v0.2.0

1. Fix bug in LinguisticEncoder. See this commit for details.
2. Apply several techniques to improve world-to-phoneme alignment and output accuracy. Check out the Note section in README for details.
3. Update pre-trained models with some demo of the update.

v0.1.0

update pre-trained models with audio demos