app.py

import streamlit as st
import moviepy.editor as mp
import speech_recognition as sr
from deep_translator import GoogleTranslator
import tempfile
import os
from pydub import AudioSegment
import torch
from TTS.api import TTS
import pyttsx3
import numpy as np
from scipy.io import wavfile
import soundfile as sf

class EnhancedVideoTranslator:
    def __init__(self):
        self.device = "cuda" if torch.cuda.is_available() else "cpu"
        # Initialize Coqui TTS
        try:
            self.tts = TTS(model_name="tts_models/multilingual/multi-dataset/xtts_v2").to(self.device)
        except:
            # Fallback to a smaller model if XTTS fails
            self.tts = TTS(model_name="tts_models/multilingual/multi-dataset/bark").to(self.device)
        
        # Initialize pyttsx3 as backup
        self.pyttsx3_engine = pyttsx3.init()

    def extract_audio(self, video_path):
        video = mp.VideoFileClip(video_path)
        audio = video.audio
        temp_audio = tempfile.NamedTemporaryFile(delete=False, suffix='.wav')
        audio.write_audiofile(temp_audio.name)
        return temp_audio.name

    def enhance_audio(self, audio_path):
        audio = AudioSegment.from_wav(audio_path)
        
        # Noise reduction and enhancement
        enhanced = audio.high_pass_filter(80)
        enhanced = enhanced.low_pass_filter(7500)
        enhanced = enhanced.normalize()
        
        # Apply compression
        enhanced = enhanced.compress_dynamic_range()
        
        enhanced_path = audio_path.replace('.wav', '_enhanced.wav')
        enhanced.export(enhanced_path, format="wav")
        return enhanced_path

    def speech_to_text(self, audio_path):
        recognizer = sr.Recognizer()
        
        with sr.AudioFile(audio_path) as source:
            recognizer.adjust_for_ambient_noise(source)
            audio = recognizer.record(source)
            
        try:
            # Try multiple language detection
            text = recognizer.recognize_google(audio)
            return text
        except Exception as e:
            return str(e)

    def translate_text(self, text, target_lang):
        translator = GoogleTranslator(source='auto', target=target_lang)
        return translator.translate(text)

    def text_to_speech_coqui(self, text, lang):
        try:
            temp_audio = tempfile.NamedTemporaryFile(delete=False, suffix='.wav')
            
            # Use language-specific voices if available
            if lang == 'ta':
                speaker = "tamil_female"
            elif lang == 'hi':
                speaker = "hindi_female"
            else:
                speaker = None
            
            self.tts.tts_to_file(
                text=text,
                file_path=temp_audio.name,
                speaker=speaker,
                language=lang
            )
            
            return temp_audio.name
        except Exception as e:
            print(f"Coqui TTS failed: {e}")
            return self.text_to_speech_pyttsx3(text, lang)

    def text_to_speech_pyttsx3(self, text, lang):
        try:
            temp_audio = tempfile.NamedTemporaryFile(delete=False, suffix='.wav')
            
            # Configure pyttsx3
            engine = self.pyttsx3_engine
            
            # Set language properties
            if lang == 'ta':
                engine.setProperty('voice', 'tamil')
            elif lang == 'hi':
                engine.setProperty('voice', 'hindi')
            
            # Adjust voice properties
            engine.setProperty('rate', 150)    # Speed
            engine.setProperty('volume', 0.9)  # Volume
            
            # Save to file
            engine.save_to_file(text, temp_audio.name)
            engine.runAndWait()
            
            return temp_audio.name
        except Exception as e:
            print(f"pyttsx3 TTS failed: {e}")
            return None

    def improve_audio_quality(self, audio_path):
        # Load audio
        audio, sr = sf.read(audio_path)
        
        # Apply basic audio improvements
        audio = audio * 1.5  # Increase volume
        audio = np.clip(audio, -1, 1)  # Prevent clipping
        
        # Save improved audio
        improved_path = audio_path.replace('.wav', '_improved.wav')
        sf.write(improved_path, audio, sr)
        return improved_path

def main():
    st.title("Enhanced AI Video Translator")
    st.write("Free and Open Source Video Translation with Realistic TTS")

    LANGUAGES = {
        'English': 'en',
        'Tamil': 'ta',
        'Hindi': 'hi',
        'Telugu': 'te',
        'Malayalam': 'ml',
        'Kannada': 'kn',
        'Spanish': 'es',
        'French': 'fr',
        'German': 'de',
        'Japanese': 'ja',
        'Chinese': 'zh',
        'Korean': 'ko'
    }

    translator = EnhancedVideoTranslator()

    # Improved UI
    st.markdown("""
    <style>
    .stButton>button {
        background-color: #4CAF50;
        color: white;
        padding: 10px 24px;
        border-radius: 5px;
    }
    </style>
    """, unsafe_allow_html=True)

    video_file = st.file_uploader(
        "Upload your video", 
        type=['mp4', 'avi', 'mov'],
        help="Supported formats: MP4, AVI, MOV"
    )
    
    if video_file:
        st.video(video_file)
        
        col1, col2 = st.columns(2)
        with col1:
            target_language = st.selectbox(
                "Target Language",
                list(LANGUAGES.keys())
            )
        
        with col2:
            tts_engine = st.selectbox(
                "TTS Engine",
                ["Coqui TTS", "pyttsx3"]
            )

        # Advanced options
        with st.expander("Advanced Settings"):
            quality_enhancement = st.checkbox("Enable Audio Enhancement", True)
            speed = st.slider("Speech Speed", 0.5, 2.0, 1.0, 0.1)
            volume = st.slider("Volume", 0.0, 2.0, 1.0, 0.1)

        if st.button("Translate Video"):
            try:
                progress_bar = st.progress(0)
                status = st.empty()

                # Process video
                temp_video = tempfile.NamedTemporaryFile(delete=False, suffix='.mp4')
                temp_video.write(video_file.read())

                # Extract audio
                status.text("Extracting audio...")
                progress_bar.progress(20)
                audio_path = translator.extract_audio(temp_video.name)

                if quality_enhancement:
                    audio_path = translator.enhance_audio(audio_path)

                # Speech to text
                status.text("Converting speech to text...")
                progress_bar.progress(40)
                original_text = translator.speech_to_text(audio_path)

                # Translate
                status.text("Translating...")
                progress_bar.progress(60)
                translated_text = translator.translate_text(
                    original_text, 
                    LANGUAGES[target_language]
                )

                # Text to speech
                status.text("Generating speech...")
                progress_bar.progress(80)
                
                if tts_engine == "Coqui TTS":
                    translated_audio = translator.text_to_speech_coqui(
                        translated_text,
                        LANGUAGES[target_language]
                    )
                else:
                    translated_audio = translator.text_to_speech_pyttsx3(
                        translated_text,
                        LANGUAGES[target_language]
                    )

                if quality_enhancement:
                    translated_audio = translator.improve_audio_quality(translated_audio)

                # Create final video
                status.text("Creating final video...")
                progress_bar.progress(90)
                
                output_path = "translated_video.mp4"
                video = mp.VideoFileClip(temp_video.name)
                audio = mp.AudioFileClip(translated_audio)
                
                final_video = video.set_audio(audio)
                final_video.write_videofile(output_path)

                progress_bar.progress(100)
                status.text("Complete!")

                # Display results
                st.success("Translation completed!")
                
                col1, col2 = st.columns(2)
                with col1:
                    st.subheader("Original Text")
                    st.write(original_text)
                with col2:
                    st.subheader("Translated Text")
                    st.write(translated_text)

                st.subheader("Translated Video")
                st.video(output_path)

                # Cleanup
                for file in [temp_video.name, audio_path, translated_audio, output_path]:
                    if os.path.exists(file):
                        os.unlink(file)

            except Exception as e:
                st.error(f"An error occurred: {str(e)}")

if __name__ == "__main__":
    main()