Spaces:

---

parler-tts

/

parler_tts

like 820

Running on Zero

For reference, here's the classic mistakes the model seems to make at the moment:

• Can't pronounce abbreviations, seems ok if we separate letters
• Mispronounce prompts' last word if it doesn't end with a punctuation mark
• Can't pronounce numbers
• Mistakes hyphens for long pauses (e.g text-to-speech)

This PR shows how to correct most of these issues with proper prompt normalization