Update app.py

app.py

@@ -6,22 +6,65 @@ from transformers import pipeline
 from gtts import gTTS
 from pydub import AudioSegment
 from pydub.silence import detect_nonsilent
-from transformers import AutoConfig
+from transformers import AutoConfig  # Import AutoConfig for the config object
 import time
 from waitress import serve
 from simple_salesforce import Salesforce
-import requests
+import requests  # Import requests for exception handling
 
 app = Flask(__name__, template_folder="templates")
 app.secret_key = os.urandom(24)
 
-# ✅ Salesforce Connection Setup
+# Use whisper-small for faster processing and better speed
+device = "cuda" if torch.cuda.is_available() else "cpu"
+
+# Create config object to set timeout and other parameters
+config = AutoConfig.from_pretrained("openai/whisper-small")
+config.update({"timeout": 60})  # Set timeout to 60 seconds
+
+# Function to generate audio prompts
+def generate_audio_prompt(text, filename):
+    try:
+        tts = gTTS(text)
+        tts.save(os.path.join("static", filename))
+    except Exception as e:
+        print(f"Error: {e}")
+        time.sleep(5)  # Wait before retrying
+        generate_audio_prompt(text, filename)
+
+# Generate required voice prompts
+prompts = {
+    "welcome": "Welcome to Biryani Hub.",
+    "ask_name": "Tell me your name.",
+    "ask_email": "Please provide your email address.",
+    "thank_you": "Thank you for registration."
+}
+
+for key, text in prompts.items():
+    generate_audio_prompt(text, f"{key}.mp3")
+
+# Function to convert audio to WAV format
+def convert_to_wav(input_path, output_path):
+    try:
+        audio = AudioSegment.from_file(input_path)
+        audio = audio.set_frame_rate(16000).set_channels(1)  # Convert to 16kHz, mono
+        audio.export(output_path, format="wav")
+    except Exception as e:
+        raise Exception(f"Audio conversion failed: {str(e)}")
+
+# Function to check if audio contains actual speech
+def is_silent_audio(audio_path):
+    audio = AudioSegment.from_wav(audio_path)
+    nonsilent_parts = detect_nonsilent(audio, min_silence_len=500, silence_thresh=audio.dBFS-16)
+    return len(nonsilent_parts) == 0
+
+# Salesforce connection details
 try:
     print("Attempting to connect to Salesforce...")
     sf = Salesforce(username='[email protected]', password='Sati@1020', security_token='sSSjyhInIsUohKpG8sHzty2q')
-    print("✅ Connected to Salesforce successfully!")
+    print("Connected to Salesforce successfully!")
 except Exception as e:
-    print(f"❌ Failed to connect to Salesforce: {str(e)}")
+    print(f"Failed to connect to Salesforce: {str(e)}")
 
 # ✅ HOME ROUTE (Loads `index.html`)
 @app.route("/", methods=["GET"])
@@ -33,7 +76,7 @@ def index():
 def dashboard():
     return render_template("dashboard.html")
 
-# ✅ MENU PAGE ROUTE (NEWLY ADDED)
+# ✅ MENU PAGE ROUTE
 @app.route("/menu_page", methods=["GET"])
 def menu_page():
     return render_template("menu_page.html")
@@ -80,6 +123,46 @@ def submit():
     except Exception as e:
         return jsonify({'error': str(e)}), 500
 
+# ✅ TRANSCRIBE AUDIO API
+@app.route("/transcribe", methods=["POST"])
+def transcribe():
+    if "audio" not in request.files:
+        return jsonify({"error": "No audio file provided"}), 400
+
+    audio_file = request.files["audio"]
+    input_audio_path = os.path.join("static", "temp_input.wav")
+    output_audio_path = os.path.join("static", "temp.wav")
+    audio_file.save(input_audio_path)
+
+    try:
+        convert_to_wav(input_audio_path, output_audio_path)
+        if is_silent_audio(output_audio_path):
+            return jsonify({"error": "No speech detected. Please try again."}), 400
+
+        asr_pipeline = pipeline("automatic-speech-recognition", model="openai/whisper-small", device=0 if torch.cuda.is_available() else -1, config=config)
+        result = asr_pipeline(output_audio_path)
+
+        transcribed_text = result["text"].strip().capitalize()
+
+        parts = transcribed_text.split()
+        name = parts[0] if len(parts) > 0 else "Unknown Name"
+        email = parts[1] if '@' in parts[1] else "[email protected]"
+        phone_number = parts[2] if len(parts) > 2 else "0000000000"
+
+        confirmation = f"Is this correct? Name: {name}, Email: {email}, Phone: {phone_number}"
+        generate_audio_prompt(confirmation, "confirmation.mp3")
+
+        salesforce_response = sf.Customer_Login__c.create({
+            'Name': name,
+            'Email__c': email,
+            'Phone_Number__c': phone_number
+        })
+
+        return jsonify({"text": transcribed_text, "salesforce_record": salesforce_response})
+
+    except Exception as e:
+        return jsonify({"error": f"Speech recognition error: {str(e)}"}), 500
+
 # ✅ MENU API
 @app.route("/menu", methods=["GET"])
 def get_menu():