config main

main.py

@@ -48,9 +48,7 @@ def get_fragment_list(result):
 @app.route('/predict', methods=['POST'])
 @cross_origin()  
 def inference():
-    audio = request.files['audio']
-    audio.save('audio.wav')
-    # audio = bytearray(audio)
+    audio = request.files['audio'].read()
     openai_api_key = request.form['openai_api_key']
     print(audio, request.form['openai_api_key'])
     
@@ -59,7 +57,7 @@ def inference():
     if openai_api_key:
         chat = ChatOpenAI(temperature=0, openai_api_key=openai_api_key.strip())
         batch_size = 16 # reduce if low on GPU mem
-        audio = whisperx.load_audio("./audio.wav")
+        audio = whisperx.load_audio(io.BytesIO(audio))
         result = model.transcribe(audio, batch_size=batch_size)
 
         text = get_all_text(result)