closed

[quakeman00]

nope

[erew123]

@quakeman00 Have you read my statement on Support and why I am restricted/cannot spend all my time supporting people?

Do you mean like these on the Wiki? https://github.com/erew123/alltalk_tts/wiki/Install-%E2%80%90-Standalone-Installation#quick-setup---video-guide

And the new version has very clear documentation on every page:

What exactly are you installing into/which guide are you following? If you are following the Standard installation, you shouldnt have any messages about "fairseq-0.12.4-cp311-cp311-win_amd64.whl is not a supported wheel on this platform" which would typically come up because you are installing it in a non Python 3.11 environment OR trying to install into a non-windows environment. You are welcome to post the full error up and explain what exactly you are doing and Ill help you thorough it.

[erew123]

Hi @quakeman00 Please glance the over the quick intro to Python environments this may explain that to you and it a 2 minute read.

Because AllTalk is installing so many requirements for so many TTS engines, the requirements have to be carefully managed as other software may upgrade/downgrade packages causing instability, conflicts, so many unknown situations. This is unfortunately the way Python is.

The reason you don't hand out zip files is due to both the size it would need to be e.g. here is the Python environment folder that AllTalk needs/would be built for its environment by atsetup (10GB):

And, because of the open nature of Python code, and packages, they settled on Python PIP as the manager, which is a messy sod at times dealing with conflict resolution issues. Down the chain of AllTalk there is probably code from 1,000-2,000 other projects that PIP figures out what works with what and what to do if one of those people update/change their code because of a bug or security issue (this is called dependency resolution). In short its a complicated mess, but, it is the way Python code/scripts work.

The utility atsetup.bat or atsetup.sh will create its own custom python environment. It should just be loaded at a normal command prompt or terminal, not in a currently loaded Python virtual environment, just a plain old command prompt. e.g it should look exactly like this when you start out:

Over the next few days I will be updating atsetup.bat and atsetup.sh to not only log out its progress, but also carry a little bit more reporting on errors/issues it may encounter.

Youre welcome to get me a screenshot of an error code or copy/paste the text, typically this would be say the final 10-20 lines of the error.

Re Discord or similar.

I would still have an overhead to manage it at a base level. Hopefully some other people will come on board one day and want to take on that responsibility, but I just have too much on my plate at the moment. So the slower pace and one central point of contact allows me to deal with things a bit easier. However, people do respond on the Discussions forums, so you can always post there.

Let me know.

Thanks

[erew123]

Hi @quakeman00

I think I have it figured based on the above. I am going to make some assumptions from that error which are:

1. You are wanting to use AllTalk with Text-generation webui
2. You have copied AllTalk into your ...\text-generation-webui\extensions\alltalk_tts\ folder

---

Installing AllTalk

Be it that the 2nd one is correct or not, Ill tell you the fix and then I will tell you the why.

• Do not git clone Alltalk into ...\text-generation-webui\extensions\alltalk_tts\ remove the alltalk_tts folder if you currently have it in there.
• We will need a separate folder, somewhere else on your disk, that is not inside another applications folder. As such lets say you keep your text-generation-webui in a folder called AI, at the command prompt we will go into the AI folder and clone AllTalk there. So in my example you would go from this:

C:\
├── AI
│    ├── models
│    ├── SillyTavern
│    ├── stuff
│    └── text-generation-webui     # This is where TGWUI is currently installed.
├── boot
├── Program files
├── Windows
etc.....

to looking like this:

C:\
├── AI
│    ├── alltalk_tts    # This is our newly cloned alltalk_tts folder NOT inside another applications folder
│    ├── models
│    ├── SillyTavern
│    ├── stuff
│    └── text-generation-webui
├── boot
├── Program files
├── Windows
etc.....

Once you are have cloned the alltalk_tts folder and then in your command prompt moved into that folder, you will run atsetup.bat and you will choose Standalone Installation

And then Install AllTalk as a Standalone Application

This will create a full custom python environment and do all the things necessary. It should run through to the end without any problems and you should get a screen at the end of that installation which says something like this:

That will be AllTalk installed as a standalone.

When you want to use AllTalk, you will go into its folder and run "starts_alltalk.bat`

---

Getting AllTalk working with Text-generation-webui

Now that AllTalk is running, I am assuming you want it working with Text-generation-webui? To do that you will follow the installation instructions for Text‐generation‐webui Remote Extension

This will install a small extension into TGWUI, that has all the functionality etc.

---

What has the issue been

The atsetup, when running the "Text Generation Webui" install, and not the "Standalone Installation", the TGWUI installation will NOT install a custom Python environment, because you would be installing into TGWUI's custom Python environment.

Because TGWUI's custom Python environment's package requirements have moved its version requirements so high up, Installing Alltalk's main installation directly into its Python environment would damage it.

I reference this issue https://github.com/erew123/alltalk_tts/wiki/Install-%E2%80%90-Text%E2%80%90generation%E2%80%90webui-Installation#read-before-installing-into-tgwuis-python-environment

Here https://github.com/erew123/alltalk_tts/wiki/Install-%E2%80%90-Text%E2%80%90generation%E2%80%90webui-Installation#quick-setup---text-generation-webui-installations

And here #377

---

Hopefully that will answer your question and get you up and running. If so, please feel free to close the ticket.

Once you are up and running, you can check things like the quick start guide to help you though any initial questions https://github.com/erew123/alltalk_tts/wiki

Thanks

[erew123]

This error message here???

ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
gruut 2.4.0 requires jsonlines~=1.2.0, but you have jsonlines 4.0.0 which is incompatible.

Requirement already satisfied: pure-eval in c:\users\dispenser\appdata\local\programs\python\python311\lib\site-packages (from stack-data->ipython->descript-audiotools@ git+https://github.com/descriptinc/audiotools->parler_tts==0.2.1->-r system\requirements\requirements_parler.txt (line 1)) (0.2.3)
** Faiss **
The system cannot find the path specified.
** FFmpeg **
The system cannot find the path specified.

   There was an error installing the requirements.
   Have you started your Text-gen-webui Python environment
   with cmd_{yourOS} before running atsetup.bat?
   Press any key to return to the menu.

where at the bottom of the error message it says:

with cmd_{yourOS} before running atsetup.bat?

This is the error message you are getting still?

[erew123]

And you are telling me that AllTalk the standalone installation has worked? And you can now start AllTalk? so you get this screen?

[erew123]

Ok. At no point should you be using that menu option:

I assume you DID remove the AllTalk folder that you had previously created? (from my instructions above):

And then you install the TGWUI remote extension (from my instructions above):

Please follow these instructions TGWUI remote extension (from my instructions above):

Thanks

[erew123]

I dont understand what you mean by where does it spawn. The instructions in for the TGWUI Remote extension tell you to create that folder:

In the TGWUI Instructions it EXPLICITLY says what to do and links to the relevant sections:
https://github.com/erew123/alltalk_tts/wiki/Install-%E2%80%90-Text%E2%80%90generation%E2%80%90webui-Installation#read-before-installing-into-tgwuis-python-environment

It repeats that instruction, on the same page to draw your attention to it:

Nowhere in the instructions for Standalone installation OR the TGWUI Remote Extension does it tell you to go back into the installation utility atsetup and go to the TGWUI menu.

Do you think I need to explicitly caveat that?

[erew123]

If you are using Text-generation-webui from Oobabooga https://github.com/oobabooga/text-generation-webui that is the standard folder name it will create when you follow their installation instructions. https://github.com/oobabooga/text-generation-webui?tab=readme-ov-file#how-to-install

[erew123]

If you are NOT using Text-generation-webui, then you wouldn't follow the instructions for that.

[erew123]

You don't need another one! If you just want it to use it with SillyTavern, you just want to install it as a Standalone.

https://github.com/erew123/alltalk_tts/wiki/Install-%E2%80%90-Standalone-Installation

TGWUI is a completely different thing. No-where in the Standalone instructions does it say to install the TGWUI requirements, or install it in Google Colab, or Docker etc.

This is an Options list that you pick the option you want/need to match your requirements:

If all you want it AllTalk to work with SillyTavern and you have whatever backend loader for your AI model (OpenAI, Claude, Kobold, TGWUI, whatever) you can just install AllTalk as a Standalone install and it will speak to SillyTavern which and SillyTavern will speak to your AI model however you set that up.

AllTalk is used across 50+ different projects on the internet, its not just for SillyTavern, so I cannot caveat every individual persons custom scenario on a - If you have XX and Also YY and want ZZ then you need to install ABC way. Unfortunately that is far too complicated for me to cover every scenario for every person.

I will take on board what you say though and what you found difficult and think about how to re-visit it. Though I did check the Sillytavern documentation and it does state the choice in their documentation https://docs.sillytavern.app/extensions/alltalk/ which is explicit instructions for SillyTavern and links to the correct AllTalk instructions as you need.

Are you at least up and running now?

[erew123]

Please go to this link and click the download (as shown in the image below): https://github.com/erew123/alltalk_tts/blob/alltalkbeta/system/TGWUI_Extension/script.py

You want to save that over the top (replacing) the other script.py file in c:\Users\DISPENSER\Desktop\genstuff_00\text-generaetion-webui-main\extensions\alltalk_tts\ so replace the scipt.py in that location with the one you just downloaded.

That will solve your first problem.

As for your second problem, have you looked at the quick start guide?: https://github.com/erew123/alltalk_tts/wiki/AllTalk-V2-QuickStart-Guide

Each TTS engine has various locations for different things and there is help throughout the whole of the Gradio interface. Typically most voices are stored in the voices folder.

As mentioned in the quickstart guide, you can press Ctrl+left click on the clickable links to to open it:

Each TTS engine has its own help page/guide e.g.

And pretty much every single page in the interface has an expandable help section detailing what is on that page and how to use it:

Beyond that, the Wiki is full of lots of other information.

[erew123]

All TTS engines are managed in the central interface. You should use the download tab for each engine e.g.

[CaseyleeIRL]

I just wanted to comment - that Alltalk's install and download - as well as the UI for training has a been a breeze to work, max appreciate. And dev showed so much patience on this thread x.x

[ldavis9000aws]

I just wanted to comment - that Alltalk's install and download - as well as the UI for training has a been a breeze to work, max appreciate. And dev showed so much patience on this thread x.x

I know...dev needs an award for sticking this one out, lol...