Modular Hybrid Autoregressive Transducer

• Published in: 2022 IEEE Spoken Language Technology Workshop (SLT) pp. 197 - 204
• Main Authors: Meng, Zhong, Chen, Tongzhou, Prabhavalkar, Rohit, Zhang, Yu, Wang, Gary, Audhkhasi, Kartik, Emond, Jesse, Strohman, Trevor, Ramabhadran, Bhuvana, Huang, W. Ronny, Variani, Ehsan, Huang, Yinghui, Moreno, Pedro J.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 09.01.2023
• Subjects: Acoustics; Adaptation models; Conferences; Decoding; hybrid autoregressive transducer; Production; Speech recognition; text-only adaptation; Transducers

Text-only adaptation of a transducer model remains challenging for end-to-end speech recognition since the transducer has no clearly separated acoustic model (AM), language model (LM) or blank model. In this work, we propose a modular hybrid autoregressive transducer (MHAT) that has structurally sep...