fix

api/services/app_model_config_service.py

@@ -4,6 +4,7 @@ import uuid
 from core.constant import llm_constant
 from models.account import Account
 from services.dataset_service import DatasetService
+from core.llm.llm_builder import LLMBuilder
 
 
 class AppModelConfigService:
@@ -123,6 +124,11 @@ class AppModelConfigService:
 
         if not isinstance(config["speech_to_text"]["enabled"], bool):
             raise ValueError("enabled in speech_to_text must be of boolean type")
+        
+        provider_name = LLMBuilder.get_default_provider(account.current_tenant_id)
+
+        if config["speech_to_text"]["enabled"] and provider_name != 'openai':
+            raise ValueError("provider not support speech to text")
 
         # more_like_this
         if 'more_like_this' not in config or not config["more_like_this"]:

api/services/audio_service.py

@@ -4,6 +4,7 @@ from werkzeug.datastructures import FileStorage
 from core.llm.llm_builder import LLMBuilder
 from core.llm.provider.llm_provider_service import LLMProviderService
 from services.errors.audio import NoAudioUploadedError, AudioTooLargeError, UnsupportedAudioTypeError
+from openai.error import InvalidRequestError
 
 FILE_SIZE_LIMIT = 1 * 1024 * 1024
 ALLOWED_EXTENSIONS = ['mp3', 'mp4', 'mpeg', 'mpga', 'm4a', 'wav', 'webm']
@@ -32,7 +33,8 @@ class AudioService:
         buffer = io.BytesIO(file_content)
         buffer.name = 'temp.wav'
 
-        transcript = openai.Audio.transcribe(
+        try:
+            transcript = openai.Audio.transcribe(
                         model='whisper-1', 
                         file=buffer,
                         api_key=credentials.get('openai_api_key'),
@@ -42,4 +44,11 @@ class AudioService:
                         params=params
                     )
         
-        return transcript
\ No newline at end of file
+            return transcript
+        except InvalidRequestError as e:
+            return {'message': str(e)}, 400
+    
+
+
+        
+        
\ No newline at end of file

web/app/components/app/configuration/config/feature/choose-feature/index.tsx

@@ -21,6 +21,7 @@ export type IChooseFeatureProps = {
   config: IConfig
   isChatApp: boolean
   onChange: (key: string, value: boolean) => void
+  showSpeechToTextItem?: boolean
 }
 
 const OpeningStatementIcon = (
@@ -35,6 +36,7 @@ const ChooseFeature: FC<IChooseFeatureProps> = ({
   isChatApp,
   config,
   onChange,
+  showSpeechToTextItem,
 }) => {
   const { t } = useTranslation()
 
@@ -71,14 +73,18 @@ const ChooseFeature: FC<IChooseFeatureProps> = ({
                 value={config.suggestedQuestionsAfterAnswer}
                 onChange={value => onChange('suggestedQuestionsAfterAnswer', value)}
               />
-              <FeatureItem
-                icon={<Microphone01 className='w-4 h-4 text-[#7839EE]' />}
-                previewImgClassName='speechToTextPreview'
-                title={t('appDebug.feature.speechToText.title')}
-                description={t('appDebug.feature.speechToText.description')}
-                value={config.speechToText}
-                onChange={value => onChange('speechToText', value)}
-              />
+              {
+                showSpeechToTextItem && (
+                  <FeatureItem
+                    icon={<Microphone01 className='w-4 h-4 text-[#7839EE]' />}
+                    previewImgClassName='speechToTextPreview'
+                    title={t('appDebug.feature.speechToText.title')}
+                    description={t('appDebug.feature.speechToText.description')}
+                    value={config.speechToText}
+                    onChange={value => onChange('speechToText', value)}
+                  />
+                )
+              }
             </>
           </FeatureGroup>
         )}

web/app/components/app/configuration/config/index.tsx

@@ -4,6 +4,7 @@ import React from 'react'
 import { useContext } from 'use-context-selector'
 import produce from 'immer'
 import { useBoolean } from 'ahooks'
+import useSWR from 'swr'
 import DatasetConfig from '../dataset-config'
 import ChatGroup from '../features/chat-group'
 import ExperienceEnchanceGroup from '../features/experience-enchance-group'
@@ -19,6 +20,7 @@ import ConfigPrompt from '@/app/components/app/configuration/config-prompt'
 import ConfigVar from '@/app/components/app/configuration/config-var'
 import type { PromptVariable } from '@/models/debug'
 import { AppType } from '@/types/app'
+import { fetchTenantInfo } from '@/service/common'
 
 const Config: FC = () => {
   const {
@@ -37,6 +39,8 @@ const Config: FC = () => {
     setSpeechToTextConfig,
   } = useContext(ConfigContext)
   const isChatApp = mode === AppType.chat
+  const { data: userInfo } = useSWR({ url: '/info' }, fetchTenantInfo)
+  const targetProvider = userInfo?.providers?.find(({ token_is_set, is_valid }) => token_is_set && is_valid)
 
   const promptTemplate = modelConfig.configs.prompt_template
   const promptVariables = modelConfig.configs.prompt_variables
@@ -88,7 +92,7 @@ const Config: FC = () => {
     },
   })
 
-  const hasChatConfig = isChatApp && (featureConfig.openingStatement || featureConfig.suggestedQuestionsAfterAnswer || featureConfig.speechToText)
+  const hasChatConfig = isChatApp && (featureConfig.openingStatement || featureConfig.suggestedQuestionsAfterAnswer || (featureConfig.speechToText && targetProvider?.provider_name === 'openai'))
   const hasToolbox = false
 
   const [showAutomatic, { setTrue: showAutomaticTrue, setFalse: showAutomaticFalse }] = useBoolean(false)
@@ -118,6 +122,7 @@ const Config: FC = () => {
             isChatApp={isChatApp}
             config={featureConfig}
             onChange={handleFeatureChange}
+            showSpeechToTextItem={targetProvider?.provider_name === 'openai'}
           />
         )}
         {showAutomatic && (