Wrapper script for corpus generation about deepspeech-italian-model

[Mte90]

We have one but i think that need a bit of testing https://github.com/MozillaItalia/DeepSpeech-Italian-Model/blob/master/MITADS/merge_txt.sh

[dag7dev]

Wouldn't it be better to do this in Python?

[Mte90]

I was thinking that executing program instead of a script to sort the content and keep unique lines is more fast then a python script.
Anyway I think that we can test it.

[Mte90]

As it is now the bash script:

• https://github.com/lostutils/uq is used because is more fast of sort|uniq, require to be compiled
• Remove the space at beginning of sentences (I think that we can remove it)

So we need:

• Ignore sentences with [^A-Za-z0-9àÈèìéòù,;:'.! ] characters from not this set
• extend to all the txt files of output folder

[Mte90]

Just as mention, this work is in progress in a dedicated branch https://github.com/MozillaItalia/DeepSpeech-Italian-Model/tree/wrapper

[Mte90]

As today the wrapper is missing wikisource and opensubtitles