# Miniconda Setup

If you had installed wrong version nor even you haven't installed python yet and you want multiple python versions then follow my guide on how to install Miniconda.

## Installing on your machine.

So let's start off with going to this page:

{% embed url="<https://docs.conda.io/en/latest/miniconda.html>" %}

Here, we are in the main website on instructions on how to:

![](/files/WY3eQM0u7PQqKC20lLF6)

And to be frankly enough... it's even available for Mac and Linux, but we'll focus on Windows okay? Works Linux too for training but for Macs... I don't think so... Synthesis will work but training is probably not... (maybe...) :disappointed: Sorry Mac users.

If we scroll down till we find this page:

![](/files/SxJxkxK81wU7pqQQZ2Cb)

Now, all we need to get is Python 3.7 AND 64-bit (because, I mean who wants to use 32-bit anyways.)

You will be greeted by the installer itself:

![](/files/FZxn7C9natE7TNtlImR3)

I'll tell you what to do in step by step:

1. Click Next.
2. Click I Agree.
3. Click Next (Unless if you want install for yourself).
4. Change location if you want to, but you can click Next anyways.
5. This is optional but check Add Miniconda3 to the PATH, otherwise click Install.

Now wait till installs.

Once that's done, you can uncheck these two and click finish.

## Usage of Miniconda

So, now that you got installed, it's time to go ahead and try to use it.

> But awesome, where and how will I use Miniconda?

That's a good question.

The real deal is that you gonna have to open it trough Start Menu on your Windows (Or use Terminal if you are in Linux).

![](/files/xAG7S2EPlOHQC1OqRh5z)

It may not so hard, so go ahead and click on Anaconda Prompt.

![](/files/AgIXCtrrtbrt1DyOqGJt)

Welcome to the Anaconda Terminal (or Miniconda3, whatever is.)

Right now you need to make an virtual environment for it. So one thing you should be doing is that installing python 3.7.X and all that stuff. (Do not install new versions or older, they don't work. ONLY INSTALL 3.7.X Or you may get an errors.)

So to make a virtual environment just type like this:

```
conda create -n *whatever name you want but i suggest tacotron2* python=3.7
```

So let's go one by one:

* conda - is basically command you should use
* create - this creates an environment
* -n - is the name enviroment you should name in to
* python=3.7 - this will force to install python 3.7.X so make sure you write that (this may work on latest version too don't worry about it)

Now that you have to wait until it's finished.

Once you have done, go ahead and type this command:

```
conda activate *the name that you wrote*
```

So basically what it does, it changes base environment in to your environment.

It even says in command line:

![](/files/GNwT5bcuyEUleKCMkoCt)

To make sure if your python the same version as we are, type:

```
 python --version
```

![](/files/OdEGCdVuJFTqwluH9WUn)

If you have 3.7 version then congratulations you can continue on, if not, go ahead try again.

Lastly but not at least, you should be able to install PyTorch and other dependencies. Which we definitely required for this or else you may run in to some errors.

To do this, you first need to check what cuda version you have, open up cmd and run this command:

```
nvidia-smi
```

Once that's done you should see an CUDA version right here:

![](/files/1UH5s31bD6nMMvITdio1)

Go to the PyTorch website:

{% embed url="<https://pytorch.org/get-started/locally>" %}

And you should see it something like this:

![](/files/0b9RclBiX9qEi4XUm4f2)

Now, remember that CUDA version that we did? here's what you should do:

If you have CUDA Version 10.2 to 11.3, select CUDA 10.2

Otherwise if you have CUDA 11.3 or up then choose that one.

Other than that, you cannot install PyTorch if you have lower version. (sorry)

For myself I choose CUDA 11.3 so, select that one and copy the entire command, run inside conda terminal and press enter. Now this may take a while, so be patient...

![](/files/t6Wib565LrBnlPFOhhnn)

After that, type this:

```
pip install matplotlib tensorflow==1.15.2 numpy inflect librosa==0.8.1 scipy Unidecode pillow tqdm
```

Now it should be installing some dependencies.


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://uberduck-guide.gitbook.io/uberduck-written-tutorial/installing-tacotron-2-locally/miniconda-setup.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.