runtime downloading of piper voice model

[g3ar-v]

Problem statement
From the codebase here, it seems new voice models can't be downloaded if piper already exists because the only time a voice model would be downloaded is if piper doesn't exist.

Solution
Check if the PIPER_VOICE file declared in .env exists in the piper directory. if not download it, if the link and name's are provided in .env

[g3ar-v]

I have implemented a solution for this and I could create a pull request. If necessary.

[tyfiero]

I just merged some changes that help with piper downloading, can you check to see if the latest merges fix the problem? If not please creat a pull request with your fix. Thanks!

[g3ar-v]

Just checked it. The merge wouldn't. I'll go ahead with the pull request.

[andrewfr]

@g3ar-v @tyfiero

I think it would help (including speed things up a bit) if the model (voice) was preloaded in the piper directory as a part of the installation.

[g3ar-v]

Valid point, but I believe that the aim was to avoid a bloated codebase, so it wasn't done that way.

[andrewfr]

@g3ar-v Okay. I don't consider the models the code base, per se. I've been writing small programs with Piper to get a feel for it. My sense is if you need a model, why not pre-install it, rather than install it upon set up?