TTS Module Missing Attributes

[ZingBlue]

Describe the bug
When importing TTS in Python, you should be able to access attributes via call (e.g: TTS.model). However, most attributes are inaccessible. A few attributes are still accessible such as f, version, and __name__, which all return expected results.

To Reproduce

1. Open your CLI and install TTS via Pip or install manually via the instructions in the Readme (doesn't matter in this case, I tested both).
2. After installation, open a Python environment and import TTS.
3. Try accessing any of the core classes (e.g: TTS.utils)
4. See result: AttributeError: module 'TTS' has no attribute 'utils'

Expected behavior
You should be able to access all of TTS' attributes.

Environment:

• OS Platform and Distribution: Windows 10, 21H1
• PyTorch or TensorFlow version: PyTorch 1.9.0
• Python version: Python 3.9.6
• CUDA/cuDNN version: N / A
• GPU model and memory: N / A
• Exact command to reproduce: N / A

Additional context

1. I tried this on two similar Windows machines and had the same issue.
2. It is also worth noting that TTS functions in the CLI. I can use commands such as TTS --text "The meaning of life is 42." without issue.
3. As of now, Gruut and Gruut-IPA do not install on Windows via Pip. This is because of a single line error in setup.py, I've sent over PRs for both repos and as of now 1/2 have been merged. You will need to manually install Gruut and Gruut-IPA until these changes are merged and released. This is an unrelated issue however, only mentioned here to ensure this does not stall testing.

[erogol]

It is not a bug since PyPIrelease is only for inference. If you want to use TTSfor development, you need to install it from GitHub.

If you want to install it from PyPIeven for development, you can post a feature request.

[ZingBlue]

It is not a bug since PyPIrelease is only for inference. If you want to use TTSfor development, you need to install it from GitHub.

If you want to install it from PyPIeven for development, you can post a feature request.

No, I installed by cloning the repo.

[ZingBlue]

As I said in the post, if you do the commands
git clone https://github.com/coqui-ai/TTS
pip install -e .[all,dev,notebooks,tf] # Select the relevant extras
It will install and have the same issue regardless.

[ZingBlue]

Even pip install -e .\ in the TTS top directory does not allow access to all attributes.

[ZingBlue]

I just went on Mac and installed using the same instructions. Still, no attribute access.

[erogol]

I can't replicate the issue since all works on my side.

What is your system? Do you use conda or something else?

[ZingBlue]

I feel like I must be missing something, to install TTS with access to attributes on a UNIX/Linux system (Mac in this case), all I have to do is clone the repo, and go into the TTS dir and install via pip install -e .[all,dev,notebooks,tf] right? That's all I'm doing yet I still can't access any attributes.

[erogol]

Exactly. To help you in that I should be able to replicate the issue but I cannot.

Try using conda and create a new env and install TTS from scratch by pip install -e .