MDX-NET model:MDX23C-InstVoc HQ error

[chasswow0413]

Using Direct ML Version with GPU: Intel Arc Graphics, while the program came into a error "runtime error".

Last Error Received:

Process: MDX-Net

If this error persists, please contact the developers with the error details.

Raw Error Details:

RuntimeError: ""
Traceback Error: "
File "UVR.py", line 9217, in process_start
File "separate.py", line 681, in seperate
File "separate.py", line 960, in demix
File "torch\nn\modules\module.py", line 1501, in _call_impl
File "lib_v5\tfc_tdf_v3.py", line 238, in forward
File "torch\nn\modules\module.py", line 1501, in _call_impl
File "lib_v5\tfc_tdf_v3.py", line 154, in forward
File "torch\nn\modules\module.py", line 1501, in _call_impl
File "torch\nn\modules\container.py", line 217, in forward
File "torch\nn\modules\module.py", line 1501, in _call_impl
File "torch\nn\modules\instancenorm.py", line 74, in forward
File "torch\nn\modules\instancenorm.py", line 34, in _apply_instance_norm
File "torch\nn\functional.py", line 2495, in instance_norm
"

Error Time Stamp [2025-05-10 16:25:27]

Full Application Settings:

vr_model: Choose Model
aggression_setting: 5
window_size: 512
mdx_segment_size: Default
batch_size: Default
crop_size: 256
is_tta: False
is_output_image: False
is_post_process: False
is_high_end_process: False
post_process_threshold: 0.2
vr_voc_inst_secondary_model: No Model Selected
vr_other_secondary_model: No Model Selected
vr_bass_secondary_model: No Model Selected
vr_drums_secondary_model: No Model Selected
vr_is_secondary_model_activate: False
vr_voc_inst_secondary_model_scale: 0.9
vr_other_secondary_model_scale: 0.7
vr_bass_secondary_model_scale: 0.5
vr_drums_secondary_model_scale: 0.5
demucs_model: Choose Model
segment: Default
overlap: 0.25
overlap_mdx: Default
overlap_mdx23: 2
shifts: 2
chunks_demucs: Auto
margin_demucs: 44100
is_chunk_demucs: False
is_chunk_mdxnet: False
is_primary_stem_only_Demucs: False
is_secondary_stem_only_Demucs: False
is_split_mode: True
is_demucs_combine_stems: True
is_mdx23_combine_stems: True
demucs_voc_inst_secondary_model: No Model Selected
demucs_other_secondary_model: No Model Selected
demucs_bass_secondary_model: No Model Selected
demucs_drums_secondary_model: No Model Selected
demucs_is_secondary_model_activate: False
demucs_voc_inst_secondary_model_scale: 0.9
demucs_other_secondary_model_scale: 0.7
demucs_bass_secondary_model_scale: 0.5
demucs_drums_secondary_model_scale: 0.5
demucs_pre_proc_model: No Model Selected
is_demucs_pre_proc_model_activate: False
is_demucs_pre_proc_model_inst_mix: False
mdx_net_model: MDX23C-InstVoc HQ
chunks: Auto
margin: 44100
compensate: Auto
denoise_option: None
is_match_frequency_pitch: True
phase_option: Automatic
phase_shifts: None
is_save_align: False
is_match_silence: True
is_spec_match: False
is_mdx_c_seg_def: True
is_use_torch_inference_mode: False
is_invert_spec: False
is_deverb_vocals: False
deverb_vocal_opt: Main Vocals Only
voc_split_save_opt: Lead Only
is_mixer_mode: False
mdx_batch_size: Default
mdx_voc_inst_secondary_model: No Model Selected
mdx_other_secondary_model: No Model Selected
mdx_bass_secondary_model: No Model Selected
mdx_drums_secondary_model: No Model Selected
mdx_is_secondary_model_activate: False
mdx_voc_inst_secondary_model_scale: 0.9
mdx_other_secondary_model_scale: 0.7
mdx_bass_secondary_model_scale: 0.5
mdx_drums_secondary_model_scale: 0.5
is_save_all_outputs_ensemble: True
is_append_ensemble_name: False
chosen_audio_tool: Manual Ensemble
choose_algorithm: Min Spec
time_stretch_rate: 2.0
pitch_rate: 2.0
is_time_correction: True
is_gpu_conversion: True
is_primary_stem_only: False
is_secondary_stem_only: False
is_testing_audio: False
is_auto_update_model_params: True
is_add_model_name: False
is_accept_any_input: False
is_save_to_input_path: False
apollo_overlap: 2
apollo_chunk_size: 5
apollo_model: Choose Model
is_task_complete: False
is_normalization: False
is_use_directml: True
is_wav_ensemble: False
is_create_model_folder: False
mp3_bit_set: 320k
semitone_shift: 0
save_format: WAV
wav_type_set: PCM_16
device_set: Default
help_hints_var: True
set_vocal_splitter: No Model Selected
is_set_vocal_splitter: False
is_save_inst_set_vocal_splitter: False
model_sample_mode: False
model_sample_mode_duration: 30
demucs_stems: All Stems
mdx_stems: Vocals
Patch Version: UVR_Patch_1_15_25_22_30_BETA

[deton24]

Make sure you have the latest GPU drivers and/or Windows.
If still the same, copy DirectML.dll 1.9.1.0 to UVR\torch_directml folder, or search for the one attached with your GPU drivers installed in your system. It will increase noise from MDX-Net v2 (HQ and inst1-3, Kim inst) models which won't be fully get rid by denoise standard and denoise model options.