[FEAT] Optionally normalize/transliterate characters

[NeonDaniel]

Objective

Non-latin languages will not generate audio for Latin characters. There should be a standard method for transliteration here to help make text synthesis possible. Consider also normalizing characters (i.e. @, ') which are not mapped in all language models.

Initial Implementation Requirements

• Add config option to enable transliteration
• Add some standard transliteration mapping for uk (and other languages?)

Other Considerations

neon-tts-plugin-coqui/neon_tts_plugin_coqui/__init__.py

Line 114 in de50afa

wav_data = synthesizer.tts(sentence, **tts_kwargs)