Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,27 +1,52 @@
|
|
| 1 |
-
import torch
|
| 2 |
-
from flask import Flask, render_template, request, jsonify
|
| 3 |
-
import json
|
| 4 |
import os
|
| 5 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6 |
from gtts import gTTS
|
| 7 |
from pydub import AudioSegment
|
| 8 |
from pydub.silence import detect_nonsilent
|
| 9 |
-
from transformers import AutoConfig # Import AutoConfig for the config object
|
| 10 |
-
import time
|
| 11 |
from waitress import serve
|
| 12 |
-
from simple_salesforce import Salesforce
|
| 13 |
-
import requests # Import requests for exception handling
|
| 14 |
|
| 15 |
app = Flask(__name__)
|
| 16 |
|
| 17 |
-
#
|
| 18 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 19 |
|
| 20 |
-
#
|
|
|
|
| 21 |
config = AutoConfig.from_pretrained("openai/whisper-small")
|
| 22 |
-
config.update({"timeout": 60})
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 23 |
|
| 24 |
-
# Your function where you generate and save the audio
|
| 25 |
def generate_audio_prompt(text, filename):
|
| 26 |
try:
|
| 27 |
tts = gTTS(text)
|
|
@@ -29,96 +54,83 @@ def generate_audio_prompt(text, filename):
|
|
| 29 |
except gtts.tts.gTTSError as e:
|
| 30 |
print(f"Error: {e}")
|
| 31 |
print("Retrying after 5 seconds...")
|
| 32 |
-
time.sleep(5)
|
| 33 |
generate_audio_prompt(text, filename)
|
| 34 |
|
| 35 |
-
# Generate required voice prompts
|
| 36 |
-
prompts = {
|
| 37 |
-
"welcome": "Welcome to Biryani Hub.",
|
| 38 |
-
"ask_name": "Tell me your name.",
|
| 39 |
-
"ask_email": "Please provide your email address.",
|
| 40 |
-
"thank_you": "Thank you for registration."
|
| 41 |
-
}
|
| 42 |
-
|
| 43 |
for key, text in prompts.items():
|
| 44 |
generate_audio_prompt(text, f"{key}.mp3")
|
| 45 |
|
| 46 |
-
# Symbol mapping for proper recognition
|
| 47 |
-
SYMBOL_MAPPING = {
|
| 48 |
-
"at the rate": "@",
|
| 49 |
-
"at": "@",
|
| 50 |
-
"dot": ".",
|
| 51 |
-
"underscore": "_",
|
| 52 |
-
"hash": "#",
|
| 53 |
-
"plus": "+",
|
| 54 |
-
"dash": "-",
|
| 55 |
-
"comma": ",",
|
| 56 |
-
"space": " "
|
| 57 |
-
}
|
| 58 |
-
|
| 59 |
# Function to convert audio to WAV format
|
| 60 |
def convert_to_wav(input_path, output_path):
|
| 61 |
try:
|
| 62 |
audio = AudioSegment.from_file(input_path)
|
| 63 |
-
audio = audio.set_frame_rate(16000).set_channels(1)
|
| 64 |
audio.export(output_path, format="wav")
|
| 65 |
except Exception as e:
|
| 66 |
-
print(f"Error: {str(e)}")
|
| 67 |
raise Exception(f"Audio conversion failed: {str(e)}")
|
| 68 |
|
| 69 |
# Function to check if audio contains actual speech
|
| 70 |
def is_silent_audio(audio_path):
|
| 71 |
audio = AudioSegment.from_wav(audio_path)
|
| 72 |
-
nonsilent_parts = detect_nonsilent(audio, min_silence_len=500, silence_thresh=audio.dBFS-16)
|
| 73 |
-
|
| 74 |
-
return len(nonsilent_parts) == 0 # If no speech detected
|
| 75 |
|
| 76 |
-
|
| 77 |
-
|
| 78 |
-
|
| 79 |
-
sf = Salesforce(username='diggavalli98@gmail.com', password='Sati@1020', security_token='sSSjyhInIsUohKpG8sHzty2q')
|
| 80 |
-
print("Connected to Salesforce successfully!")
|
| 81 |
-
print("User Info:", sf.UserInfo) # Log the user info to verify the connection
|
| 82 |
-
except Exception as e:
|
| 83 |
-
print(f"Failed to connect to Salesforce: {str(e)}")
|
| 84 |
|
| 85 |
-
#
|
| 86 |
-
# API endpoint to receive data from voice bot
|
| 87 |
@app.route('/login', methods=['POST'])
|
| 88 |
def login():
|
| 89 |
-
|
| 90 |
-
|
| 91 |
-
|
| 92 |
-
name = data.get('name')
|
| 93 |
-
email = data.get('email')
|
| 94 |
-
phone_number = data.get('phone_number')
|
| 95 |
|
| 96 |
-
if not
|
| 97 |
-
return jsonify({'error': 'Missing
|
| 98 |
|
| 99 |
-
# Create a record in Salesforce
|
| 100 |
try:
|
| 101 |
-
|
| 102 |
-
|
| 103 |
-
|
| 104 |
-
|
| 105 |
-
|
| 106 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 107 |
except Exception as e:
|
| 108 |
-
|
|
|
|
| 109 |
|
| 110 |
-
|
| 111 |
-
|
|
|
|
| 112 |
data = request.json
|
| 113 |
name = data.get('name')
|
| 114 |
-
email = data.get('email')
|
| 115 |
-
phone = data.get('phone')
|
| 116 |
|
| 117 |
if not name or not email or not phone:
|
| 118 |
return jsonify({'error': 'Missing data'}), 400
|
| 119 |
|
| 120 |
try:
|
| 121 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 122 |
customer_login = sf.Customer_Login__c.create({
|
| 123 |
'Name': name,
|
| 124 |
'Email__c': email,
|
|
@@ -126,22 +138,17 @@ def submit():
|
|
| 126 |
})
|
| 127 |
|
| 128 |
if customer_login.get('id'):
|
| 129 |
-
return jsonify({'success': True})
|
| 130 |
else:
|
| 131 |
return jsonify({'error': 'Failed to create record'}), 500
|
| 132 |
|
| 133 |
except Exception as e:
|
| 134 |
return jsonify({'error': str(e)}), 500
|
| 135 |
|
| 136 |
-
|
| 137 |
-
@app.route("/")
|
| 138 |
-
def index():
|
| 139 |
-
return render_template("index.html")
|
| 140 |
-
|
| 141 |
@app.route("/transcribe", methods=["POST"])
|
| 142 |
def transcribe():
|
| 143 |
if "audio" not in request.files:
|
| 144 |
-
print("No audio file provided")
|
| 145 |
return jsonify({"error": "No audio file provided"}), 400
|
| 146 |
|
| 147 |
audio_file = request.files["audio"]
|
|
@@ -150,56 +157,17 @@ def transcribe():
|
|
| 150 |
audio_file.save(input_audio_path)
|
| 151 |
|
| 152 |
try:
|
| 153 |
-
# Convert to WAV
|
| 154 |
convert_to_wav(input_audio_path, output_audio_path)
|
| 155 |
|
| 156 |
-
# Check for silence
|
| 157 |
if is_silent_audio(output_audio_path):
|
| 158 |
return jsonify({"error": "No speech detected. Please try again."}), 400
|
| 159 |
-
|
| 160 |
-
|
| 161 |
-
|
| 162 |
-
|
| 163 |
-
|
| 164 |
-
retry_attempts = 3
|
| 165 |
-
for attempt in range(retry_attempts):
|
| 166 |
-
try:
|
| 167 |
-
result = pipeline("automatic-speech-recognition", model="openai/whisper-small", device=0 if torch.cuda.is_available() else -1, config=config)
|
| 168 |
-
print(f"Transcribed text: {result['text']}")
|
| 169 |
-
break
|
| 170 |
-
except requests.exceptions.ReadTimeout:
|
| 171 |
-
print(f"Timeout occurred, retrying attempt {attempt + 1}/{retry_attempts}...")
|
| 172 |
-
time.sleep(5)
|
| 173 |
-
|
| 174 |
-
if result is None:
|
| 175 |
-
return jsonify({"error": "Unable to transcribe audio after retries."}), 500
|
| 176 |
-
|
| 177 |
-
transcribed_text = result["text"].strip().capitalize()
|
| 178 |
-
print(f"Transcribed text: {transcribed_text}")
|
| 179 |
-
|
| 180 |
-
# Extract name, email, and phone number from the transcribed text
|
| 181 |
-
parts = transcribed_text.split()
|
| 182 |
-
name = parts[0] if len(parts) > 0 else "Unknown Name"
|
| 183 |
-
email = parts[1] if '@' in parts[1] else "unknown@domain.com"
|
| 184 |
-
phone_number = parts[2] if len(parts) > 2 else "0000000000"
|
| 185 |
-
print(f"Parsed data - Name: {name}, Email: {email}, Phone Number: {phone_number}")
|
| 186 |
-
|
| 187 |
-
# Create record in Salesforce
|
| 188 |
-
salesforce_response = create_salesforce_record(name, email, phone_number)
|
| 189 |
-
|
| 190 |
-
# Log the Salesforce response
|
| 191 |
-
print(f"Salesforce record creation response: {salesforce_response}")
|
| 192 |
-
|
| 193 |
-
# Check if the response contains an error
|
| 194 |
-
if "error" in salesforce_response:
|
| 195 |
-
print(f"Error creating record in Salesforce: {salesforce_response['error']}")
|
| 196 |
-
return jsonify(salesforce_response), 500
|
| 197 |
-
|
| 198 |
-
# If creation was successful, return the details
|
| 199 |
-
return jsonify({"text": transcribed_text, "salesforce_record": salesforce_response})
|
| 200 |
|
| 201 |
except Exception as e:
|
| 202 |
-
print(f"Error in transcribing or processing: {str(e)}")
|
| 203 |
return jsonify({"error": f"Speech recognition error: {str(e)}"}), 500
|
| 204 |
|
| 205 |
# Start Production Server
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
import os
|
| 2 |
+
import time
|
| 3 |
+
import logging
|
| 4 |
+
import json
|
| 5 |
+
import requests
|
| 6 |
+
import torch
|
| 7 |
+
from flask import Flask, render_template, request, jsonify, session
|
| 8 |
+
from flask_session import Session
|
| 9 |
+
from simple_salesforce import Salesforce
|
| 10 |
+
from transformers import pipeline, AutoConfig
|
| 11 |
from gtts import gTTS
|
| 12 |
from pydub import AudioSegment
|
| 13 |
from pydub.silence import detect_nonsilent
|
|
|
|
|
|
|
| 14 |
from waitress import serve
|
|
|
|
|
|
|
| 15 |
|
| 16 |
app = Flask(__name__)
|
| 17 |
|
| 18 |
+
# Configure Flask session
|
| 19 |
+
app.secret_key = os.getenv("SECRET_KEY", "sSSjyhInIsUohKpG8sHzty2q")
|
| 20 |
+
app.config["SESSION_TYPE"] = "filesystem"
|
| 21 |
+
app.config["SESSION_COOKIE_NAME"] = "my_session"
|
| 22 |
+
app.config["SESSION_COOKIE_SECURE"] = True
|
| 23 |
+
app.config["SESSION_COOKIE_SAMESITE"] = "None"
|
| 24 |
+
Session(app)
|
| 25 |
+
|
| 26 |
+
# Set up logging
|
| 27 |
+
logging.basicConfig(level=logging.INFO)
|
| 28 |
+
|
| 29 |
+
# Connect to Salesforce
|
| 30 |
+
try:
|
| 31 |
+
print("Attempting to connect to Salesforce...")
|
| 32 |
+
sf = Salesforce(username='diggavalli98@gmail.com', password='Sati@1020', security_token='sSSjyhInIsUohKpG8sHzty2q')
|
| 33 |
+
print("Connected to Salesforce successfully!")
|
| 34 |
+
except Exception as e:
|
| 35 |
+
print(f"Failed to connect to Salesforce: {str(e)}")
|
| 36 |
|
| 37 |
+
# Whisper ASR Configuration
|
| 38 |
+
device = "cuda" if torch.cuda.is_available() else "cpu"
|
| 39 |
config = AutoConfig.from_pretrained("openai/whisper-small")
|
| 40 |
+
config.update({"timeout": 60})
|
| 41 |
+
|
| 42 |
+
# Voice prompts for registration
|
| 43 |
+
prompts = {
|
| 44 |
+
"welcome": "Welcome to Biryani Hub.",
|
| 45 |
+
"ask_name": "Tell me your name.",
|
| 46 |
+
"ask_email": "Please provide your email address.",
|
| 47 |
+
"thank_you": "Thank you for registration."
|
| 48 |
+
}
|
| 49 |
|
|
|
|
| 50 |
def generate_audio_prompt(text, filename):
|
| 51 |
try:
|
| 52 |
tts = gTTS(text)
|
|
|
|
| 54 |
except gtts.tts.gTTSError as e:
|
| 55 |
print(f"Error: {e}")
|
| 56 |
print("Retrying after 5 seconds...")
|
| 57 |
+
time.sleep(5)
|
| 58 |
generate_audio_prompt(text, filename)
|
| 59 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 60 |
for key, text in prompts.items():
|
| 61 |
generate_audio_prompt(text, f"{key}.mp3")
|
| 62 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 63 |
# Function to convert audio to WAV format
|
| 64 |
def convert_to_wav(input_path, output_path):
|
| 65 |
try:
|
| 66 |
audio = AudioSegment.from_file(input_path)
|
| 67 |
+
audio = audio.set_frame_rate(16000).set_channels(1)
|
| 68 |
audio.export(output_path, format="wav")
|
| 69 |
except Exception as e:
|
|
|
|
| 70 |
raise Exception(f"Audio conversion failed: {str(e)}")
|
| 71 |
|
| 72 |
# Function to check if audio contains actual speech
|
| 73 |
def is_silent_audio(audio_path):
|
| 74 |
audio = AudioSegment.from_wav(audio_path)
|
| 75 |
+
nonsilent_parts = detect_nonsilent(audio, min_silence_len=500, silence_thresh=audio.dBFS-16)
|
| 76 |
+
return len(nonsilent_parts) == 0
|
|
|
|
| 77 |
|
| 78 |
+
@app.route("/")
|
| 79 |
+
def index():
|
| 80 |
+
return render_template("index.html")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 81 |
|
| 82 |
+
# β
LOGIN ENDPOINT (Validates Existing Users)
|
|
|
|
| 83 |
@app.route('/login', methods=['POST'])
|
| 84 |
def login():
|
| 85 |
+
data = request.json
|
| 86 |
+
email = data.get('email').strip().lower()
|
| 87 |
+
phone_number = data.get('phone_number').strip()
|
|
|
|
|
|
|
|
|
|
| 88 |
|
| 89 |
+
if not email or not phone_number:
|
| 90 |
+
return jsonify({'error': 'Missing email or phone number'}), 400
|
| 91 |
|
|
|
|
| 92 |
try:
|
| 93 |
+
print(f"π Checking login for Email: {email}, Phone: {phone_number}")
|
| 94 |
+
|
| 95 |
+
query = f"SELECT Id, Name FROM Customer_Login__c WHERE LOWER(Email__c) = '{email}' AND Phone_Number__c = '{phone_number}' LIMIT 1"
|
| 96 |
+
result = sf.query(query)
|
| 97 |
+
|
| 98 |
+
if result['totalSize'] == 0:
|
| 99 |
+
print("β No matching records found!")
|
| 100 |
+
return jsonify({'error': 'Invalid email or phone number. User not found'}), 401
|
| 101 |
+
|
| 102 |
+
user_data = result['records'][0]
|
| 103 |
+
session['user_id'] = user_data['Id']
|
| 104 |
+
session['name'] = user_data['Name']
|
| 105 |
+
print("β
User found:", user_data)
|
| 106 |
+
|
| 107 |
+
return jsonify({'success': True, 'message': 'Login successful', 'user_id': user_data['Id'], 'name': user_data['Name']}), 200
|
| 108 |
+
|
| 109 |
+
except requests.exceptions.RequestException as req_error:
|
| 110 |
+
print("π΄ Salesforce Connection Error:", req_error)
|
| 111 |
+
return jsonify({'error': f'Salesforce connection error: {str(req_error)}'}), 500
|
| 112 |
except Exception as e:
|
| 113 |
+
print("π¨ Unexpected Error:", e)
|
| 114 |
+
return jsonify({'error': f'Unexpected error: {str(e)}'}), 500
|
| 115 |
|
| 116 |
+
# β
REGISTRATION ENDPOINT (Creates New User)
|
| 117 |
+
@app.route("/register", methods=["POST"])
|
| 118 |
+
def register():
|
| 119 |
data = request.json
|
| 120 |
name = data.get('name')
|
| 121 |
+
email = data.get('email').strip().lower()
|
| 122 |
+
phone = data.get('phone').strip()
|
| 123 |
|
| 124 |
if not name or not email or not phone:
|
| 125 |
return jsonify({'error': 'Missing data'}), 400
|
| 126 |
|
| 127 |
try:
|
| 128 |
+
query = f"SELECT Id FROM Customer_Login__c WHERE LOWER(Email__c) = '{email}' AND Phone_Number__c = '{phone}' LIMIT 1"
|
| 129 |
+
existing_user = sf.query(query)
|
| 130 |
+
|
| 131 |
+
if existing_user['totalSize'] > 0:
|
| 132 |
+
return jsonify({'error': 'User already exists'}), 409
|
| 133 |
+
|
| 134 |
customer_login = sf.Customer_Login__c.create({
|
| 135 |
'Name': name,
|
| 136 |
'Email__c': email,
|
|
|
|
| 138 |
})
|
| 139 |
|
| 140 |
if customer_login.get('id'):
|
| 141 |
+
return jsonify({'success': True, 'user_id': customer_login['id']}), 200
|
| 142 |
else:
|
| 143 |
return jsonify({'error': 'Failed to create record'}), 500
|
| 144 |
|
| 145 |
except Exception as e:
|
| 146 |
return jsonify({'error': str(e)}), 500
|
| 147 |
|
| 148 |
+
# β
TRANSCRIPTION ENDPOINT (Converts Speech to Text)
|
|
|
|
|
|
|
|
|
|
|
|
|
| 149 |
@app.route("/transcribe", methods=["POST"])
|
| 150 |
def transcribe():
|
| 151 |
if "audio" not in request.files:
|
|
|
|
| 152 |
return jsonify({"error": "No audio file provided"}), 400
|
| 153 |
|
| 154 |
audio_file = request.files["audio"]
|
|
|
|
| 157 |
audio_file.save(input_audio_path)
|
| 158 |
|
| 159 |
try:
|
|
|
|
| 160 |
convert_to_wav(input_audio_path, output_audio_path)
|
| 161 |
|
|
|
|
| 162 |
if is_silent_audio(output_audio_path):
|
| 163 |
return jsonify({"error": "No speech detected. Please try again."}), 400
|
| 164 |
+
|
| 165 |
+
result = pipeline("automatic-speech-recognition", model="openai/whisper-small", device=0 if torch.cuda.is_available() else -1, config=config)
|
| 166 |
+
transcribed_text = result(output_audio_path)["text"].strip().capitalize()
|
| 167 |
+
|
| 168 |
+
return jsonify({"text": transcribed_text})
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 169 |
|
| 170 |
except Exception as e:
|
|
|
|
| 171 |
return jsonify({"error": f"Speech recognition error: {str(e)}"}), 500
|
| 172 |
|
| 173 |
# Start Production Server
|