app.py

# DEPENDENCIES

import time
import pyperclip
import streamlit as st
from config.config import Config
from src.utils.logger import LoggerSetup
from src.utils.model_loader import ModelLoader
from src.audio_recorder.record_audio import AudioRecorder
from src.audio_transcriber.transcribe_audio import AudioTranscriber

# LOGGER SETUP
app_logger = LoggerSetup(logger_name = "app.py", log_filename_prefix = "app").get_logger()

# STREAMLIT CONFIGURATION
st.set_page_config(page_title = "Speechify", page_icon = "🎙️", layout = "wide")

# CUSTOM CSS STYLING
st.markdown(
    """
    <style>
    body {
        background-color: #f4f4f4;
    }
    .main-title {
        font-size: 40px;
        font-weight: bold;
        color: #6A1B9A;
        text-align: center;
    }
    .record-button {
        background-color: #6200EE;
        color: white;
        font-size: 18px;
        padding: 12px;
        border-radius: 10px;
    }
    .transcription-box {
        background-color: #ffffff;
        padding: 15px;
        border-radius: 10px;
        box-shadow: 2px 2px 10px rgba(0, 0, 0, 0.1);
    }
    </style>
    """,
    unsafe_allow_html = True
)

@st.cache_resource
def load_model():
    """Loads the model and processor and caches them."""
    
    try:

        app_logger.info("Loading model...")
        
        model_loader = ModelLoader(model_name=Config.MODEL_NAME)

        app_logger.info("Model loaded successfully.")
        
        return model_loader.load_model()
    
    except Exception as e:
        
        app_logger.error(f"Error loading model: {repr(e)}")
        
        return None, None

try:

    # Centered Container
    with st.container():
        
        st.markdown("<div class = 'container'>", unsafe_allow_html = True)

        st.markdown("<h1 class='main-title'>🎙️ Speechify</h1>", unsafe_allow_html = True)
        st.markdown("<p style='text-align:center; font-size:18px;'><b>Click the button below to record your voice and get an instant transcription.</b></p>",
                     unsafe_allow_html = True)
        

        recording_seconds                  = st.slider("Recording Duration (in seconds):", 
                                                       min_value  = 1, 
                                                       max_value  = 60, 
                                                       value      = Config.RECORD_SECONDS, 
                                                       step       = 1)

        # Load model and processor from cache
        processor, model                   = load_model()

        # Initialize recorder and transcriber
        recorder                           = AudioRecorder(rate            = Config.SAMPLE_RATE, 
                                                           channels        = Config.CHANNELS, 
                                                           chunk           = Config.CHUNK, 
                                                           record_seconds  = recording_seconds
                                                           )
        
        transcriber                        = AudioTranscriber(processor = processor, model = model)

        # Ensure session state stores transcription
        if "transcription" not in st.session_state:
            st.session_state.transcription = ""
        
        if "audio_path" not in st.session_state:
            st.session_state.audio_path    = None

        if st.button("🎙️ Record and Transcribe", key = "record", help = "Click to start recording"):

            st.write("🎤 **Recording started...**")
                    
            audio_path                     = recorder.record_audio()  
            st.session_state.audio_path    = audio_path

            transcription                  = transcriber.transcribe_audio(audio_path, sample_rate = Config.SAMPLE_RATE) 
            st.session_state.transcription = transcription
            
            st.write("✅ **Recording complete. Transcribing...**")

        audio_area, space, transcription_area = st.columns([4, 1, 4])

        with audio_area:
           
            st.subheader("🔊 Recorded Audio")
            
            if st.session_state.audio_path:
                st.audio(st.session_state.audio_path, format = "audio/wav")

        with transcription_area:  

            st.subheader("Transcription:")
            
            st.write(st.session_state.transcription)

            copy_button, download_button, space = st.columns([1, 1, 6])

            with copy_button:

                if st.session_state.transcription:
                
                    if st.button("📋", key = "copy_text", help = "Copy your Text to Clipboard"):
                        
                        pyperclip.copy(st.session_state.transcription)  
                        
                        st.toast("✅ Text copied to clipboard!")

            with download_button:

                if st.session_state.transcription:
                
                    st.download_button(label      = "📥",
                                       data       = st.session_state.transcription,
                                       file_name  = "transcription.txt",
                                       mime       = "text/plain",
                                       help       = "Download your Transcription"
                                       )


        # Footer
        st.markdown("---")
        st.markdown("<p style='text-align:center;'>Developed with ❤️ by Priyam Pal [AI and Data Science Engineer] using OpenAI-Whisper</p>", 
                    unsafe_allow_html = True)

except Exception as e:
    app_logger.error(f"An error occurred: {repr(e)}")

    st.error("An error occurred. Please check the logs for more information.")