Update app.py

app.py

@@ -6,10 +6,13 @@ from datetime import datetime
 from pydub import AudioSegment
 import pdfplumber, docx, pandas as pd
 from PIL import Image
+import whisper  # Local whisper for transcription
 
 openai.api_key = os.environ.get("GROQ_API_KEY")
 openai.api_base = "https://api.groq.com/openai/v1"
 
+model_whisper = whisper.load_model("base")
+
 # Chat with Groq
 def chat_with_groq(message, state):
     if state is None:
@@ -70,23 +73,23 @@ def load_chat(name):
     except Exception as e:
         return [], {"oai_history": [], "chatbot_ui": []}, f"❌ Could not load {name}: {e}"
 
-# Transcription
-
+# Transcription using local Whisper
 def transcribe_audio(file):
     if not file:
         return ""
     try:
-        with open(file, "rb") as audio_file:
-            transcript = openai.Audio.transcribe("whisper-1", audio_file)
-        return transcript["text"]
+        audio = whisper.load_audio(file)
+        audio = whisper.pad_or_trim(audio)
+        mel = whisper.log_mel_spectrogram(audio).to(model_whisper.device)
+        result = model_whisper.decode(mel)
+        return result.text
     except Exception as e:
         return f"❌ Transcription error: {e}"
 
-# File processor – now sends file content to model for follow-up conversation
-
+# File processor – send file content to textbox for user to ask about
 def file_to_message(file, state):
     if file is None:
-        return ""
+        return "", state
     filename = file.name
     ext = filename.split(".")[-1].lower()
     try:
@@ -107,9 +110,6 @@ def file_to_message(file, state):
         if len(text) > 8000:
             text = text[:8000]
 
-        state["oai_history"].append({"role": "user", "content": f"File uploaded:\n{text}"})
-        state["chatbot_ui"].append({"role": "user", "content": f"File uploaded:\n{text}"})
-
         return text, state
     except Exception as e:
         return f"❌ Error processing file: {e}", state
@@ -157,4 +157,4 @@ textarea, input[type='text'] { background: #f0f0f0; border-radius: 8px; }
     save_btn.click(lambda: gr.update(choices=list_saved_files()), [], [dropdown])
     load_btn.click(load_chat, [dropdown], [chatbot, state, status])
 
-demo.launch()
+demo.launch()