app.py

import gradio as gr
import librosa
import numpy as np
import torch
from transformers import Wav2Vec2Processor, Wav2Vec2Model
from simple_salesforce import Salesforce
import os
from datetime import datetime

# Salesforce credentials (store securely in environment variables)
SF_USERNAME = os.getenv("SF_USERNAME", "[email protected]")
SF_PASSWORD = os.getenv("SF_PASSWORD", "voicebot1")
SF_SECURITY_TOKEN = os.getenv("SF_SECURITY_TOKEN", "jq4VVHUFti6TmzJDjjegv2h6b")
SF_INSTANCE_URL = os.getenv("SF_INSTANCE_URL", "https://voicebot-dev-ed.develop.lightning.force.com/lightning/setup/SetupOneHome/home")

# Initialize Salesforce connection
try:
    sf = Salesforce(
        username=SF_USERNAME,
        password=SF_PASSWORD,
        security_token=SF_SECURITY_TOKEN,
        instance_url=SF_INSTANCE_URL
    )
except Exception as e:
    print(f"Failed to connect to Salesforce: {str(e)}")
    sf = None

# Load Wav2Vec2 model for speech feature extraction
processor = Wav2Vec2Processor.from_pretrained("facebook/wav2vec2-base-960h")
model = Wav2Vec2Model.from_pretrained("facebook/wav2vec2-base-960h")

def analyze_voice(audio_file):
    """Analyze voice for health indicators."""
    try:
        # Load audio file
        audio, sr = librosa.load(audio_file, sr=16000)
        
        # Process audio for Wav2Vec2
        inputs = processor(audio, sampling_rate=16000, return_tensors="pt", padding=True)
        with torch.no_grad():
            outputs = model(**inputs)
        
        # Extract features (simplified for demo; real-world needs trained classifier)
        features = outputs.last_hidden_state.mean(dim=1).numpy()
        
        # Placeholder health analysis (replace with trained model)
        respiratory_score = np.mean(features)  # Mock score
        mental_health_score = np.std(features)  # Mock score
        feedback = ""
        if respiratory_score > 0.5:
            feedback += "Possible respiratory issue detected; consult a doctor. "
        if mental_health_score > 0.3:
            feedback += "Possible stress indicators detected; consider professional advice. "
        
        if not feedback:
            feedback = "No significant health indicators detected."
        
        feedback += "\n\n**Disclaimer**: This is not a diagnostic tool. Consult a healthcare provider for medical advice."
        
        # Store in Salesforce
        if sf:
            store_in_salesforce(audio_file, feedback, respiratory_score, mental_health_score)
        
        return feedback
    except Exception as e:
        return f"Error processing audio: {str(e)}"

def store_in_salesforce(audio_file, feedback, respiratory_score, mental_health_score):
    """Store analysis results in Salesforce."""
    try:
        sf.HealthAssessment__c.create({
            "AssessmentDate__c": datetime.utcnow().isoformat(),
            "Feedback__c": feedback,
            "RespiratoryScore__c": float(respiratory_score),
            "MentalHealthScore__c": float(mental_health_score),
            "AudioFileName__c": os.path.basename(audio_file)
        })
    except Exception as e:
        print(f"Failed to store in Salesforce: {str(e)}")

# Gradio interface
iface = gr.Interface(
    fn=analyze_voice,
    inputs=gr.Audio(type="filepath", label="Record or Upload Voice"),
    outputs=gr.Textbox(label="Health Assessment Feedback"),
    title="Health Voice Analyzer",
    description="Record or upload a voice sample for preliminary health assessment. Supports English, Spanish, Hindi, Mandarin."
)

if __name__ == "__main__":
    iface.launch(server_name="0.0.0.0", server_port=7860)