app.py

import gradio as gr
from audio_separator.separator import Separator

separator = Separator()

roformer_models = {
        'BS-Roformer-Viperx-1297.ckpt': 'model_bs_roformer_ep_317_sdr_12.9755.ckpt',
        'BS-Roformer-Viperx-1296.ckpt': 'model_bs_roformer_ep_368_sdr_12.9628.ckpt',
        'BS-Roformer-Viperx-1053.ckpt': 'model_bs_roformer_ep_937_sdr_10.5309.ckpt',
        'Mel-Roformer-Viperx-1143.ckpt': 'model_mel_band_roformer_ep_3005_sdr_11.4360.ckpt'
}

mdx23c_models = [
    'MDX23C_D1581.ckpt',
    'MDX23C-8KFFT-InstVoc_HQ.ckpt',
    'MDX23C-8KFFT-InstVoc_HQ_2.ckpt',
]

mdxnet_models = [
    'UVR-MDX-NET-Inst_full_292.onnx',
    'UVR-MDX-NET_Inst_187_beta.onnx',
    'UVR-MDX-NET_Inst_82_beta.onnx',
    'UVR-MDX-NET_Inst_90_beta.onnx',
    'UVR-MDX-NET_Main_340.onnx',
    'UVR-MDX-NET_Main_390.onnx',
    'UVR-MDX-NET_Main_406.onnx',
    'UVR-MDX-NET_Main_427.onnx',
    'UVR-MDX-NET_Main_438.onnx',
    'UVR-MDX-NET-Inst_HQ_1.onnx',
    'UVR-MDX-NET-Inst_HQ_2.onnx',
    'UVR-MDX-NET-Inst_HQ_3.onnx',
    'UVR-MDX-NET-Inst_HQ_4.onnx',
    'UVR_MDXNET_Main.onnx',
    'UVR-MDX-NET-Inst_Main.onnx',
    'UVR_MDXNET_1_9703.onnx',
    'UVR_MDXNET_2_9682.onnx',
    'UVR_MDXNET_3_9662.onnx',
    'UVR-MDX-NET-Inst_1.onnx',
    'UVR-MDX-NET-Inst_2.onnx',
    'UVR-MDX-NET-Inst_3.onnx',
    'UVR_MDXNET_KARA.onnx',
    'UVR_MDXNET_KARA_2.onnx',
    'UVR_MDXNET_9482.onnx',
    'UVR-MDX-NET-Voc_FT.onnx',
    'Kim_Vocal_1.onnx',
    'Kim_Vocal_2.onnx',
    'Kim_Inst.onnx',
    'Reverb_HQ_By_FoxJoy.onnx',
    'UVR-MDX-NET_Crowd_HQ_1.onnx',
    'kuielab_a_vocals.onnx',
    'kuielab_a_other.onnx',
    'kuielab_a_bass.onnx',
    'kuielab_a_drums.onnx',
    'kuielab_b_vocals.onnx',
    'kuielab_b_other.onnx',
    'kuielab_b_bass.onnx',
    'kuielab_b_drums.onnx',
]

vrarch_models = [
    '1_HP-UVR.pth',
    '2_HP-UVR.pth',
    '3_HP-Vocal-UVR.pth',
    '4_HP-Vocal-UVR.pth',
    '5_HP-Karaoke-UVR.pth',
    '6_HP-Karaoke-UVR.pth',
    '7_HP2-UVR.pth',
    '8_HP2-UVR.pth',
    '9_HP2-UVR.pth',
    '10_SP-UVR-2B-32000-1.pth',
    '11_SP-UVR-2B-32000-2.pth',
    '12_SP-UVR-3B-44100.pth',
    '13_SP-UVR-4B-44100-1.pth',
    '14_SP-UVR-4B-44100-2.pth',
    '15_SP-UVR-MID-44100-1.pth',
    '16_SP-UVR-MID-44100-2.pth',
    '17_HP-Wind_Inst-UVR.pth',
    'UVR-De-Echo-Aggressive.pth',
    'UVR-De-Echo-Normal.pth',
    'UVR-DeEcho-DeReverb.pth',
    'UVR-DeNoise-Lite.pth',
    'UVR-DeNoise.pth',
    'UVR-BVE-4B_SN-44100-1.pth',
    'MGM_HIGHEND_v4.pth',
    'MGM_LOWEND_A_v4.pth',
    'MGM_LOWEND_B_v4.pth',
    'MGM_MAIN_v4.pth',
]

demucs_models = [
    'htdemucs_ft.yaml',
    'htdemucs.yaml',
    'hdemucs_mmi.yaml',
]

def roformer_separator(audio, checkpoint_name):
    full_checkpoint_name = roformer_models[checkpoint_name]
    separator.load_model(full_checkpoint_name)
    output_files = separator.separate(audio)
    stem1 = output_files[0]
    stem2 = output_files[1]
    return stem1, stem2

def mdx_vr_separator(audio, checkpoint_name):
    separator.load_model(checkpoint_name)
    output_files = separator.separate(audio)
    stem1 = output_files[0]
    stem2 = output_files[1]
    return stem1, stem2

def demucs_separator(audio, checkpoint_name):
    separator.load_model(checkpoint_name)
    output_files = separator.separate(audio)
    stem1 = output_files[0]
    stem2 = output_files[1]
    stem3 = output_files[2]
    stem4 = output_files[3]
    return stem1, stem2, stem3, stem4

with gr.Blocks(title="🎵 UVR5 UI 🎵") as demo:
    gr.Markdown("<h1> 🎵 UVR5 UI 🎵 </h1>")
    with gr.Tab("Vocal Separator (UVR)"):
        gr.Markdown("Separate vocals and instruments from an audio file using UVR models.")
        with gr.Tab("Mel/BS Roformer"):
            roformer_audio_file = gr.Audio(label="Audio File", type="filepath")
            with gr.Row():
                roformer_model = gr.Dropdown(label="Model", choices=list(roformer_models.keys()))
                roformer_button = gr.Button("Separate", variant="primary")
            roformer_stem1 = gr.Audio(type="filepath", label="Stem 1")
            roformer_stem2 = gr.Audio(type="filepath", label="Stem 2")

            roformer_button.click(roformer_separator, [roformer_audio_file, roformer_model], [roformer_stem1, roformer_stem2])
        with gr.Tab("MDX23C"):
            mdx23c_audio_file = gr.Audio(label="Audio File", type="filepath")
            with gr.Row():
                mdx23c_model = gr.Dropdown(label="Model", choices=mdx23c_models)
                mdx23c_button = gr.Button("Separate", variant="primary")
            mdx23c_stem1 = gr.Audio(type="filepath", label="Stem 1")
            mdx23c_stem2 = gr.Audio(type="filepath", label="Stem 2")

            mdx23c_button.click(mdx_vr_separator, [mdx23c_audio_file, mdx23c_model], [mdx23c_stem1, mdx23c_stem2])
        with gr.Tab("MDX-NET"):
            mdxnet_audio_file = gr.Audio(label="Audio File", type="filepath")
            with gr.Row():
                mdxnet_model = gr.Dropdown(label="Model", choices=mdxnet_models)
                mdxnet_button = gr.Button("Separate", variant="primary")
            mdxnet_stem1 = gr.Audio(type="filepath", label="Stem 1")
            mdxnet_stem2 = gr.Audio(type="filepath", label="Stem 2")

            mdxnet_button.click(mdx_vr_separator, [mdxnet_audio_file, mdxnet_model], [mdxnet_stem1, mdxnet_stem2])
        with gr.Tab("VR-ARCH"):
            vr_audio_file = gr.Audio(label="Audio File", type="filepath")
            with gr.Row():
                vr_model = gr.Dropdown(label="Model", choices=vrarch_models)
                vr_button = gr.Button("Separate", variant="primary")
            vr_stem1 = gr.Audio(type="filepath", label="Stem 1")
            vr_stem2 = gr.Audio(type="filepath", label="Stem 2")

            vr_button.click(mdx_vr_separator, [vr_audio_file, vr_model], [vr_stem1, vr_stem2])
        with gr.Tab("Demucs"):
            demucs_audio_file = gr.Audio(label="Audio File", type="filepath")
            with gr.Row():
                demucs_model = gr.Dropdown(label="Model", choices=demucs_models)
                demucs_button = gr.Button("Separate", variant="primary")
            demucs_stem1 = gr.Audio(type="filepath", label="Stem 1")
            demucs_stem2 = gr.Audio(type="filepath", label="Stem 2")
            demucs_stem3 = gr.Audio(type="filepath", label="Stem 3")
            demucs_stem4 = gr.Audio(type="filepath", label="Stem 4")

            demucs_button.click(demucs_separator, [demucs_audio_file, demucs_model], [demucs_stem1, demucs_stem2, demucs_stem3, demucs_stem4])

demo.launch()