app.py

# Install dependencies
!pip install gTTS python-docx
!apt-get install -y ffmpeg

# Imports
from gtts import gTTS
from IPython.display import Audio, display, HTML, clear_output
import os
from google.colab import files
import tempfile
import docx
import ipywidgets as widgets

# Global state
lines = []
current_speed = 1.0

# Function to extract text from docx
def get_text_from_docx(file_path):
    doc = docx.Document(file_path)
    text = []
    for paragraph in doc.paragraphs:
        text.append(paragraph.text)
    return '\n'.join(text)

# Function to create and play TTS audio
def speak_urdu(text, speed=1.0):
    tts = gTTS(text=text, lang='ur')

    with tempfile.NamedTemporaryFile(delete=False, suffix=".mp3") as tmp_file:
        tts.save(tmp_file.name)
        audio_path = tmp_file.name

    # Adjust speed using ffmpeg
    speed_str = str(speed)
    adjusted_audio = audio_path.replace(".mp3", "_fast.mp3")
    os.system(f"ffmpeg -i {audio_path} -filter:a \"atempo={speed_str}\" -vn {adjusted_audio} -y -loglevel quiet")

    # Play the audio
    display(Audio(adjusted_audio, autoplay=True))

# Upload .docx file using Colab method
uploaded = files.upload()
filename = next(iter(uploaded))

# Process the uploaded .docx file
with tempfile.NamedTemporaryFile(delete=False, suffix=".docx") as tmp_file:
    tmp_file.write(uploaded[filename])
    tmp_file_path = tmp_file.name

text = get_text_from_docx(tmp_file_path)
lines = [line.strip() for line in text.split('\n') if line.strip()]

print("✅ File uploaded and processed successfully!")

# Speed slider
speed_slider = widgets.FloatSlider(
    value=1.0,
    min=0.5,
    max=1.5,
    step=0.1,
    description='Speed:',
    continuous_update=False
)

def on_speed_change(change):
    global current_speed
    current_speed = change['new']

speed_slider.observe(on_speed_change, names='value')
display(speed_slider)

# Function to read the text aloud
def display_line(line, idx):
    styled_line = f"<h3 style='font-family: sans-serif; color: #333;'>📖 Line {idx + 1}: {line}</h3>"
    display(HTML(styled_line))

def read_text(_=None):
    clear_output(wait=True)
    display(speed_slider)
    display(read_button)
    for idx, line in enumerate(lines):
        display_line(line, idx)
        speak_urdu(line, speed=current_speed)
        input("🔸 Press Enter to read next line...")

# Read button
read_button = widgets.Button(description="🔊 Read Text")
read_button.on_click(read_text)
display(read_button)