datasets
librosa
pycountry
scipy
sentencepiece
transformers>=2.5
torch
IPython
TTS
pandas
wave
#py-espeak-ng
espeakng