/aliyun_bai_lian_model_provider/model/tts.py

@@ -44,3 +44,6 @@ class AliyunBaiLianTextToSpeech(MaxKBBaseModel, BaseTextToSpeech):
             print(audio)
             raise Exception(audio)
         return audio
+
+    def is_cache_model(self):
+        return False

apps/setting/models_provider/impl/openai_model_provider/model/tts.py

@@ -56,3 +56,6 @@ class OpenAITextToSpeech(MaxKBBaseModel, BaseTextToSpeech):
                 input=text,
         ) as response:
             return response.read()
+
+    def is_cache_model(self):
+        return False

apps/setting/models_provider/impl/volcanic_engine_model_provider/model/tts.py

@@ -102,6 +102,9 @@ class VolcanicEngineTextToSpeech(MaxKBBaseModel, BaseTextToSpeech):
 
         return asyncio.run(self.submit(request_json, text))
 
+    def is_cache_model(self):
+        return False
+
     def token_auth(self):
         return {'Authorization': 'Bearer; {}'.format(self.volcanic_token)}
 

apps/setting/models_provider/impl/xf_model_provider/model/tts.py

@@ -113,6 +113,9 @@ class XFSparkTextToSpeech(MaxKBBaseModel, BaseTextToSpeech):
 
         return asyncio.run(handle())
 
+    def is_cache_model(self):
+        return False
+
     @staticmethod
     async def handle_message(ws):
         audio_bytes: bytes = b''

apps/setting/models_provider/impl/xinference_model_provider/model/tts.py

@@ -58,3 +58,6 @@ class XInferenceTextToSpeech(MaxKBBaseModel, BaseTextToSpeech):
                 input=text,
         ) as response:
             return response.read()
+
+    def is_cache_model(self):
+        return False

ui/src/api/application.ts

@@ -344,7 +344,7 @@ const postSpeechToText: (
 }
 
 /**
- * 语音转文本
+ * 文本转语音
  */
 const postTextToSpeech: (
   application_id: String,
@@ -353,6 +353,17 @@ const postTextToSpeech: (
 ) => Promise<Result<any>> = (application_id, data, loading) => {
   return download(`${prefix}/${application_id}/text_to_speech`, 'post', data, undefined, loading)
 }
+
+/**
+ * 播放测试文本
+ */
+const playDemoText: (
+  application_id: String,
+  data: any,
+  loading?: Ref<boolean>
+) => Promise<Result<any>> = (application_id, data, loading) => {
+  return download(`${prefix}/${application_id}/play_demo_text`, 'post', data, undefined, loading)
+}
 /**
  * 获取平台状态
  */
@@ -430,5 +441,6 @@ export default {
   getPlatformConfig,
   updatePlatformConfig,
   updatePlatformStatus,
-  validatePassword
+  validatePassword,
+  playDemoText
 }

ui/src/views/application/ApplicationSetting.vue

@@ -533,7 +533,7 @@
     </el-row>
 
     <AIModeParamSettingDialog ref="AIModeParamSettingDialogRef" @refresh="refreshForm" />
-    <AIModeParamSettingDialog ref="TTSModeParamSettingDialogRef" @refresh="refreshTTSForm" />
+    <TTSModeParamSettingDialog ref="TTSModeParamSettingDialogRef" @refresh="refreshTTSForm" />
     <ParamSettingDialog ref="ParamSettingDialogRef" @refresh="refreshParam" />
     <AddDatasetDialog
       ref="AddDatasetDialogRef"
@@ -573,6 +573,7 @@ import { relatedObject } from '@/utils/utils'
 import { MsgSuccess, MsgWarning } from '@/utils/message'
 import useStore from '@/stores'
 import { t } from '@/locales'
+import TTSModeParamSettingDialog from './component/TTSModeParamSettingDialog.vue'
 
 const { model, application } = useStore()
 
@@ -587,7 +588,7 @@ const defaultPrompt = t('views.application.prompt.defaultPrompt', {
 })
 
 const AIModeParamSettingDialogRef = ref<InstanceType<typeof AIModeParamSettingDialog>>()
-const TTSModeParamSettingDialogRef = ref<InstanceType<typeof AIModeParamSettingDialog>>()
+const TTSModeParamSettingDialogRef = ref<InstanceType<typeof TTSModeParamSettingDialog>>()
 const ParamSettingDialogRef = ref<InstanceType<typeof ParamSettingDialog>>()
 const createModelRef = ref<InstanceType<typeof CreateModelDialog>>()
 const selectProviderRef = ref<InstanceType<typeof SelectProviderDialog>>()

ui/src/views/application/component/TTSModeParamSettingDialog.vue

@@ -0,0 +1,156 @@
+<template>
+  <el-dialog
+    align-center
+    :title="$t('views.application.applicationForm.dialogues.paramSettings')"
+    class="aiMode-param-dialog"
+    v-model="dialogVisible"
+    style="width: 550px"
+    append-to-body
+    :close-on-click-modal="false"
+    :close-on-press-escape="false"
+  >
+    <DynamicsForm
+      v-model="form_data"
+      :model="form_data"
+      label-position="top"
+      require-asterisk-position="right"
+      :render_data="model_form_field"
+      ref="dynamicsFormRef"
+    >
+    </DynamicsForm>
+
+    <template #footer>
+      <div class="flex-between">
+        <span class="p-16">
+          <el-button @click="testPlay" :loading="playLoading">
+            <AppIcon iconName="app-video-play"></AppIcon>
+            试听
+          </el-button>
+        </span>
+        <span class="dialog-footer p-16">
+        <el-button @click.prevent="dialogVisible = false">
+          {{ $t('views.application.applicationForm.buttons.cancel') }}
+        </el-button>
+        <el-button type="primary" @click="submit" :loading="loading">
+          {{ $t('views.application.applicationForm.buttons.confirm') }}
+        </el-button>
+      </span>
+      </div>
+    </template>
+  </el-dialog>
+  <!-- 先渲染，不然不能播放   -->
+  <audio ref="audioPlayer" controls hidden="hidden"></audio>
+</template>
+
+<script setup lang="ts">
+import { ref } from 'vue'
+import type { FormField } from '@/components/dynamics-form/type'
+import modelAPi from '@/api/model'
+import applicationApi from '@/api/application'
+import DynamicsForm from '@/components/dynamics-form/index.vue'
+import { keys } from 'lodash'
+import { app } from '@/main'
+
+const {
+  params: { id }
+} = app.config.globalProperties.$route as any
+
+const model_form_field = ref<Array<FormField>>([])
+const emit = defineEmits(['refresh'])
+const dynamicsFormRef = ref<InstanceType<typeof DynamicsForm>>()
+const form_data = ref<any>({})
+const dialogVisible = ref(false)
+const loading = ref(false)
+const playLoading = ref(false)
+const getApi = (model_id: string, application_id?: string) => {
+  return application_id
+    ? applicationApi.getModelParamsForm(application_id, model_id, loading)
+    : modelAPi.getModelParamsForm(model_id, loading)
+}
+const open = (model_id: string, application_id?: string, model_setting_data?: any) => {
+  form_data.value = {}
+  const api = getApi(model_id, application_id)
+  api.then((ok) => {
+    model_form_field.value = ok.data
+    model_setting_data =
+      model_setting_data && keys(model_setting_data).length > 0
+        ? model_setting_data
+        : ok.data
+          .map((item: any) => ({ [item.field]: item.default_value }))
+          .reduce((x, y) => ({ ...x, ...y }), {})
+    // 渲染动态表单
+    dynamicsFormRef.value?.render(model_form_field.value, model_setting_data)
+  })
+  dialogVisible.value = true
+}
+
+const reset_default = (model_id: string, application_id?: string) => {
+  const api = getApi(model_id, application_id)
+  api.then((ok) => {
+    model_form_field.value = ok.data
+    const model_setting_data = ok.data
+      .map((item) => ({ [item.field]: item.default_value }))
+      .reduce((x, y) => ({ ...x, ...y }), {})
+
+    emit('refresh', model_setting_data)
+  })
+}
+
+const submit = async () => {
+  emit('refresh', form_data.value)
+  dialogVisible.value = false
+}
+
+
+const audioPlayer = ref<HTMLAudioElement | null>(null)
+const testPlay = () => {
+  applicationApi
+    .playDemoText(id as string, form_data.value, playLoading)
+    .then((res: any) => {
+      // 创建 Blob 对象
+      const blob = new Blob([res], { type: 'audio/mp3' })
+
+      // 创建对象 URL
+      const url = URL.createObjectURL(blob)
+
+      // 检查 audioPlayer 是否已经引用了 DOM 元素
+      if (audioPlayer.value instanceof HTMLAudioElement) {
+        audioPlayer.value.src = url
+        audioPlayer.value.play() // 自动播放音频
+      } else {
+        console.error('audioPlayer.value is not an instance of HTMLAudioElement')
+      }
+    })
+    .catch((err) => {
+      console.log('err: ', err)
+    })
+
+}
+
+
+defineExpose({ open, reset_default })
+</script>
+
+<style lang="scss" scoped>
+.aiMode-param-dialog {
+  padding: 8px 8px 24px 8px;
+
+  .el-dialog__header {
+    padding: 16px 16px 0 16px;
+  }
+
+  .el-dialog__body {
+    padding: 16px !important;
+  }
+
+  .dialog-max-height {
+    height: 550px;
+  }
+
+  .custom-slider {
+    .el-input-number.is-without-controls .el-input__wrapper {
+      padding: 0 !important;
+    }
+  }
+}
+</style>

ui/src/workflow/nodes/base-node/index.vue

@@ -258,7 +258,7 @@
 
     <FieldFormDialog ref="FieldFormDialogRef" @refresh="refreshFieldList" />
   </NodeContainer>
-  <AIModeParamSettingDialog ref="TTSModeParamSettingDialogRef" @refresh="refreshTTSForm" />
+  <TTSModeParamSettingDialog ref="TTSModeParamSettingDialogRef" @refresh="refreshTTSForm" />
 </template>
 <script setup lang="ts">
 import { app } from '@/main'
@@ -273,7 +273,7 @@ import type { Provider } from '@/api/type/model'
 import FieldFormDialog from './component/FieldFormDialog.vue'
 import { MsgError, MsgSuccess, MsgWarning } from '@/utils/message'
 import { t } from '@/locales'
-import AIModeParamSettingDialog from '@/views/application/component/AIModeParamSettingDialog.vue'
+import TTSModeParamSettingDialog from '@/views/application/component/TTSModeParamSettingDialog.vue'
 const { model } = useStore()
 
 const {