app.py

import io
from typing import List
import gradio as gr
from pydub import AudioSegment
from gtts import gTTS

# ---------------------------
# Helper functions
# ---------------------------

def load_audio(file_obj) -> AudioSegment:
    """Load audio from file-like object or path, any ffmpeg-supported format."""
    # If file_obj has read() method (BytesIO or UploadedFile)
    if hasattr(file_obj, "read"):
        file_bytes = io.BytesIO(file_obj.read())
        file_bytes.seek(0)
    else:
        # fallback if already bytes
        file_bytes = io.BytesIO(file_obj)
    
    # Determine extension
    name = getattr(file_obj, "name", "audio.mp3")
    ext = name.rsplit(".", 1)[1].lower() if "." in name else "mp3"
    if ext == "mpg":
        ext = "mpeg"

    file_bytes.seek(0)
    return AudioSegment.from_file(file_bytes, format=ext)

def replace_vocals(original: AudioSegment, new_text: str, lang='en') -> AudioSegment:
    """Generate speech from text and overlay on original audio."""
    tts = gTTS(new_text, lang=lang)
    tts_io = io.BytesIO()
    tts.save(tts_io)
    tts_io.seek(0)
    voice = AudioSegment.from_file(tts_io, format="mp3")
    return original.overlay(voice)

def edit_audio(files: List, new_texts: List[str], output_format: str):
    """Edit multiple audio files with optional vocal replacement."""
    results = []
    for i, f in enumerate(files):
        try:
            audio = load_audio(f)
        except Exception as e:
            return None, f"Failed to load file #{i+1}: {e}"

        if i < len(new_texts) and new_texts[i].strip():
            audio = replace_vocals(audio, new_texts[i])

        out_io = io.BytesIO()
        fmt = output_format.lower() if output_format.lower() in ["mp3","wav","ogg","flac"] else "mp3"
        audio.export(out_io, format=fmt)
        out_io.seek(0)
        results.append((out_io, f"edited_track_{i+1}.{fmt}"))
    return results, None

# ---------------------------
# Gradio UI
# ---------------------------

def build_ui():
    with gr.Blocks() as demo:
        gr.Markdown("# 🎵 Audio Editor with Vocal Replacement")

        files = gr.File(
            label="Upload audio files (any ffmpeg-supported format)", 
            file_count="multiple", 
            file_types=None
        )
        output_format = gr.Dropdown(
            label="Output format", 
            choices=["mp3","wav","ogg","flac"], 
            value="mp3"
        )

        textboxes_box = gr.Column()
        make_btn = gr.Button("Edit Audio")
        status = gr.Textbox(label="Status", interactive=False)
        out_files = gr.File(label="Download Edited Files", file_types=None, interactive=False)

        # Dynamically create textboxes for each uploaded track
        def create_text_boxes(files_list):
            textboxes_box.clear()
            if not files_list:
                return
            for i, f in enumerate(files_list):
                tb = gr.Textbox(
                    label=f"New lyrics / words for Track {i+1}", 
                    placeholder="Leave empty to keep original"
                )
                textboxes_box.append(tb)

        files.change(create_text_boxes, inputs=files, outputs=None)

        # Collect all textbox values dynamically
        def on_edit(files_list, output_format_in, *textbox_values):
            if not files_list:
                return None, "Please upload audio files."
            results, err = edit_audio(files_list, list(textbox_values), output_format_in)
            if err:
                return None, err
            return results, "Edited audio ready!"

        make_btn.click(
            on_edit,
            inputs=[files, output_format] + textboxes_box.children,  # dynamic textboxes
            outputs=[out_files, status]
        )

    return demo

if __name__=="__main__":
    app = build_ui()
    app.launch(debug=True, share=True)