GPU conversion support for Intel and AMD on Linux version

[AhuracMusic]

I managed to run UVR on NixOS, which was kind of a big deal. Now I want to know if there is a way to use the GPU conversion feature when your GPU is not Nvidia on Linux. On Windows, I know there's a build using DirectML to provide this: we could try to work on a way to achieve this on Linux using another API like oneAPI or OpenCL.

[nPrevail]

Greetings,

I'm trying to figure out how to run UVR on NixOS. Do you mind sharing a few pointers on how you did it? Or a nix-shell script?

[AhuracMusic]

I pushed a branch on my personal forge with the Nix shell : https://git.ahur.ac/ahurac/ultimatevocalremovergui/src/branch/feat/nix-shell. Reach out if you have any trouble using it.

[nPrevail]

Thank you for sharing!

Building it in terminal so to speak, via nix-shell

[nPrevail]

When trying to use any of the demucs v4 models to split All Stems I encounter this error:

Last Error Received:

Process: Demucs

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

Raw Error Details:

UnpicklingError: "Weights only load failed. This file can still be loaded, to do so you have two options, �[1mdo those steps only if you trust the source of the checkpoint�[0m. 
	(1) In PyTorch 2.6, we changed the default value of the `weights_only` argument in `torch.load` from `False` to `True`. Re-running `torch.load` with `weights_only` set to `False` will likely succeed, but it can result in arbitrary code execution. Do it only if you got the file from a trusted source.
	(2) Alternatively, to load with `weights_only=True` please check the recommended steps in the following error message.
	WeightsUnpickler error: Unsupported global: GLOBAL demucs.htdemucs.HTDemucs was not an allowed global by default. Please use `torch.serialization.add_safe_globals([demucs.htdemucs.HTDemucs])` or the `torch.serialization.safe_globals([demucs.htdemucs.HTDemucs])` context manager to allowlist this global if you trust this class/function.

Check the documentation of torch.load to learn more about types accepted by default with weights_only https://pytorch.org/docs/stable/generated/torch.load.html."
Traceback Error: "
  File "/home/nprevail/ultimatevocalremovergui/UVR.py", line 6640, in process_start
    seperator.seperate()
  File "/home/nprevail/ultimatevocalremovergui/separate.py", line 834, in seperate
    self.demucs = _gm(name=os.path.splitext(os.path.basename(self.model_path))[0],
  File "/home/nprevail/ultimatevocalremovergui/demucs/pretrained.py", line 81, in get_model
    model = any_repo.get_model(name)
  File "/home/nprevail/ultimatevocalremovergui/demucs/repo.py", line 148, in get_model
    return self.bag_repo.get_model(name_or_sig)
  File "/home/nprevail/ultimatevocalremovergui/demucs/repo.py", line 129, in get_model
    models = [self.model_repo.get_model(sig) for sig in signatures]
  File "/home/nprevail/ultimatevocalremovergui/demucs/repo.py", line 129, in <listcomp>
    models = [self.model_repo.get_model(sig) for sig in signatures]
  File "/home/nprevail/ultimatevocalremovergui/demucs/repo.py", line 100, in get_model
    return load_model(file)
  File "/home/nprevail/ultimatevocalremovergui/demucs/states.py", line 46, in load_model
    package = torch.load(path, 'cpu')
  File "/home/nprevail/ultimatevocalremovergui/.venv/lib/python3.10/site-packages/torch/serialization.py", line 1548, in load
    raise pickle.UnpicklingError(_get_wo_message(str(e))) from None
"

Error Time Stamp [2026-03-12 14:06:50]

Full Application Settings:

vr_model: Choose Model
aggression_setting: 5
window_size: 512
mdx_segment_size: 256
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: v4 | htdemucs_ft
segment: Default
overlap: 0.25
overlap_mdx: Default
overlap_mdx23: 8
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: UVR-MDX-NET Main
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: 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: False
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_task_complete: False
is_normalization: False
is_use_opencl: False
is_wav_ensemble: False
is_create_model_folder: False
mp3_bit_set: 320k
semitone_shift: 0
save_format: FLAC
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: True
model_sample_mode_duration: 30
demucs_stems: All Stems
mdx_stems: All Stems

Similarly:

#2090

This was one person's solution:
#2090 (comment)

Any idea how to update UVR Roformer, or apply the solution to this issue?

[AhuracMusic]

Well, I have no take on fixing this issue... But you managed to run UVR at least?

[nPrevail]

Yup! I was able to run UVR.

…

On Thu, Mar 12, 2026 at 4:20 PM AhuracMusic ***@***.***> wrote: Well, I have no take on fixing this issue... But you managed to run UVR at least? — Reply to this email directly, view it on GitHub <#2095 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ANYETW4TVGOHLJZUY7ZGA3T4QNA3FAVCNFSM6AAAAACQYPK5DCVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTMMJQGU3DQOA> . You are receiving this because you commented.Message ID: <Anjok07/ultimatevocalremovergui/repo-discussions/2095/comments/16105688@ github.com>

-- Nelson "nPrevail" Nailat (224) 36N-PREV (224) 366-7738 www.nprevail.bandcamp.com www.mixcloud.com/hasi_la_pheeque "You got to have vision."

[nPrevail]

Thanks again!

…

On Thu, Mar 12, 2026 at 6:31 PM nPrev ***@***.***> wrote: Yup! I was able to run UVR. On Thu, Mar 12, 2026 at 4:20 PM AhuracMusic ***@***.***> wrote: > Well, I have no take on fixing this issue... But you managed to run UVR > at least? > > — > Reply to this email directly, view it on GitHub > <#2095 (reply in thread)>, > or unsubscribe > <https://github.com/notifications/unsubscribe-auth/ANYETW4TVGOHLJZUY7ZGA3T4QNA3FAVCNFSM6AAAAACQYPK5DCVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTMMJQGU3DQOA> > . > You are receiving this because you commented.Message ID: > <Anjok07/ultimatevocalremovergui/repo-discussions/2095/comments/16105688@ > github.com> > -- Nelson "nPrevail" Nailat (224) 36N-PREV (224) 366-7738 www.nprevail.bandcamp.com www.mixcloud.com/hasi_la_pheeque "You got to have vision."

-- Nelson "nPrevail" Nailat (224) 36N-PREV (224) 366-7738 www.nprevail.bandcamp.com www.mixcloud.com/hasi_la_pheeque "You got to have vision."

[AhuracMusic]

No problem!