tts models support (#2033)

Co-authored-by: luowei <glpat-EjySCyNjWiLqAED-YmwM>
Co-authored-by: crazywoola <427733928@qq.com>
Co-authored-by: crazywoola <100913391+crazywoola@users.noreply.github.com>
Co-authored-by: Yeuoly <45712896+Yeuoly@users.noreply.github.com>

api/services/audio_service.py

@@ -1,22 +1,26 @@
 import io
+from typing import Optional
 
 from core.model_manager import ModelManager
 from core.model_runtime.entities.model_entities import ModelType
-from services.errors.audio import (AudioTooLargeServiceError, NoAudioUploadedServiceError,
-                                   ProviderNotSupportSpeechToTextServiceError, UnsupportedAudioTypeServiceError)
+from services.errors.audio import (AudioTooLargeServiceError,
+                                   NoAudioUploadedServiceError,
+                                   ProviderNotSupportTextToSpeechServiceError,
+                                   ProviderNotSupportSpeechToTextServiceError,
+                                   UnsupportedAudioTypeServiceError)
 from werkzeug.datastructures import FileStorage
 
 FILE_SIZE = 15
 FILE_SIZE_LIMIT = FILE_SIZE * 1024 * 1024
-ALLOWED_EXTENSIONS = ['mp3', 'mp4', 'mpeg', 'mpga', 'm4a', 'wav', 'webm']
+ALLOWED_EXTENSIONS = ['mp3', 'mp4', 'mpeg', 'mpga', 'm4a', 'wav', 'webm', 'amr']
 
 
 class AudioService:
     @classmethod
-    def transcript(cls, tenant_id: str, file: FileStorage):
+    def transcript_asr(cls, tenant_id: str, file: FileStorage, end_user: Optional[str] = None):
         if file is None:
             raise NoAudioUploadedServiceError()
-        
+
         extension = file.mimetype
         if extension not in [f'audio/{ext}' for ext in ALLOWED_EXTENSIONS]:
             raise UnsupportedAudioTypeServiceError()
@@ -33,8 +37,26 @@ class AudioService:
             tenant_id=tenant_id,
             model_type=ModelType.SPEECH2TEXT
         )
+        if model_instance is None:
+            raise ProviderNotSupportSpeechToTextServiceError()
 
         buffer = io.BytesIO(file_content)
         buffer.name = 'temp.mp3'
 
-        return {"text": model_instance.invoke_speech2text(buffer)}
+        return {"text": model_instance.invoke_speech2text(file=buffer, user=end_user)}
+
+    @classmethod
+    def transcript_tts(cls, tenant_id: str, text: str, streaming: bool, end_user: Optional[str] = None):
+        model_manager = ModelManager()
+        model_instance = model_manager.get_default_model_instance(
+            tenant_id=tenant_id,
+            model_type=ModelType.TTS
+        )
+        if model_instance is None:
+            raise ProviderNotSupportTextToSpeechServiceError()
+
+        try:
+            audio_response = model_instance.invoke_tts(content_text=text.strip(), user=end_user, streaming=streaming)
+            return audio_response
+        except Exception as e:
+            raise e