Chang H.-JLiu A.HLee H.-YHUNG-YI LEELIN-SHAN LEE2021-09-022021-09-022021https://www.scopus.com/inward/record.uri?eid=2-s2.0-85103924610&doi=10.1109%2fSLT48900.2021.9383595&partnerID=40&md5=5eee06475284ca9a7e017c50c81ca4echttps://scholars.lib.ntu.edu.tw/handle/123456789/580906Whispering is an important mode of human speech, but no end-to-end recognition results for it were reported yet, probably due to the scarcity of available whispered speech data. In this paper, we present several approaches for end-to-end (E2E) recognition of whispered speech considering the special characteristics of whispered speech and the scarcity of data. This includes a frequency-weighted SpecAugment policy and a frequency-divided CNN feature extractor for better capturing the high-frequency structures of whispered speech, and a layer-wise transfer learning approach to pre-train a model with normal or normal-to-whispered converted speech then fine-tune it with whispered speech to bridge the gap between whispered and normal speech. We achieve an overall relative reduction of 19.8% in PER and 44.4% in CER on a relatively small whispered TIMIT corpus. The results indicate as long as we have a good E2E model pre-trained on normal or pseudo-whispered speech, a relatively small set of whispered speech may suffice to obtain a reasonably good E2E whispered speech recognizer. ? 2021 IEEE.Speech; Transfer learning; Feature extractor; Frequency weighted; High frequency HF; Human speech; Layer-wise; Pre-training; Relative reduction; Whispered speech; Speech recognitionconference paper10.1109/SLT48900.2021.93835952-s2.0-85103924610