Table 3.

Model configurations and descriptions.

ModelDescription
Baseline models
Clean model baselineTrained on clean transcriptions
Noisy model baselineTrained on noisy transcriptions
TQL-aware models
w/ UTT TQLWith utterance-level TQL
w/ CHAR TQLWith character-level TQL
w/ UTT TQL + CHAR TQLWith utterance and character TQL
Position-aware models
w/ APEWith absolute positional encoding
w/ P’UTT-TQLWith APE and utterance-level TQL
w/ P’CHAR-TQLWith APE and character-level TQL
w/ P’UTT-TQL + CHAR-TQLWith APE and combined TQL

Note(s): TQL- and position-aware models are trained on noisy transcriptions

Source(s): Authors’ own work

or Create an Account

Close subscription notice
Close access options