Hacker News new | ask | show | jobs
by snakers41 1466 days ago
I am not sure, what can be more simple than 1 LOC invocation + minimal imports.

It is true that the model is based on PyTorch + python, but the majority of complexity (like SSML parsing) is tucked inside of the model.

Theoretically one can make a simplified model without any of those features in plain PyTorch or ONNX, but so far we did not have proper motivation to do.

As for CLI, this also seems simple enough, but out of scope for us.
2 comments
zelphirkalt 1466 days ago
In such situations it could be useful to provide a container image or nix or guix shell setup, to make sure people have the dependencies they need.
link
ur-whale 1466 days ago
> I am not sure, what can be more simple than 1 LOC invocation + minimal imports.

Let me make it "embarrassingly simple" for you:

    /bin/bash text_to_speech.sh file.txt file.wav
Also, I'm not entirely sure what "out of scope" mean?

Do you mean you run your software on computers that can't run bash?

Do you develop machine learning algorithms on your phone?

link
snakers41 1466 days ago
> Do you mean you run your software on computers that can't run bash?

It is explicitly stated, that PyTorch is the only real requirement. Bash is not required, i.e. models can be run on Windows or ARM with PyTorch.

> Also, I'm not entirely sure what "out of scope" mean?

There was no tangible benefit in making a bash CLI for us.

link