[feature]Improve each api.

server.py

@@ -1,13 +1,17 @@
+import base64
+import io
+import logging
 import os
 import typing
 from contextlib import asynccontextmanager
 
 import uvicorn
-from fastapi import FastAPI, Request, UploadFile, File
+from fastapi import FastAPI, Request, UploadFile, File, WebSocket
 from fastapi.middleware.cors import CORSMiddleware
 from fastapi.middleware.gzip import GZipMiddleware
 from faster_whisper import WhisperModel
-from pydantic import BaseModel, Field, FilePath
+from pydantic import BaseModel, Field, ValidationError
+from starlette.websockets import WebSocketState
 
 
 @asynccontextmanager
@@ -52,36 +56,100 @@ whisper_model: typing.Optional[WhisperModel] = WhisperModel(model_size, device='
 
 class TranscribeRequestParams(BaseModel):
     uuid: str = Field(title='Request Unique Id.')
-    audio_file: FilePath = Field()
+    audio_file: str
     language: typing.Literal['en', 'zh',]
 
 
 @app.post('/transcribe')
-def transcribe_api(
+async def transcribe_api(
         request: Request,
         obj: TranscribeRequestParams
 ):
-    transcribed_text = whisper_model.transcribe(obj.audio_file, language=obj.language)
-    return {
-        "if_success": True,
-        'uuid': obj.uuid,
-        'transcribed_text': transcribed_text
-    }
+    try:
+        audio_file = io.BytesIO(base64.b64decode(obj.audio_file))
+
+        transcribed_text = whisper_model.transcribe(audio_file, language=obj.language)
+    except Exception as exc:
+        logging.exception(exc)
+        response_body = {
+            "if_success": False,
+            'uuid': obj.uuid,
+            'msg': f'{exc}'
+        }
+    else:
+        response_body = {
+            "if_success": True,
+            'uuid': obj.uuid,
+            'transcribed_text': transcribed_text
+        }
+    return response_body
 
 
 @app.post('/transcribe-file')
-def transcribe_file_api(
+async def transcribe_file_api(
         request: Request,
         uuid: str,
         audio_file: typing.Annotated[UploadFile, File()],
         language: typing.Literal['en', 'zh']
 ):
-    transcribed_text = whisper_model.transcribe(audio_file.file, language=language)
-    return {
-        "if_success": True,
-        'uuid': uuid,
-        'transcribed_text': transcribed_text
-    }
+    try:
+        transcribed_text = whisper_model.transcribe(audio_file.file, language=language)
+    except Exception as exc:
+        logging.exception(exc)
+        response_body = {
+            "if_success": False,
+            'uuid': uuid,
+            'msg': f'{exc}'
+        }
+    else:
+        response_body = {
+            "if_success": True,
+            'uuid': uuid,
+            'transcribed_text': transcribed_text
+        }
+
+    return response_body
+
+
+@app.websocket('/transcribe')
+async def transcribe_ws_api(
+        websocket: WebSocket
+):
+    await websocket.accept()
+
+    while websocket.client_state == WebSocketState.CONNECTED:
+        request_params = await websocket.receive_json()
+
+        try:
+            form = TranscribeRequestParams.model_validate(request_params)
+        except ValidationError as exc:
+            logging.exception(exc)
+            await websocket.send_json({
+                "if_success": False,
+                'uuid': request_params.get('uuid', ''),
+                'msg': f'{exc}'
+            })
+            continue
+
+        try:
+            audio_file = io.BytesIO(base64.b64decode(form.audio_file))
+
+            transcribed_text = whisper_model.transcribe(audio_file, language=form.language)
+        except Exception as exc:
+            logging.exception(exc)
+            response_body = {
+                "if_success": False,
+                'uuid': form.uuid,
+                'msg': f'{exc}'
+            }
+        else:
+            response_body = {
+                "if_success": True,
+                'uuid': form.uuid,
+                'transcribed_text': transcribed_text
+            }
+
+        await websocket.send_json(response_body)
 
 
 if __name__ == '__main__':