Unable to launch in Windows 11

[larryhardin]

Hello,

After setting up my conda environment I received the error:
"No module named 'transformers.modeling_layers'"
So, I installed peft==0.17.1.

I have installed triton-windows.

Now I receive this error:
TypeError: Audio.init() got an unexpected keyword argument 'show_download_button'

Here is the log from the run:
acestep --port 7865
Traceback (most recent call last):
File "C:\Users\lwhar\miniconda3\envs\ace_step\lib\runpy.py", line 196, in _run_module_as_main
return _run_code(code, main_globals, None,
File "C:\Users\lwhar\miniconda3\envs\ace_step\lib\runpy.py", line 86, in run_code
exec(code, run_globals)
File "C:\Users\lwhar\miniconda3\envs\ace_step\Scripts\acestep.exe_main.py", line 6, in
sys.exit(main())
File "C:\Users\lwhar\miniconda3\envs\ace_step\lib\site-packages\click\core.py", line 1485, in call
return self.main(*args, **kwargs)
File "C:\Users\lwhar\miniconda3\envs\ace_step\lib\site-packages\click\core.py", line 1406, in main
rv = self.invoke(ctx)
File "C:\Users\lwhar\miniconda3\envs\ace_step\lib\site-packages\click\core.py", line 1269, in invoke
return ctx.invoke(self.callback, **ctx.params)
File "C:\Users\lwhar\miniconda3\envs\ace_step\lib\site-packages\click\core.py", line 824, in invoke
return callback(*args, **kwargs)
File "G:\AI_Projects\ACE-Step\acestep\gui.py", line 70, in main
demo = create_main_demo_ui(
File "G:\AI_Projects\ACE-Step\acestep\ui\components.py", line 1006, in create_main_demo_ui
create_text2music_ui(
File "G:\AI_Projects\ACE-Step\acestep\ui\components.py", line 137, in create_text2music_ui
ref_audio_input = gr.Audio(type="filepath", label="Reference Audio (for Audio2Audio)", visible=False, elem_id="ref_audio_input", show_download_button=True)
File "C:\Users\lwhar\miniconda3\envs\ace_step\lib\site-packages\gradio\component_meta.py", line 194, in wrapper
return fn(self, **kwargs)
TypeError: Audio.init() got an unexpected keyword argument 'show_download_button'

[larryhardin]

If anyone else has the same problem, it is due to a versioning problem with gr.Audio().
To correct this, comment out the show download button as shown. I think it appears in 4 different places.

repaint_source_audio_upload = gr.Audio(
label="Upload Audio",
type="filepath",
visible=False,
elem_id="repaint_source_audio_upload",
#show_download_button=True,
)

[whiteknight6766]

Error also occurs on ubuntu 22.04 inside a fresh venv. After applying the requirements.txt fix from another post. I am assuming "show_download_button" is a versioning error as well. I would like to eventually download the audio. What version should gradio be on? The original requirements has the following entry:

gradio

Showing gradio should be on the latest version. but 6.1.0 came out 09DEC25 and this redo is 4 months old. So i went back into gradio's releases and found that four months ago 5.47.2 was the active version. Pinning gradio to 5.47.2 allowed the application to launch. my requirements became:

datasets==3.4.1
diffusers>=0.33.0
gradio==5.47.2
librosa==0.11.0
loguru==0.7.3
matplotlib==3.10.1
numpy
pypinyin==0.53.0
pytorch_lightning==2.5.1
soundfile==0.13.1
torch
torchaudio
torchcodec
torchvision
tqdm
transformers==4.49.0
py3langid==0.3.0
hangul-romanize==0.1.0
num2words==0.5.14
spacy==3.8.4
accelerate==1.6.0
cutlet
fugashi[unidic-lite]
click
peft==0.17.1
tensorboard
tensorboardX

I am also using cuda 128 vice 126 as suggested in the install script. this caused a version conflict with libtorchcodec so be sure to watch out for that.

[larryhardin]

Error also occurs on ubuntu 22.04 inside a fresh venv. After applying the requirements.txt fix from another post. I am assuming "show_download_button" is a versioning error as well. I would like to eventually download the audio. What version should gradio be on? The original requirements has the following entry:

gradio

Showing gradio should be on the latest version. but 6.1.0 came out 09DEC25 and this redo is 4 months old. So i went back into gradio's releases and found that four months ago 5.47.2 was the active version. Pinning gradio to 5.47.2 allowed the application to launch. my requirements became:

datasets==3.4.1 diffusers>=0.33.0 gradio==5.47.2 librosa==0.11.0 loguru==0.7.3 matplotlib==3.10.1 numpy pypinyin==0.53.0 pytorch_lightning==2.5.1 soundfile==0.13.1 torch torchaudio torchcodec torchvision tqdm transformers==4.49.0 py3langid==0.3.0 hangul-romanize==0.1.0 num2words==0.5.14 spacy==3.8.4 accelerate==1.6.0 cutlet fugashi[unidic-lite] click peft==0.17.1 tensorboard tensorboardX

I am also using cuda 128 vice 126 as suggested in the install script. this caused a version conflict with libtorchcodec so be sure to watch out for that.

Thanks! That worked. RTX3080 12GB on Windows 11.
I stuck with the 126 cuda to avoid the conflict with libtorchcodec. Kinda bad they didn't include that in the requirements file.
I very much appreciate it!

[w44p]

Error also occurs on ubuntu 22.04 inside a fresh venv. After applying the requirements.txt fix from another post. I am assuming "show_download_button" is a versioning error as well. I would like to eventually download the audio. What version should gradio be on? The original requirements has the following entry:

gradio

Showing gradio should be on the latest version. but 6.1.0 came out 09DEC25 and this redo is 4 months old. So i went back into gradio's releases and found that four months ago 5.47.2 was the active version. Pinning gradio to 5.47.2 allowed the application to launch. my requirements became:

datasets==3.4.1 diffusers>=0.33.0 gradio==5.47.2 librosa==0.11.0 loguru==0.7.3 matplotlib==3.10.1 numpy pypinyin==0.53.0 pytorch_lightning==2.5.1 soundfile==0.13.1 torch torchaudio torchcodec torchvision tqdm transformers==4.49.0 py3langid==0.3.0 hangul-romanize==0.1.0 num2words==0.5.14 spacy==3.8.4 accelerate==1.6.0 cutlet fugashi[unidic-lite] click peft==0.17.1 tensorboard tensorboardX

I am also using cuda 128 vice 126 as suggested in the install script. this caused a version conflict with libtorchcodec so be sure to watch out for that.

Thanks! RTX 4080S on Windows 10.
cuda 130

requirements.txt:
datasets==3.4.1
diffusers>=0.33.0
gradio==5.47.2
librosa==0.11.0
loguru==0.7.3
matplotlib==3.10.1
numpy
pypinyin==0.53.0
pytorch_lightning==2.5.1
soundfile==0.13.1
torch
torchaudio
torchvision
tqdm
transformers==4.50.0
py3langid==0.3.0
hangul-romanize==0.1.0
num2words==0.5.14
spacy==3.8.4
accelerate==1.6.0
cutlet
fugashi[unidic-lite]
click
peft==0.17.1
tensorboard
tensorboardX