The new "deepspeed" requires env PATH CUDA_HOME about xtts-webui

[daswer123]

About CUDA, yes I have both 11.8 and 12.1 in PATH, I change them when I need it.

xtts_webui.py can work with CUDA 12.1 , but xtts_finetune_webui can't do it, it needs CUDA 11.8 because of the faster-whisper library, that's why I have CUDA 11.8 in the project requirements.

There are prebuild wheels for CUDA 12.1 for windows, you can find them here
https://github.com/erew123/alltalk_tts/releases/tag/deepspeed

[daswer123]

Thank you. I'm glad my work is useful.

The only bad news is that they are not registered by the WebUI display. This realization occured prior to FFMPEG so it had nothing to do with that. I can live without the ultra high quality versions being displayed in WebUI as I specifically use the files regardless.

I plan to add a status bar to display progress status and error messages

[daswer123]

Hi, thanks for letting me know, I'll add the instructions so people know.

I changed and then returned, now I need cuda 11.8 , I'm not going to go higher because of the dependence of some libraries

https://developer.nvidia.com/cuda-11-8-0-download-archive

[MackNcD]

Did you just change back recently, I believe I may have the 12.1...
Also, is there a CUDA_HOME already in your PATH?
If you didn't have to download it, I'm curious as to whether or not it's necessary -- or if that link is the correct solution.

If you don't have CUDA_HOME and it's working for you, I'll try find out the core issue here, so we can ensure people can get deepspeed to run without extra steps.

[MackNcD]

ALERT! DO NOT ADD THE LINK TO THE README!

Okay, for anyone searching issues that was having to manually enter the dependencies that'd ordinarily be DL'd with the pip install -r requirements.txt
When you're install your dependencies, DO NOT simply pip install deepspeed like any other module.

Instead of pip install deepspeed

Just run/enter
python scripts/modeldownloader.py
That's the full length of text, dawser has put the install instructions inside that python file.

Make sure python is in your PATH -- check with python --version in whatever cmd prompt you do your python stuff in.
IF NOTHING COMES UP, LINK YOUR PATH DIRECTLY TO YOUR PYTHON.EXE, NOT THE SCRIPTS DIRECTORY. NORMALLY YOU WON'T HAVE TO DO THIS, BUT IF SUCH A THING OCCURS, THIS IS A WAY TO GET AROUND IT.

[daswer123]

I'll add that to the instructions for manual installation, thanks for checking everything out

[MackNcD]

No problem!
One last thing. For anyone launching from a code editor, remember to remove the commented out 'launch demo' line, it should be at the very bottom of the xtts_webui.py -- also, 'untab it' -- bring it to the far left side. For those unfamiliar with python, but happen to be running an editor, this will help, just a bit, lol.

[daswer123]

One last thing. For anyone launching from a code editor, remember to remove the commented out 'launch demo' line, it should be at the very bottom of the xtts_webui.py

Removed the comment so others wouldn't have to do it

[MackNcD]

Spasibo!

Now that the new build is launching, I hate to come to you with more issues -- I'll work on them but you know your own UI better than I do, so maybe you'll spot the line more quickly... (The normal inferencing is working superbly, BTW)
This is with the 'improve output quality/reduce noise' optoin checked:

Exception: None
2023-12-21 22:38:26.268 | INFO | scripts.tts_funcs:local_generation:236 - Processing time: 15.75 seconds. #without
2023-12-21 22:39:13.120 | INFO | scripts.tts_funcs:local_generation:236 - Processing time: 15.05 seconds. #without
2023-12-21 22:40:08.728 | INFO | scripts.tts_funcs:local_generation:236 - Processing time: 14.05 seconds. #with ...
Traceback (most recent call last):
File "C:\Users\user\xtts-webui-main.venv\lib\site-packages\gradio\queueing.py", line 459, in call_prediction
output = await route_utils.call_process_api(
File "C:\Users\user\xtts-webui-main.venv\lib\site-packages\gradio\route_utils.py", line 232, in call_process_api
output = await app.get_blocks().process_api(
File "C:\Users\user\xtts-webui-main.venv\lib\site-packages\gradio\blocks.py", line 1533, in process_api
result = await self.call_function(
File "C:\Users\user\xtts-webui-main.venv\lib\site-packages\gradio\blocks.py", line 1151, in call_function
prediction = await anyio.to_thread.run_sync(
File "C:\Users\user\xtts-webui-main.venv\lib\site-packages\anyio\to_thread.py", line 33, in run_sync
return await get_asynclib().run_sync_in_worker_thread(
File "C:\Users\user\xtts-webui-main.venv\lib\site-packages\anyio_backends_asyncio.py", line 877, in run_sync_in_worker_thread
return await future
File "C:\Users\user\xtts-webui-main.venv\lib\site-packages\anyio_backends_asyncio.py", line 807, in run
result = context.run(func, *args)
File "C:\Users\user\xtts-webui-main.venv\lib\site-packages\gradio\utils.py", line 678, in wrapper
response = f(*args, **kwargs)
File "c:\Users\user\xtts-webui-main\xtts_webui.py", line 295, in generate_audio
output_file = improve_and_convert_audio(output_file,output_type)
File "c:\Users\user\xtts-webui-main\scripts\funcs.py", line 159, in improve_and_convert_audio
ffmpeg
File "C:\Users\user\xtts-webui-main.venv\lib\site-packages\ffmpeg_run.py", line 284, in run_async
return subprocess.Popen(
File "C:\Users\user\AppData\Local\Programs\Python\Python310\lib\subprocess.py", line 971, in init
self._execute_child(args, executable, preexec_fn, close_fds,
File "C:\Users\user\AppData\Local\Programs\Python\Python310\lib\subprocess.py", line 1456, in _execute_child
hp, ht, pid, tid = _winapi.CreateProcess(executable, args,
FileNotFoundError: [WinError 2] The system cannot find the file specified
Traceback (most recent call last):
File "C:\Users\user\xtts-webui-main.venv\lib\site-packages\gradio\queueing.py", line 459, in call_prediction
output = await route_utils.call_process_api(
File "C:\Users\user\xtts-webui-main.venv\lib\site-packages\gradio\route_utils.py", line 232, in call_process_api
output = await app.get_blocks().process_api(
File "C:\Users\user\xtts-webui-main.venv\lib\site-packages\gradio\blocks.py", line 1533, in process_api
result = await self.call_function(
File "C:\Users\user\xtts-webui-main.venv\lib\site-packages\gradio\blocks.py", line 1151, in call_function
prediction = await anyio.to_thread.run_sync(
File "C:\Users\user\xtts-webui-main.venv\lib\site-packages\anyio\to_thread.py", line 33, in run_sync
return await get_asynclib().run_sync_in_worker_thread(
File "C:\Users\user\xtts-webui-main.venv\lib\site-packages\anyio_backends_asyncio.py", line 877, in run_sync_in_worker_thread
return await future
File "C:\Users\user\xtts-webui-main.venv\lib\site-packages\anyio_backends_asyncio.py", line 807, in run
result = context.run(func, *args)
File "C:\Users\user\xtts-webui-main.venv\lib\site-packages\gradio\utils.py", line 678, in wrapper
response = f(*args, **kwargs)
File "c:\Users\user\xtts-webui-main\xtts_webui.py", line 295, in generate_audio
output_file = improve_and_convert_audio(output_file,output_type)
File "c:\Users\user\xtts-webui-main\scripts\funcs.py", line 159, in improve_and_convert_audio
ffmpeg
File "C:\Users\user\xtts-webui-main.venv\lib\site-packages\ffmpeg_run.py", line 284, in run_async
return subprocess.Popen(
File "C:\Users\user\AppData\Local\Programs\Python\Python310\lib\subprocess.py", line 971, in init
self._execute_child(args, executable, preexec_fn, close_fds,
File "C:\Users\user\AppData\Local\Programs\Python\Python310\lib\subprocess.py", line 1456, in _execute_child
hp, ht, pid, tid = _winapi.CreateProcess(executable, args,
FileNotFoundError: [WinError 2] The system cannot find the file specified

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
File "C:\Users\user\xtts-webui-main.venv\lib\site-packages\gradio\queueing.py", line 497, in process_events
response = await self.call_prediction(awake_events, batch)
File "C:\Users\user\xtts-webui-main.venv\lib\site-packages\gradio\queueing.py", line 468, in call_prediction
raise Exception(str(error) if show_error else None) from error

The normal inferencing worked just fine, also the playback function is now working without the need to install an external FFMPEG.
I'm going through all other options now to make sure everything is good -- looks like you spotted the UI type-o I was going to mention as well.

Update: Attempt at fix 1: Included some Japanese/Korean token dependancies just in case they were in the line of functions somewhere. The cutlet dependancy is now the only one excluded (not by choice, the wheel wouldn't work)
No change
Attempt 2: Returned to the standard model from a fine-tuned model, again trying to replicate everything perfectly as it would be on the creator's home system.
No change
Attempt 3: Downloading FFMPEG again--looking at replicating configurations perfectly--bit of a lengthy download on the mirror I chose, will update again with more later. WML
Awaiting results

[MackNcD]

Very good news and slightly bad news:
(When turning on resemblance/denoiser) The outputs have indeed been added to the output folder, and with much quicker speed, and with higher quality.

The only bad news is that they are not registered by the WebUI display. This realization occured prior to FFMPEG so it had nothing to do with that. I can live without the ultra high quality versions being displayed in WebUI as I specifically use files -- I'm not simply reading myself bedtime stories :)

Both resemblance and the noise reduction function work like a charm -- aside from webUI direct preview.

Great work so far man, this is really well done.