Linux Mint LTS NVIDIA installation bug reports and update README #433
Comments
|
Do you have the installation log? I'm seeing that the packages are
installed in two different places. That could be one reason for this error.
home/desktoppc/.local/lib/python3.10/site-packages/scipy/io/wavfile.py
And
home/desktoppc/Desktop/audiocomfyui/tts-generation-webui-main/installer_files/env/lib/python3.10/site-packages/gradio/queueing.py
…On Sat, Dec 21, 2024, 8:10 PM bastard ***@***.***> wrote:
Traceback (most recent call last):
File "/home/desktoppc/Desktop/audiocomfyui/tts-generation-webui-main/installer_files/env/lib/python3.10/site-packages/gradio/queueing.py", line 624, in process_events
response = await route_utils.call_process_api(
File "/home/desktoppc/Desktop/audiocomfyui/tts-generation-webui-main/installer_files/env/lib/python3.10/site-packages/gradio/route_utils.py", line 323, in call_process_api
output = await app.get_blocks().process_api(
File "/home/desktoppc/Desktop/audiocomfyui/tts-generation-webui-main/installer_files/env/lib/python3.10/site-packages/gradio/blocks.py", line 2015, in process_api
result = await self.call_function(
File "/home/desktoppc/Desktop/audiocomfyui/tts-generation-webui-main/installer_files/env/lib/python3.10/site-packages/gradio/blocks.py", line 1562, in call_function
prediction = await anyio.to_thread.run_sync( # type: ignore
File "/home/desktoppc/.local/lib/python3.10/site-packages/anyio/to_thread.py", line 33, in run_sync
return await get_asynclib().run_sync_in_worker_thread(
File "/home/desktoppc/.local/lib/python3.10/site-packages/anyio/_backends/_asyncio.py", line 877, in run_sync_in_worker_thread
return await future
File "/home/desktoppc/.local/lib/python3.10/site-packages/anyio/_backends/_asyncio.py", line 807, in run
result = context.run(func, *args)
File "/home/desktoppc/Desktop/audiocomfyui/tts-generation-webui-main/installer_files/env/lib/python3.10/site-packages/gradio/utils.py", line 865, in wrapper
response = f(*args, **kwargs)
File "/home/desktoppc/Desktop/audiocomfyui/tts-generation-webui-main/tts_webui/decorators/gradio_dict_decorator.py", line 133, in wrapper
result_dict = fn(**_get_mapped_args(inputs, list_args), outputs=outputs)
File "/home/desktoppc/Desktop/audiocomfyui/tts-generation-webui-main/extensions/builtin/extension_decorator_save_ffmpeg/main.py", line 75, in wrapper
result_dict = fn(*args, **kwargs)
File "/home/desktoppc/Desktop/audiocomfyui/tts-generation-webui-main/extensions/builtin/extension_decorator_save_ffmpeg/main.py", line 65, in wrapper
result_dict = fn(*args, **kwargs)
File "/home/desktoppc/Desktop/audiocomfyui/tts-generation-webui-main/extensions/builtin/extension_decorator_save_waveform/main.py", line 29, in wrapper
result_dict = fn(*args, **kwargs)
File "/home/desktoppc/Desktop/audiocomfyui/tts-generation-webui-main/extensions/builtin/extension_decorator_average_execution_time/main.py", line 27, in wrapper
result = fn(*args, **kwargs)
File "/home/desktoppc/Desktop/audiocomfyui/tts-generation-webui-main/tts_webui/rvc_tab/rvc_tab.py", line 58, in wrapper
return fn(*args, **kwargs)
File "/home/desktoppc/Desktop/audiocomfyui/tts-generation-webui-main/tts_webui/decorators/decorator_apply_torch_seed.py", line 7, in wrapper
return fn(*args, **kwargs)
File "/home/desktoppc/Desktop/audiocomfyui/tts-generation-webui-main/tts_webui/decorators/decorator_save_metadata.py", line 8, in wrapper
result_dict = fn(*args, **kwargs)
File "/home/desktoppc/Desktop/audiocomfyui/tts-generation-webui-main/tts_webui/decorators/decorator_save_wav.py", line 16, in wrapper
_save_wav(result_dict)
File "/home/desktoppc/Desktop/audiocomfyui/tts-generation-webui-main/tts_webui/decorators/decorator_save_wav.py", line 10, in _save_wav
write_wav(path, SAMPLE_RATE, audio_array)
File "/home/desktoppc/.local/lib/python3.10/site-packages/scipy/io/wavfile.py", line 798, in write
dkind = data.dtype.kind
AttributeError: 'NoneType' object has no attribute 'dtype'
AttributeError: 'NoneType' object has no attribute 'dtype', ensure that
the variable audio_array is not None before passing it to the write_wav
function.
—
Reply to this email directly, view it on GitHub
<#433>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/ABTRXI6XLY6CK35YTRUMO6T2GWVIDAVCNFSM6AAAAABUA2R47KVHI2DSMVQWIX3LMV43ASLTON2WKOZSG42TIMRZHA3DOMY>
.
You are receiving this because you are subscribed to this thread.Message
ID: ***@***.***>
|
|
no install log, ive wiped clean and am trying again for the fourth time! surprised that the app is accessing packages outside of the env im on Linux Mint LTS with a NVIDIA GPU. A lot of the applications aren't working, is there a debug mode where I can send you the log output? heres my new install, hope that log is the log youre looking for |
|
getting this error with some of the programs ive tried doing this and then this in my python env but that doesnt work the only thing that works is the legacy version of RVC, nothing else runs. i installed with the installer |
|
RustoMCSpit
changed the title
is there anything else you need? |
|
Thanks, I dug into the logs to see that it is supposedly installing the correct versions: Python 3.10, Cuda 11.8. Error: Detected that PyTorch and TorchAudio were compiled with different CUDA versions. PyTorch has CUDA version 12.1 whereas TorchAudio has CUDA version 11.8. Please install the TorchAudio version that matches your PyTorch version. This implies that the pytorch used by the app is a different one from the conda installation; however, I cannot see any evidence of it in the install log, it all appears to be installed under tts-generation-webui. By the way, this is wrong: |
|
I'll try this all later! |
By removing the cu121 and doing |
|
Without +cu it will most probably install the CPU only version. I forgot torchvision, if you add it to the install torch command, it will select the correct version automatically. |
can you give me the command so i can copy and paste it, im a bit confused at this point |
|
? |
|
|
|
|
when do i do this after a fresh git clone? before or after |
You could probably do it on the old installation and fix it. But remember that you need to do it in the virtual environment of the project, so if you use start_tts then first install, then run this with installer_scripts/conda_env_bash.sh to have the environment active and install PyTorch in the right place. |
i see this file but i have never once used it. what i did was simple.
what is the file above for |
|
|
So, if you use start_tts it has its own virtual environment. |
wait so if i just run start_tts, thats a virtual environment itself? i made my own venv and then when into start_tts. so the venv i made has the correct pytorch versions, but the start_tts venv does not.
how do i run this command after running start_tts (the terminal does not let me input anything while it it running) |
and what do i do with it |
|
you should specify that start_tts is its own venv in the readme and that you do not need and shouldnt use another venv before accessing it or it will break it |
|
? |
|
Sorry, I thought you were able to solve your issue. I will update the readme with the info you provided. Inside conda env bash pip install torch==2.3.1 torchvision==0.18.1 torchaudio==2.3.1 --index-url https://download.pytorch.org/whl/cu121 |
|
okay please update readme |
just to be clear, this is the env of start_tts? so do i just run it and do i tried running it in terminal and got this |
|
Oh, please try it from one directory up then: |
im on linux if that means anything |
|
|
|
it let me enter there. just to be clear, was that a venv or not? it didnt tell me and it had no notification output. the only reason i knew it ran was that it gave no error message and when i tried to close my terminal the system warned me a process was running. regardless, still getting the same problems: |
|
Yes, that command should be the vent active, you could try to confirm it by doing a |
|
I will do a fresh install. Please do update the readme to clarify that users SHOULD NOT use their own virtual environments as start_tts is a virtual environment in itself |
|
Good, we see the problem, but not the reason. It clearly shows that it's using the wrong Pytorch, in your .local folder not tts webui. |
I will, once I'm back to a workstation, still 3-4 days away. |
|
If LLMs are not lying, this could be the fix: This might force pip to ignore the .local Pytorch and have you install one locally. |
RustoMCSpit
changed the title
|
this re-install has taken 40 minutes, do you need to do this for every update? |
|
Totally fresh install, I did not use an environment this time and just used start_tts. I haven't installed any extra requirements or done anything else. I just git cloned and ran start_tts. |
|
No no, this is just because we're trying to figure out why it prefers the wrong Pytorch. But clearly it does it every time. Seems like a PATH problem. Also seems like instead of conda installing python, it sometimes or always uses the system python. We can confirm this with Regular updates are quite fast. |
|
|
whois? |
|
Sorry, it should have been |
|
|
is that good? |
|
It shows the right python, while also using the wrong pip's pytorch location. Perhaps, unless it's important for other projects, you could delete that python3.10 pytorch? I'm starting to think of just manually kicking it back into shape at this point. It's a bit of a mystery for my skill level as to why exactly in this case it chooses to use a random python cache. I guess there's some PATH variable or pip cache configuration that does it. Also finally updated the README file:
Edit: also do you reckon I should make some graphs and flowcharts about the architecture of this project for when folks want to do something new or when things (not so uncommonly) break? |
go for it |
thats where all my python files are. there is nothing else in the lib folder but this. can you post a forum post here https://discuss.python.org/c/help/7 |
|
are you using linux out of curiousity? maybe you can replicate this. ive included my steps |
? |
|
No, this is not a Linux specific problem. I and other Linux users have
installed successfully without this ever happening before.
At the moment the only thing I could recommend is reinstalling Python.
HOWEVER since Python is part of Linux system utilities, it's important to
be sure that your OS does not rely on Python 3.10, but rather 3.11 or 3.12
(common nowadays).
…On Sun, Feb 23, 2025, 11:19 PM Adam ***@***.***> wrote:
Location: /home/chris/.local/lib/python3.10/site-packages
thats where all my python files are. there is nothing else in the lib
folder but this. can you post a forum post here
https://discuss.python.org/c/help/7
?
—
Reply to this email directly, view it on GitHub
<#433 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ABTRXI2YWZI2YGXC2WXGNPT2RI3PVAVCNFSM6AAAAABUA2R47KVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDMNZXGEZDCMBQGQ>
.
You are receiving this because you commented.Message ID:
***@***.***>
[image: RustoMCSpit]*RustoMCSpit* left a comment
(rsxdalv/tts-generation-webui#433)
<#433 (comment)>
Location: /home/chris/.local/lib/python3.10/site-packages
thats where all my python files are. there is nothing else in the lib
folder but this. can you post a forum post here
https://discuss.python.org/c/help/7
?
—
Reply to this email directly, view it on GitHub
<#433 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ABTRXI2YWZI2YGXC2WXGNPT2RI3PVAVCNFSM6AAAAABUA2R47KVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDMNZXGEZDCMBQGQ>
.
You are receiving this because you commented.Message ID:
***@***.***>
|
|
I have been doing some research into PATH and environment variables. In short, a moved VENV is a broken VENV. VENV uses hardcoded absolute addresses for its PATH. The PATH will resort to the system wide pip if the moved VENV's pip can't be found, such as when it is moved, hence This may be a bug conditional on how this program is installed, as you offer multiple methods, or maybe I have misdiagnosed the problem. Please investigate further. |
did you do this? |
AttributeError: 'NoneType' object has no attribute 'dtype', ensure that the variable audio_array is not None before passing it to the write_wav function.
The text was updated successfully, but these errors were encountered: