Update app.py

app.py

@@ -48,8 +48,7 @@ LOGGER.setLevel(log_level)
 
 class DemandForecasting:
     def __init__(self):
-        self.client = AzureOpenAI()
-        self.whisper_model = whisper.load_model("medium.en")
+        self.client = OpenAI()
 
 
     def get_column(self,train_csv_path: str):
@@ -319,44 +318,46 @@ class DemandForecasting:
             return None
 
     def audio_to_text(self, audio_path):
-        """
-        transcribe the audio to text.
-        """
-
-
-        result = self.whisper_model.transcribe(audio_path)
-        print("audio_to_text",result["text"])
-        return result["text"]
+        audio_file= open(audio_path, "rb")
+        transcription = self.client.audio.transcriptions.create(
+        model="whisper-1", 
+        file=audio_file)
+        print(transcription.text)
+        return transcription.text
 
 
     def parse_text(self, text, column_list):
 
         # Define the prompt or input for the model
         conversation =[{"role": "system", "content": ""},
-        {"role": "user", "content":f""" extract the {column_list}. al
-        l values should be intiger data type. if date in there the format is dd-mm-YYYY.
+        {"role": "user", "content":f""" Extract the values for this given column list:{column_list}, from the given text. all values should be integer data type. if date in given text, the date format should be in dd-mm-YYYY.
         text```{text}```
+        the text may contains other name key and values, use consine similarity to map with column list.
+        the column names should be keys.
         return result should be in JSON format:
-
         """
-        }]
+        }] 
 
         # Generate a response from the GPT-3 model
         chat_completion = self.client.chat.completions.create(
-            model = "GPT-3",
+            model = "gpt-3.5-turbo",
             messages = conversation,
             max_tokens=500,
             temperature=0,
             n=1,
             stop=None,
+            response_format={ "type": "json_object" },
         )
 
         # Extract the generated text from the API response
         generated_text = chat_completion.choices[0].message.content
-
-        # Assuming jsonString is your JSON string
-        json_data = json.loads(generated_text)
-        print("parse_text",json_data)
+        print(generated_text)
+        # # Assuming jsonString is your JSON string
+        try:
+            json_data = json.loads(generated_text)
+        except Exception as e:
+            return e
+        # print("parse_text",json_data)
         return json_data
 
     def main(self, train_csv_path: str, audio_path, target_column, column_list) -> None: