For example, running G2P on "text-to-speech" returns ['T', 'EH1', 'K', 'S', 'T', 'S', 'P', 'EH2', 'K'], the same as "texttospeech", when it should return something closer to ['T', 'EH1', 'K', 'S', 'T', ' ', 'T', 'UW1', ' ', 'S', 'P', 'IY1', 'CH'], the result for "text to speech" (though the stress could use some adjustment).
Simple workaround for now: use .replace("-", " ") on the input being passed in.
For example, running G2P on "text-to-speech" returns
['T', 'EH1', 'K', 'S', 'T', 'S', 'P', 'EH2', 'K']
, the same as "texttospeech", when it should return something closer to['T', 'EH1', 'K', 'S', 'T', ' ', 'T', 'UW1', ' ', 'S', 'P', 'IY1', 'CH']
, the result for "text to speech" (though the stress could use some adjustment).Simple workaround for now: use
.replace("-", " ")
on the input being passed in.