Update app.py

app.py

@@ -6,48 +6,106 @@ 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__)
 
-# Check GPU availability
+# Use whisper-small for faster processing and better speed
 device = "cuda" if torch.cuda.is_available() else "cpu"
 
-# Whisper ASR config
+# Create config object to set timeout and other parameters
 config = AutoConfig.from_pretrained("openai/whisper-small")
-config.update({"timeout": 60})
+config.update({"timeout": 60})  # Set timeout to 60 seconds
 
-# Salesforce connection
+# Your function where you generate and save the audio
+def generate_audio_prompt(text, filename):
+    try:
+        tts = gTTS(text)
+        tts.save(os.path.join("static", filename))
+    except gtts.tts.gTTSError as e:
+        print(f"Error: {e}")
+        print("Retrying after 5 seconds...")
+        time.sleep(5)  # Wait for 5 seconds 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")
+
+# Symbol mapping for proper recognition
+SYMBOL_MAPPING = {
+    "at the rate": "@",
+    "at": "@",
+    "dot": ".",
+    "underscore": "_",
+    "hash": "#",
+    "plus": "+",
+    "dash": "-",
+    "comma": ",",
+    "space": " "
+}
+
+# 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:
+        print(f"Error: {str(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)  # Reduced silence duration
+    print(f"Detected nonsilent parts: {nonsilent_parts}")
+    return len(nonsilent_parts) == 0  # If no speech detected
+
+# Salesforce connection details
 try:
-    print("Connecting to Salesforce...")
+    print("Attempting to connect to Salesforce...")
     sf = Salesforce(username='[email protected]', password='Sati@1020', security_token='sSSjyhInIsUohKpG8sHzty2q')
-    print("Connected successfully!")
+    print("Connected to Salesforce successfully!")
+    print("User Info:", sf.UserInfo)  # Log the user info to verify the connection
 except Exception as e:
-    print(f"Salesforce connection failed: {str(e)}")
-
-# API Routes
+    print(f"Failed to connect to Salesforce: {str(e)}")
 
+# Function to create Salesforce record
+# API endpoint to receive data from voice bot
 @app.route('/login', methods=['POST'])
 def login():
-    data = request.json
+    # Get data from voice bot (name, email, phone number)
+    data = request.json  # Assuming voice bot sends JSON data
+    
+    name = data.get('name')
     email = data.get('email')
     phone_number = data.get('phone_number')
 
-    if not email or not phone_number:
+    if not name or not email or not phone_number:
         return jsonify({'error': 'Missing required fields'}), 400
 
+    # Create a record in Salesforce
     try:
-        existing_customer = sf.query(f"SELECT Id, Name FROM Customer_Login__c WHERE Email__c = '{email}' AND Phone_Number__c = '{phone_number}'")
-        if existing_customer['totalSize'] > 0:
-            return jsonify({'success': True, 'message': 'Login successful!', 'customer': existing_customer['records'][0]})
-        else:
-            return jsonify({'error': 'No user found. Please register first.'}), 404
+        customer_login = sf.Customer_Login__c.create({
+            'Name': name,
+            'Email__c': email,
+            'Phone_Number__c': phone_number
+        })
+        return jsonify({'success': True, 'id': customer_login['id']}), 200
     except Exception as e:
-        return jsonify({'error': f'Failed to fetch user from Salesforce: {str(e)}'}), 500
+        return jsonify({'error': f'Failed to create record in Salesforce: {str(e)}'}), 500
 
 @app.route("/submit", methods=["POST"])
 def submit():
@@ -60,25 +118,90 @@ def submit():
         return jsonify({'error': 'Missing data'}), 400
 
     try:
-        existing_customer = sf.query(f"SELECT Id FROM Customer_Login__c WHERE Email__c = '{email}' AND Phone_Number__c = '{phone}'")
-        if existing_customer['totalSize'] > 0:
-            return jsonify({'success': False, 'message': 'User already registered. Please log in.'}), 409
-
+        # Create Salesforce record
         customer_login = sf.Customer_Login__c.create({
             'Name': name,
             'Email__c': email,
             'Phone_Number__c': phone
         })
 
-        return jsonify({'success': True, 'message': 'Registration successful!', 'id': customer_login['id']}), 200
+        if customer_login.get('id'):
+            return jsonify({'success': True})
+        else:
+            return jsonify({'error': 'Failed to create record'}), 500
 
     except Exception as e:
         return jsonify({'error': str(e)}), 500
 
+
 @app.route("/")
 def index():
     return render_template("index.html")
 
-# Start server
+@app.route("/transcribe", methods=["POST"])
+def transcribe():
+    if "audio" not in request.files:
+        print("No audio file provided")
+        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
+        convert_to_wav(input_audio_path, output_audio_path)
+
+        # Check for silence
+        if is_silent_audio(output_audio_path):
+            return jsonify({"error": "No speech detected. Please try again."}), 400
+        else:
+            print("Audio contains speech, proceeding with transcription.")
+
+        # Use Whisper ASR model for transcription
+        result = None
+        retry_attempts = 3
+        for attempt in range(retry_attempts):
+            try:
+                result = pipeline("automatic-speech-recognition", model="openai/whisper-small", device=0 if torch.cuda.is_available() else -1, config=config)
+                print(f"Transcribed text: {result['text']}")
+                break
+            except requests.exceptions.ReadTimeout:
+                print(f"Timeout occurred, retrying attempt {attempt + 1}/{retry_attempts}...")
+                time.sleep(5)
+
+        if result is None:
+            return jsonify({"error": "Unable to transcribe audio after retries."}), 500
+
+        transcribed_text = result["text"].strip().capitalize()
+        print(f"Transcribed text: {transcribed_text}")
+
+        # Extract name, email, and phone number from the transcribed text
+        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"
+        print(f"Parsed data - Name: {name}, Email: {email}, Phone Number: {phone_number}")
+
+        # Create record in Salesforce
+        salesforce_response = create_salesforce_record(name, email, phone_number)
+
+        # Log the Salesforce response
+        print(f"Salesforce record creation response: {salesforce_response}")
+
+        # Check if the response contains an error
+        if "error" in salesforce_response:
+            print(f"Error creating record in Salesforce: {salesforce_response['error']}")
+            return jsonify(salesforce_response), 500
+
+        # If creation was successful, return the details
+        return jsonify({"text": transcribed_text, "salesforce_record": salesforce_response})
+
+    except Exception as e:
+        print(f"Error in transcribing or processing: {str(e)}")
+        return jsonify({"error": f"Speech recognition error: {str(e)}"}), 500
+
+# Start Production Server
 if __name__ == "__main__":
-    serve(app, host="0.0.0.0", port=7860)
+    serve(app, host="0.0.0.0", port=7860)