app.py

import gradio as gr
import openai
import numpy as np
import re
import time
import emoji
import os
from transformers import pipeline
from deep_translator import GoogleTranslator

# Initialize emotion detection model
emotion_classifier = pipeline(
    "text-classification", 
    model="j-hartmann/emotion-english-distilroberta-base",
    top_k=1
)

# Emotion to emoji mapping
EMOTION_EMOJI_MAP = {
    "anger": "😠",
    "disgust": "🤢",
    "fear": "😨",
    "joy": "😄",
    "neutral": "😐",
    "sadness": "😢",
    "surprise": "😲"
}

# OpenAI TTS voices
VOICES = ["alloy", "echo", "fable", "onyx", "nova", "shimmer"]
DEFAULT_SPEED = 1.0

def detect_emotion(text):
    """Detect emotion from text and return corresponding emoji"""
    if not text.strip():
        return "📝"  # Return pencil emoji for empty text
    
    # Handle case where text is too long
    truncated_text = text[:512]
    
    # Safe emotion detection
    try:
        result = emotion_classifier(truncated_text)
        emotion = result[0][0]['label'].lower()
        return EMOTION_EMOJI_MAP.get(emotion, "❓")
    except Exception:
        return "❓"  # Question mark if detection fails

def translate_text(text, target_lang="en"):
    """Translate text to target language"""
    if not text.strip():
        return ""
    try:
        return GoogleTranslator(source='auto', target=target_lang).translate(text)
    except Exception:
        return text  # Return original if translation fails

def text_to_speech(api_key, text, voice, speed, translate, emotion_boost):
    """Convert text to speech with emotion detection"""
    # Validate inputs
    if not api_key.strip():
        raise gr.Error("🔑 Please enter your OpenAI API key")
    if not text.strip():
        raise gr.Error("📝 Please enter some text")
    
    openai.api_key = api_key
    
    # Translation - FIXED: Check if translate is True (boolean)
    translated_text = text
    if translate is True:  # Explicitly check if it's True
        translated_text = translate_text(text)
    
    # Emotion detection
    emoji_icon = detect_emotion(translated_text)
    
    # Emotion-based speed adjustment
    try:
        # Ensure values are numbers
        speed_val = float(speed)
        boost_val = float(emotion_boost)
        adjusted_speed = max(0.25, min(2.0, speed_val * boost_val))
    except (TypeError, ValueError):
        adjusted_speed = DEFAULT_SPEED
    
    # Generate speech
    try:
        response = openai.audio.speech.create(
            model="tts-1",
            voice=voice,
            input=translated_text,
            speed=adjusted_speed
        )
        
        # Create audio file
        timestamp = int(time.time())
        filename = f"tts_output_{timestamp}.wav"
        response.stream_to_file(filename)
        
        return filename, emoji_icon, f"Speed: {adjusted_speed:.2f}x"
    except Exception as e:
        error_msg = f"⚠️ Error: {str(e)}"
        if "rate limit" in str(e).lower():
            error_msg += "\n🚨 You've hit the rate limit. Please try again later."
        raise gr.Error(error_msg)

# Gradio UI Components
with gr.Blocks(theme=gr.themes.Soft(), title="Advanced OpenAI TTS") as demo:
    gr.Markdown("# <center>🎤 Advanced Text-to-Speech Generator</center>")
    gr.Markdown("<center>Convert text to natural-sounding speech with emotion detection</center>")
    
    with gr.Row():
        with gr.Column(scale=1):
            api_key = gr.Textbox(
                label="OpenAI API Key",
                type="password",
                placeholder="Enter your OpenAI API key...",
                info="Get your API key from [OpenAI Platform](https://platform.openai.com/account/api-keys)"
            )
            
            with gr.Accordion("⚙️ Advanced Settings", open=False):
                voice = gr.Dropdown(
                    label="Voice Style",
                    choices=VOICES,
                    value="nova",
                    interactive=True
                )
                speed = gr.Slider(
                    label="Speech Speed",
                    minimum=0.25,
                    maximum=2.0,
                    value=DEFAULT_SPEED,
                    step=0.05
                )
                emotion_boost = gr.Slider(
                    label="Emotion Intensity",
                    minimum=0.8,
                    maximum=1.5,
                    value=1.0,
                    step=0.1,
                    info="Adjust speed based on emotion"
                )
                translate = gr.Checkbox(
                    label="Auto-translate to English",
                    value=True,
                    info="Supports 100+ languages"
                )
            
            input_text = gr.TextArea(
                label="Input Text",
                placeholder="Enter text to convert to speech...",
                lines=5,
                max_lines=10
            )
            
            btn_generate = gr.Button(
                "✨ Generate Speech",
                variant="primary"
            )
            
        with gr.Column(scale=1):
            emoji_output = gr.Textbox(
                label="Detected Emotion",
                interactive=False,
                placeholder="Emoji will appear here..."
            )
            speed_info = gr.Textbox(
                label="Adjusted Speed",
                interactive=False
            )
            audio_output = gr.Audio(
                label="Generated Speech",
                interactive=False,
                format="wav"
            )
            
            # FIXED: Simplified examples without booleans
            gr.Examples(
                examples=[
                    ["I'm absolutely thrilled about this amazing opportunity!", "nova", 1.0, 1.2],
                    ["This situation makes me feel anxious and worried.", "onyx", 1.0, 1.3],
                    ["Je suis très heureux de vous rencontrer aujourd'hui.", "echo", 1.0, 1.0]
                ],
                inputs=[input_text, voice, speed, emotion_boost],
                label="Example Inputs"
            )
    
    # Event handling
    btn_generate.click(
        fn=text_to_speech,
        inputs=[api_key, input_text, voice, speed, translate, emotion_boost],
        outputs=[audio_output, emoji_output, speed_info]
    )
    
    # Footer
    gr.Markdown("---")
    gr.HTML("""
    <div style="text-align: center">
        <p>🚀 Powered by OpenAI TTS • Emotion Detection • Real-time Translation</p>
        <p>⚠️ Your API key is only used for TTS generation and not stored</p>
    </div>
    """)

# Launch with error handling
try:
    demo.launch(server_name="0.0.0.0", server_port=7860)
except Exception as e:
    print(f"Error launching app: {str(e)}")
    # Create a simple fallback interface
    gr.Interface(
        lambda: "Please check the logs for errors", 
        inputs=None, 
        outputs=gr.Textbox(label="Error")
    ).launch()