Removed all text cleaning such as number and abbreviation expansions which are dataset dependent.
Now it relies on the phonemizer cleaning unwanted chars and filter only happens on the output of the phonemizer.
Stripping and punctuation keeping switched from optional to mandatory.
Removed all text cleaning such as number and abbreviation expansions which are dataset dependent. Now it relies on the phonemizer cleaning unwanted chars and filter only happens on the output of the phonemizer. Stripping and punctuation keeping switched from optional to mandatory.