I'm getting an out of CUDA memory error when using the Upscale Audio feature, and I want to know how much CUDA memory is needed to run this feature

[Mingsirdon]

An error occurred connecting to Discord: Could not find Discord installed and running on this machine.
Theme Applio successfully loaded.
Failed to launch on port 6969, trying again on port 6968...
Running on local URL: http://127.0.0.1:6968

To create a public link, set share=True in launch().
Converting audio 'assets\audios.wav'...
Loading audio_upscaler: basic
Loading model on cuda:0
DiffusionWrapper has 258.20 M params.
An error occurred during audio conversion: CUDA out of memory. Tried to allocate 26.00 MiB. GPU
Traceback (most recent call last):
File "E:\AIGAME\Applio\rvc\infer\infer.py", line 260, in convert_audio
upscale(audio_input_path, audio_input_path)
File "E:\AIGAME\Applio\env\lib\site-packages\audio_upscaler\predict.py", line 122, in upscale
p.setup(model_name, device)
File "E:\AIGAME\Applio\env\lib\site-packages\audio_upscaler\predict.py", line 22, in setup
self.audiosr = build_model(model_name=self.model_name, device=self.device)
File "E:\AIGAME\Applio\env\lib\site-packages\audio_upscaler\pipeline.py", line 148, in build_model
latent_diffusion = latent_diffusion.to(device)
File "E:\AIGAME\Applio\env\lib\site-packages\torch\nn\modules\module.py", line 1173, in to
return self._apply(convert)
File "E:\AIGAME\Applio\env\lib\site-packages\torch\nn\modules\module.py", line 779, in _apply
module._apply(fn)
File "E:\AIGAME\Applio\env\lib\site-packages\torch\nn\modules\module.py", line 779, in _apply
module._apply(fn)
File "E:\AIGAME\Applio\env\lib\site-packages\torch\nn\modules\module.py", line 779, in _apply
module._apply(fn)
[Previous line repeated 3 more times]
File "E:\AIGAME\Applio\env\lib\site-packages\torch\nn\modules\module.py", line 804, in _apply
param_applied = fn(param)
File "E:\AIGAME\Applio\env\lib\site-packages\torch\nn\modules\module.py", line 1159, in convert
return t.to(
torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 26.00 MiB. GPU

I'm missing cuda memory when using Upscale Audio, my CUDA memory is 12GB, how much CUDA memory can I use normally?

[AznamirWoW]

upscale audio does not do anything useful, it upscales audio before processing and the processing using 16k resample anyway. So all the work done with upscaling is basically wasted.