app.py

import gradio as gr
import torch
from utils.speech_processor import SpeechProcessor
from utils.text_processor import TextProcessor
from utils.output_generator import OutputGenerator
import tempfile
import os

# Initialize processors
speech_processor = SpeechProcessor()
text_processor = TextProcessor()
output_generator = OutputGenerator()

def process_meeting(audio_file, language="id", summary_ratio=0.3):
    """
    Main pipeline untuk memproses audio meeting
    """
    try:
        # Step 1: Speech Processing
        gr.Info("🎤 Memproses audio...")
        transcript_with_speakers = speech_processor.process_audio(
            audio_file, 
            language=language
        )
        
        # Step 2: Text Processing & Summarization
        gr.Info("📝 Membuat ringkasan...")
        summary = text_processor.summarize_transcript(
            transcript_with_speakers, 
            ratio=summary_ratio
        )
        
        # Step 3: Information Extraction
        gr.Info("🔍 Mengekstrak informasi penting...")
        extracted_info = text_processor.extract_key_information(
            transcript_with_speakers
        )
        
        # Step 4: Generate Output
        gr.Info("📄 Membuat notulensi...")
        outputs = output_generator.generate_all_formats(
            transcript_with_speakers,
            summary,
            extracted_info
        )
        
        return (
            outputs['markdown'],
            outputs['json'],
            outputs['transcript_table'],
            outputs['action_items_table'],
            outputs['decisions_table']
        )
        
    except Exception as e:
        gr.Error(f"Error: {str(e)}")
        return None, None, None, None, None

# Gradio Interface
with gr.Blocks(title="🤖 AI Meeting Minutes Generator") as demo:
    gr.Markdown("""
    # 🤖 AI Meeting Minutes Generator
    
    Upload audio rapat Anda dan dapatkan notulensi otomatis dengan:
    - 🎯 Identifikasi pembicara
    - 📝 Ringkasan otomatis
    - ✅ Action items
    - 📊 Keputusan penting
    """)
    
    with gr.Row():
        with gr.Column():
            audio_input = gr.Audio(
                label="Upload Audio Rapat",
                type="filepath",
                sources=["upload", "microphone"]
            )
            
            with gr.Row():
                language = gr.Dropdown(
                    choices=[
                        ("Indonesia", "id"),
                        ("English", "en")
                    ],
                    value="id",
                    label="Bahasa"
                )
                
                summary_ratio = gr.Slider(
                    minimum=0.1,
                    maximum=0.5,
                    value=0.3,
                    step=0.05,
                    label="Rasio Ringkasan"
                )
            
            process_btn = gr.Button("🚀 Proses Audio", variant="primary")
    
    with gr.Row():
        with gr.Column():
            gr.Markdown("### 📄 Notulensi (Markdown)")
            markdown_output = gr.Textbox(
                label="Preview Notulensi",
                lines=20,
                max_lines=30
            )
            
            json_download = gr.File(
                label="📥 Download JSON"
            )
    
    with gr.Row():
        with gr.Column():
            gr.Markdown("### 📊 Transkrip Lengkap")
            transcript_table = gr.Dataframe(
                headers=["Waktu", "Pembicara", "Teks"],
                label="Transkrip dengan Pembicara"
            )
    
    with gr.Row():
        with gr.Column():
            gr.Markdown("### ✅ Action Items")
            action_items_table = gr.Dataframe(
                headers=["Action Item", "Penanggung Jawab", "Timestamp"],
                label="Daftar Action Items"
            )
        
        with gr.Column():
            gr.Markdown("### 📌 Keputusan")
            decisions_table = gr.Dataframe(
                headers=["Keputusan", "Pembicara", "Timestamp"],
                label="Daftar Keputusan"
            )
    
    # Process button action
    process_btn.click(
        fn=process_meeting,
        inputs=[audio_input, language, summary_ratio],
        outputs=[
            markdown_output,
            json_download,
            transcript_table,
            action_items_table,
            decisions_table
        ]
    )
    
    # Examples
    gr.Examples(
        examples=[
            ["examples/meeting_sample_id.wav", "id", 0.3],
            ["examples/meeting_sample_en.wav", "en", 0.25]
        ],
        inputs=[audio_input, language, summary_ratio]
    )

if __name__ == "__main__":
    demo.launch()