whisper model update + minor fixes

.gitignore

@@ -284,4 +284,4 @@ rolegenie/media/
 templates.py
 newrelic.ini
 __pycache__/
-audios/*
+audios/

requirements.txt

@@ -34,7 +34,7 @@ nvidia-cusolver-cu11==11.4.0.1
 nvidia-cusparse-cu11==11.7.4.91
 nvidia-nccl-cu11==2.14.3
 nvidia-nvtx-cu11==11.7.91
-openai-whisper==20230314
+openai-whisper==20231117
 python-dotenv==1.0.0
 regex==2023.6.3
 requests==2.31.0

routes.py

@@ -1,3 +1,4 @@
+import os
 import uuid
 import json
 
@@ -29,8 +30,10 @@ async def transcribe(request):
             status_code=400,
             content={"status": 400, "message": "Bad Request", "data": {}},
         )
-
-    file_path = f"{str(ROOT_DIR)}/audios/{str(uuid.uuid4())}.mp3"
+    audio_directory = f"{str(ROOT_DIR)}/audios"
+    if not os.path.exists(audio_directory):
+        os.makedirs(audio_directory)
+    file_path = f"{audio_directory}/{str(uuid.uuid4())}.mp3"
     with open(file_path, "wb") as f:
         f.write(audio.file.read())
 

routes_definition/transcribe/transcribe.py

@@ -1,8 +1,10 @@
+import os
 import whisper
 
 
 def transcribe_helper(model: whisper.Whisper, audio_path: str, language: str):
     result = model.transcribe(audio_path, language=language)
+    os.remove(audio_path)
 
     print(result)
     return result["text"]