import os import torch import shutil import logging import gradio as gr from audio_separator.separator import Separator device = "cuda" if torch.cuda.is_available() else "cpu" use_autocast = device == "cuda" #=========================# # Roformer Models # #=========================# ROFORMER_MODELS = { 'BS-Roformer-De-Reverb': 'deverb_bs_roformer_8_384dim_10depth.ckpt', 'BS-Roformer-Viperx-1053': 'model_bs_roformer_ep_937_sdr_10.5309.ckpt', 'BS-Roformer-Viperx-1296': 'model_bs_roformer_ep_368_sdr_12.9628.ckpt', 'BS-Roformer-Viperx-1297': 'model_bs_roformer_ep_317_sdr_12.9755.ckpt', 'Mel-Roformer-Crowd-Aufr33-Viperx': 'mel_band_roformer_crowd_aufr33_viperx_sdr_8.7144.ckpt', 'Mel-Roformer-Denoise-Aufr33': 'denoise_mel_band_roformer_aufr33_sdr_27.9959.ckpt', 'Mel-Roformer-Denoise-Aufr33-Aggr': 'denoise_mel_band_roformer_aufr33_aggr_sdr_27.9768.ckpt', 'Mel-Roformer-Karaoke-Aufr33-Viperx': 'mel_band_roformer_karaoke_aufr33_viperx_sdr_10.1956.ckpt', 'Mel-Roformer-Viperx-1143': 'model_mel_band_roformer_ep_3005_sdr_11.4360.ckpt', 'MelBand Roformer Kim | Inst V1 by Unwa': 'melband_roformer_inst_v1.ckpt', 'MelBand Roformer Kim | Inst V2 by Unwa': 'melband_roformer_inst_v2.ckpt', 'MelBand Roformer Kim | InstVoc Duality V1 by Unwa': 'melband_roformer_instvoc_duality_v1.ckpt', 'MelBand Roformer Kim | InstVoc Duality V2 by Unwa': 'melband_roformer_instvox_duality_v2.ckpt', 'Vocals Mel Band Roformer': 'vocals_mel_band_roformer.ckpt', 'Mel Band Roformer Bleed Suppressor V1': 'mel_band_roformer_bleed_suppressor_v1.ckpt', 'Mel Band Roformer SYHFT V2': 'MelBandRoformerSYHFTV2.ckpt', 'Mel Band Roformer SYHFT V2.5': 'MelBandRoformerSYHFTV2.5.ckpt', } #=========================# # MDX23C Models # #=========================# MDX23C_MODELS = [ 'MDX23C-8KFFT-InstVoc_HQ.ckpt', 'MDX23C-8KFFT-InstVoc_HQ_2.ckpt', 'MDX23C_D1581.ckpt', ] #=========================# # MDXN-NET Models # #=========================# MDXNET_MODELS = [ 'UVR-MDX-NET-Crowd_HQ_1.onnx', 'UVR-MDX-NET-Inst_1.onnx', 'UVR-MDX-NET-Inst_2.onnx', 'UVR-MDX-NET-Inst_3.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-MDX-NET-Inst_HQ_5.onnx', 'UVR-MDX-NET-Inst_full_292.onnx', 'UVR-MDX-NET-Voc_FT.onnx', 'UVR-MDX-NET_Inst_82_beta.onnx', 'UVR-MDX-NET_Inst_90_beta.onnx', 'UVR-MDX-NET_Inst_187_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_MDXNET_1_9703.onnx', 'UVR_MDXNET_2_9682.onnx', 'UVR_MDXNET_3_9662.onnx', 'UVR_MDXNET_9482.onnx', 'UVR_MDXNET_KARA.onnx', 'UVR_MDXNET_KARA_2.onnx', 'UVR_MDXNET_Main.onnx', 'kuielab_a_bass.onnx', 'kuielab_a_drums.onnx', 'kuielab_a_other.onnx', 'kuielab_a_vocals.onnx', 'kuielab_b_bass.onnx', 'kuielab_b_drums.onnx', 'kuielab_b_other.onnx', 'kuielab_b_vocals.onnx', 'Kim_Inst.onnx', 'Kim_Vocal_1.onnx', 'Kim_Vocal_2.onnx', 'Reverb_HQ_By_FoxJoy.onnx', ] #========================# # VR-ARCH Models # #========================# VR_ARCH_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', 'MGM_HIGHEND_v4.pth', 'MGM_LOWEND_A_v4.pth', 'MGM_LOWEND_B_v4.pth', 'MGM_MAIN_v4.pth', 'UVR-BVE-4B_SN-44100-1.pth', 'UVR-DeEcho-DeReverb.pth', 'UVR-De-Echo-Aggressive.pth', 'UVR-De-Echo-Normal.pth', 'UVR-DeNoise-Lite.pth', 'UVR-DeNoise.pth', ] #=======================# # DEMUCS Models # #=======================# DEMUCS_MODELS = [ 'hdemucs_mmi.yaml', 'htdemucs.yaml', 'htdemucs_6s.yaml', 'htdemucs_ft.yaml', ] def print_message(input_file, model_name): """Prints information about the audio separation process.""" base_name = os.path.splitext(os.path.basename(input_file))[0] print("\n") print("🎵 Audio-Separator 🎵") print("Input audio:", base_name) print("Separation Model:", model_name) print("Audio Separation Process...") def prepare_output_dir(input_file, output_dir): """Create a directory for the output files and clean it if it already exists.""" base_name = os.path.splitext(os.path.basename(input_file))[0] out_dir = os.path.join(output_dir, base_name) try: if os.path.exists(out_dir): shutil.rmtree(out_dir) os.makedirs(out_dir) except Exception as e: raise RuntimeError(f"Failed to prepare output directory {out_dir}: {e}") return out_dir def roformer_separator(audio, model_key, seg_size, override_seg_size, overlap, pitch_shift, model_dir, out_dir, out_format, norm_thresh, amp_thresh, batch_size, progress=gr.Progress()): """Separate audio using Roformer model.""" base_name = os.path.splitext(os.path.basename(audio))[0] print_message(audio, model_key) model = ROFORMER_MODELS[model_key] try: out_dir = prepare_output_dir(audio, out_dir) separator = Separator( log_level=logging.WARNING, model_file_dir=model_dir, output_dir=out_dir, output_format=out_format, normalization_threshold=norm_thresh, amplification_threshold=amp_thresh, use_autocast=use_autocast, mdxc_params={ "segment_size": seg_size, "override_model_segment_size": override_seg_size, "batch_size": batch_size, "overlap": overlap, "pitch_shift": pitch_shift, } ) progress(0.2, desc="Model loaded...") separator.load_model(model_filename=model) progress(0.7, desc="Audio separated...") separation = separator.separate(audio, f"{base_name}_(Stem1)", f"{base_name}_(Stem2)") print(f"Separation complete!\nResults: {', '.join(separation)}") stems = [os.path.join(out_dir, file_name) for file_name in separation] return stems[1], stems[0] except Exception as e: raise RuntimeError(f"Roformer separation failed: {e}") from e def mdx23c_separator(audio, model, seg_size, override_seg_size, overlap, pitch_shift, model_dir, out_dir, out_format, norm_thresh, amp_thresh, batch_size, progress=gr.Progress(track_tqdm=True)): """Separate audio using MDX23C model.""" base_name = os.path.splitext(os.path.basename(audio))[0] print_message(audio, model) try: out_dir = prepare_output_dir(audio, out_dir) separator = Separator( log_level=logging.WARNING, model_file_dir=model_dir, output_dir=out_dir, output_format=out_format, normalization_threshold=norm_thresh, amplification_threshold=amp_thresh, use_autocast=use_autocast, mdxc_params={ "segment_size": seg_size, "override_model_segment_size": override_seg_size, "batch_size": batch_size, "overlap": overlap, "pitch_shift": pitch_shift, } ) progress(0.2, desc="Model loaded...") separator.load_model(model_filename=model) progress(0.7, desc="Audio separated...") separation = separator.separate(audio, f"{base_name}_(Stem1)", f"{base_name}_(Stem2)") print(f"Separation complete!\nResults: {', '.join(separation)}") stems = [os.path.join(out_dir, file_name) for file_name in separation] return stems[1], stems[0] except Exception as e: raise RuntimeError(f"MDX23C separation failed: {e}") from e def mdx_separator(audio, model, hop_length, seg_size, overlap, denoise, model_dir, out_dir, out_format, norm_thresh, amp_thresh, batch_size, progress=gr.Progress()): """Separate audio using MDX-NET model.""" base_name = os.path.splitext(os.path.basename(audio))[0] print_message(audio, model) try: out_dir = prepare_output_dir(audio, out_dir) separator = Separator( log_level=logging.WARNING, model_file_dir=model_dir, output_dir=out_dir, output_format=out_format, normalization_threshold=norm_thresh, amplification_threshold=amp_thresh, use_autocast=use_autocast, mdx_params={ "hop_length": hop_length, "segment_size": seg_size, "overlap": overlap, "batch_size": batch_size, "enable_denoise": denoise, } ) progress(0.2, desc="Model loaded...") separator.load_model(model_filename=model) progress(0.7, desc="Audio separated...") separation = separator.separate(audio, f"{base_name}_(Stem1)", f"{base_name}_(Stem2)") print(f"Separation complete!\nResults: {', '.join(separation)}") stems = [os.path.join(out_dir, file_name) for file_name in separation] return stems[0], stems[1] except Exception as e: raise RuntimeError(f"MDX-NET separation failed: {e}") from e def vr_separator(audio, model, window_size, aggression, tta, post_process, post_process_threshold, high_end_process, model_dir, out_dir, out_format, norm_thresh, amp_thresh, batch_size, progress=gr.Progress()): """Separate audio using VR ARCH model.""" base_name = os.path.splitext(os.path.basename(audio))[0] print_message(audio, model) try: out_dir = prepare_output_dir(audio, out_dir) separator = Separator( log_level=logging.WARNING, model_file_dir=model_dir, output_dir=out_dir, output_format=out_format, normalization_threshold=norm_thresh, amplification_threshold=amp_thresh, use_autocast=use_autocast, vr_params={ "batch_size": batch_size, "window_size": window_size, "aggression": aggression, "enable_tta": tta, "enable_post_process": post_process, "post_process_threshold": post_process_threshold, "high_end_process": high_end_process, } ) progress(0.2, desc="Model loaded...") separator.load_model(model_filename=model) progress(0.7, desc="Audio separated...") separation = separator.separate(audio, f"{base_name}_(Stem1)", f"{base_name}_(Stem2)") print(f"Separation complete!\nResults: {', '.join(separation)}") stems = [os.path.join(out_dir, file_name) for file_name in separation] return stems[0], stems[1] except Exception as e: raise RuntimeError(f"VR ARCH separation failed: {e}") from e def demucs_separator(audio, model, seg_size, shifts, overlap, segments_enabled, model_dir, out_dir, out_format, norm_thresh, amp_thresh, progress=gr.Progress()): """Separate audio using Demucs model.""" print_message(audio, model) try: out_dir = prepare_output_dir(audio, out_dir) separator = Separator( log_level=logging.WARNING, model_file_dir=model_dir, output_dir=out_dir, output_format=out_format, normalization_threshold=norm_thresh, amplification_threshold=amp_thresh, use_autocast=use_autocast, demucs_params={ "segment_size": seg_size, "shifts": shifts, "overlap": overlap, "segments_enabled": segments_enabled, } ) progress(0.2, desc="Model loaded...") separator.load_model(model_filename=model) progress(0.7, desc="Audio separated...") separation = separator.separate(audio) print(f"Separation complete!\nResults: {', '.join(separation)}") stems = [os.path.join(out_dir, file_name) for file_name in separation] if model == "htdemucs_6s.yaml": return stems[0], stems[1], stems[2], stems[3], stems[4], stems[5] else: return stems[0], stems[1], stems[2], stems[3], None, None except Exception as e: raise RuntimeError(f"Demucs separation failed: {e}") from e def update_stems(model): if model == "htdemucs_6s.yaml": return gr.update(visible=True) else: return gr.update(visible=False) with gr.Blocks(title="🎵 Audio-Separator 🎵",theme=gr.themes.Base()) as app: gr.HTML("

🎵 Audio-Separator 🎵

") with gr.Tab("Roformer"): with gr.Group(): with gr.Row(): roformer_model = gr.Dropdown(label="Select the Model", choices=list(ROFORMER_MODELS.keys())) with gr.Row(): with gr.Row(): roformer_seg_size = gr.Slider(minimum=32, maximum=4000, step=32, value=256, label="Segment Size", info="Larger consumes more resources, but may give better results.") roformer_override_seg_size = gr.Checkbox(value=False, label="Override segment size", info="Override model default segment size instead of using the model default value.") with gr.Row(): roformer_overlap = gr.Slider(minimum=2, maximum=10, step=1, value=8, label="Overlap", info="Amount of overlap between prediction windows. Lower is better but slower.") roformer_pitch_shift = gr.Slider(minimum=-12, maximum=12, step=1, value=0, label="Pitch shift", info="Shift audio pitch by a number of semitones while processing. may improve output for deep/high vocals.") with gr.Row(): roformer_audio = gr.Audio(label="Input Audio", type="filepath") with gr.Row(): roformer_button = gr.Button("Separate!", variant="primary") with gr.Row(): roformer_stem1 = gr.Audio(label="Stem 1", type="filepath", interactive=False) roformer_stem2 = gr.Audio(label="Stem 2", type="filepath", interactive=False) with gr.Tab("MDX23C"): with gr.Group(): with gr.Row(): mdx23c_model = gr.Dropdown(label="Select the Model", choices=MDX23C_MODELS) with gr.Row(): mdx23c_seg_size = gr.Slider(minimum=32, maximum=4000, step=32, value=256, label="Segment Size", info="Larger consumes more resources, but may give better results.") mdx23c_override_seg_size = gr.Checkbox(value=False, label="Override segment size", info="Override model default segment size instead of using the model default value.") with gr.Row(): mdx23c_overlap = gr.Slider(minimum=2, maximum=50, step=1, value=8, label="Overlap", info="Amount of overlap between prediction windows. Higher is better but slower.") mdx23c_pitch_shift = gr.Slider(minimum=-12, maximum=12, step=1, value=0, label="Pitch shift", info="Shift audio pitch by a number of semitones while processing. may improve output for deep/high vocals.") with gr.Row(): mdx23c_audio = gr.Audio(label="Input Audio", type="filepath") with gr.Row(): mdx23c_button = gr.Button("Separate!", variant="primary") with gr.Row(): mdx23c_stem1 = gr.Audio(label="Stem 1", type="filepath", interactive=False) mdx23c_stem2 = gr.Audio(label="Stem 2", type="filepath", interactive=False) with gr.Tab("MDX-NET"): with gr.Group(): with gr.Row(): mdx_model = gr.Dropdown(label="Select the Model", choices=MDXNET_MODELS) with gr.Row(): mdx_hop_length = gr.Slider(minimum=32, maximum=2048, step=32, value=1024, label="Hop Length", info="Usually called stride in neural networks; only change if you know what you're doing.") mdx_seg_size = gr.Slider(minimum=32, maximum=4000, step=32, value=256, label="Segment Size", info="Larger consumes more resources, but may give better results.") with gr.Row(): mdx_overlap = gr.Slider(minimum=0.001, maximum=0.999, step=0.001, value=0.25, label="Overlap", info="Amount of overlap between prediction windows. Higher is better but slower.") mdx_denoise = gr.Checkbox(value=False, label="Denoise", info="Enable denoising after separation.") with gr.Row(): mdx_audio = gr.Audio(label="Input Audio", type="filepath") with gr.Row(): mdx_button = gr.Button("Separate!", variant="primary") with gr.Row(): mdx_stem1 = gr.Audio(label="Stem 1", type="filepath", interactive=False) mdx_stem2 = gr.Audio(label="Stem 2", type="filepath", interactive=False) with gr.Tab("VR ARCH"): with gr.Group(): with gr.Row(): vr_model = gr.Dropdown(label="Select the Model", choices=VR_ARCH_MODELS) with gr.Row(): vr_window_size = gr.Slider(minimum=320, maximum=1024, step=32, value=512, label="Window Size", info="Balance quality and speed. 1024 = fast but lower, 320 = slower but better quality.") vr_aggression = gr.Slider(minimum=1, maximum=50, step=1, value=5, label="Agression", info="Intensity of primary stem extraction.") with gr.Row(): vr_tta = gr.Checkbox(value=False, label="TTA", info="Enable Test-Time-Augmentation; slow but improves quality.") with gr.Row(): vr_post_process = gr.Checkbox(value=False, label="Post Process", info="Identify leftover artifacts within vocal output; may improve separation for some songs.") vr_post_process_threshold = gr.Slider(minimum=0.1, maximum=0.3, step=0.1, value=0.2, label="Post Process Threshold", info="Threshold for post-processing.") vr_high_end_process = gr.Checkbox(value=False, label="High End Process", info="Mirror the missing frequency range of the output.") with gr.Row(): vr_audio = gr.Audio(label="Input Audio", type="filepath") with gr.Row(): vr_button = gr.Button("Separate!", variant="primary") with gr.Row(): vr_stem1 = gr.Audio(label="Stem 1", type="filepath", interactive=False) vr_stem2 = gr.Audio(label="Stem 2", type="filepath", interactive=False) with gr.Tab("Demucs"): with gr.Group(): with gr.Row(): demucs_model = gr.Dropdown(label="Select the Model", choices=DEMUCS_MODELS) with gr.Row(): demucs_seg_size = gr.Slider(minimum=1, maximum=100, step=1, value=40, label="Segment Size", info="Size of segments into which the audio is split. Higher = slower but better quality.") demucs_shifts = gr.Slider(minimum=0, maximum=20, step=1, value=2, label="Shifts", info="Number of predictions with random shifts, higher = slower but better quality.") demucs_overlap = gr.Slider(minimum=0.001, maximum=0.999, step=0.001, value=0.25, label="Overlap", info="Overlap between prediction windows. Higher = slower but better quality.") demucs_segments_enabled = gr.Checkbox(value=True, label="Segment-wise processing", info="Enable segment-wise processing.") with gr.Row(): demucs_audio = gr.Audio(label="Input Audio", type="filepath") with gr.Row(): demucs_button = gr.Button("Separate!", variant="primary") with gr.Row(): demucs_stem1 = gr.Audio(label="Stem 1", type="filepath", interactive=False) demucs_stem2 = gr.Audio(label="Stem 2", type="filepath", interactive=False) with gr.Row(): demucs_stem3 = gr.Audio(label="Stem 3", type="filepath", interactive=False) demucs_stem4 = gr.Audio(label="Stem 4", type="filepath", interactive=False) with gr.Row(visible=False) as stem6: demucs_stem5 = gr.Audio(label="Stem 5", type="filepath", interactive=False) demucs_stem6 = gr.Audio(label="Stem 6", type="filepath", interactive=False) with gr.Tab("General settings"): with gr.Group(): model_file_dir = gr.Textbox(value="/tmp/audio-separator-models/", label="Directory to cache model files", info="The directory where model files are stored.", placeholder="/tmp/audio-separator-models/") with gr.Row(): output_dir = gr.Textbox(value="output", label="File output directory", info="The directory where output files will be saved.", placeholder="output") output_format = gr.Dropdown(value="wav", choices=["wav", "flac", "mp3"], label="Output Format", info="The format of the output audio file.") with gr.Row(): norm_threshold = gr.Slider(minimum=0.1, maximum=1, step=0.1, value=0.9, label="Normalization threshold", info="The threshold for audio normalization.") amp_threshold = gr.Slider(minimum=0.1, maximum=1, step=0.1, value=0.6, label="Amplification threshold", info="The threshold for audio amplification.") with gr.Row(): batch_size = gr.Slider(minimum=1, maximum=16, step=1, value=1, label="Batch Size", info="Larger consumes more RAM but may process slightly faster.") with gr.Tab("Credits"): gr.Markdown(""" Politrees - gradio webui\n theNeodev - mod the ui\n nomadkaraoke - original project """) demucs_model.change(update_stems, inputs=[demucs_model], outputs=stem6) roformer_button.click( roformer_separator, inputs=[ roformer_audio, roformer_model, roformer_seg_size, roformer_override_seg_size, roformer_overlap, roformer_pitch_shift, model_file_dir, output_dir, output_format, norm_threshold, amp_threshold, batch_size, ], outputs=[roformer_stem1, roformer_stem2], ) mdx23c_button.click( mdx23c_separator, inputs=[ mdx23c_audio, mdx23c_model, mdx23c_seg_size, mdx23c_override_seg_size, mdx23c_overlap, mdx23c_pitch_shift, model_file_dir, output_dir, output_format, norm_threshold, amp_threshold, batch_size, ], outputs=[mdx23c_stem1, mdx23c_stem2], ) mdx_button.click( mdx_separator, inputs=[ mdx_audio, mdx_model, mdx_hop_length, mdx_seg_size, mdx_overlap, mdx_denoise, model_file_dir, output_dir, output_format, norm_threshold, amp_threshold, batch_size, ], outputs=[mdx_stem1, mdx_stem2], ) vr_button.click( vr_separator, inputs=[ vr_audio, vr_model, vr_window_size, vr_aggression, vr_tta, vr_post_process, vr_post_process_threshold, vr_high_end_process, model_file_dir, output_dir, output_format, norm_threshold, amp_threshold, batch_size, ], outputs=[vr_stem1, vr_stem2], ) demucs_button.click( demucs_separator, inputs=[ demucs_audio, demucs_model, demucs_seg_size, demucs_shifts, demucs_overlap, demucs_segments_enabled, model_file_dir, output_dir, output_format, norm_threshold, amp_threshold, ], outputs=[demucs_stem1, demucs_stem2, demucs_stem3, demucs_stem4, demucs_stem5, demucs_stem6], ) def main(): app.launch(share=True, debug=True) if __name__ == "__main__": main()