vocal_separation.py

import gradio as gr
from spleeter.separator import Separator
import os

# Initialize Spleeter with the 5stems model (for vocals, harmony, adlibs, etc.)
separator = Separator('spleeter:5stems')

# Function to separate the audio into vocals, harmony, ad-libs, and accompaniment
def separate_vocals(input_audio):
    output_dir = "output"
    if not os.path.exists(output_dir):
        os.makedirs(output_dir)
    
    # Path for the uploaded audio file
    audio_path = input_audio.name
    separated_path = os.path.join(output_dir, 'separated')

    # Separate the audio file
    separator.separate(audio_path)
    
    # Paths for the separated tracks
    main_vocal_file = os.path.join(separated_path, 'vocals.wav')
    harmony_vocal_file = os.path.join(separated_path, 'harmonics.wav')
    adlib_vocal_file = os.path.join(separated_path, 'adlibs.wav')
    accompaniment_file = os.path.join(separated_path, 'accompaniment.wav')
    
    # Return the separated files
    return main_vocal_file, harmony_vocal_file, adlib_vocal_file, accompaniment_file

# Gradio interface setup
def create_interface():
    with gr.Blocks() as demo:
        with gr.Row():
            audio_input = gr.Audio(label="Upload Audio File", type="file")
            main_vocal_output = gr.Audio(label="Main Vocals", type="file", interactive=False)
            harmony_vocal_output = gr.Audio(label="Harmony Vocals", type="file", interactive=False)
            adlib_vocal_output = gr.Audio(label="Ad-lib Vocals", type="file", interactive=False)
            accompaniment_output = gr.Audio(label="Accompaniment", type="file", interactive=False)
        
        # When the audio file is uploaded, separate the audio and return the files
        audio_input.change(separate_vocals, inputs=audio_input, outputs=[main_vocal_output, harmony_vocal_output, adlib_vocal_output, accompaniment_output])
    
    demo.launch()

# Run the Gradio interface
create_interface()