1 paper · 1 filter
MinJu Jeon, Younghan Park, Han Sung Park +3
Text-to-speech systems increasingly process user-generated text (UGT) such as ppl and imo, whose pronunciation must be inferred from the canonical rather than surface form. We intr…