TY - GEN
T1 - Language Model Integration Based on Memory Control for Sequence to Sequence Speech Recognition
AU - Cho, Jaejin
AU - Watanabe, Shinji
AU - Hori, Takaaki
AU - Baskar, Murali Karthick
AU - Inaguma, Hirofumi
AU - Villalba, Jesus
AU - Dehak, Najim
N1 - Funding Information:
The work reported here was started during JSALT 2018, and supported by JHU with gifts from Amazon, Facebook, Google, Microsoft and Mitsubishi Electric.
PY - 2019/5
Y1 - 2019/5
N2 - In this paper, we explore several new schemes to train a seq2seq model to integrate a pre-trained language model (LM). Our proposed fusion methods focus on the memory cell state and the hidden state in the seq2seq decoder long short-term memory (LSTM), and the memory cell state is updated by the LM unlike the prior studies. This means the memory retained by the main seq2seq would be adjusted by the external LM. These fusion methods have several variants depending on the architecture of this memory cell update and the use of memory cell and hidden states which directly affects the final label inference. We performed the experiments to show the effectiveness of the proposed methods in a mono-lingual ASR setup on the Librispeech corpus and in a transfer learning setup from a multilingual ASR (MLASR) base model to a low-resourced language. In Librispeech, our best model improved WER by 3.7%, 2.4% for test clean, test other relatively to the shallow fusion baseline, with multilevel decoding. In transfer learning from an MLASR base model to the IARPA Babel Swahili model, the best scheme improved the transferred model on eval set by 9.9%, 9.8% in CER, WER relatively to the 2-stage transfer baseline.
AB - In this paper, we explore several new schemes to train a seq2seq model to integrate a pre-trained language model (LM). Our proposed fusion methods focus on the memory cell state and the hidden state in the seq2seq decoder long short-term memory (LSTM), and the memory cell state is updated by the LM unlike the prior studies. This means the memory retained by the main seq2seq would be adjusted by the external LM. These fusion methods have several variants depending on the architecture of this memory cell update and the use of memory cell and hidden states which directly affects the final label inference. We performed the experiments to show the effectiveness of the proposed methods in a mono-lingual ASR setup on the Librispeech corpus and in a transfer learning setup from a multilingual ASR (MLASR) base model to a low-resourced language. In Librispeech, our best model improved WER by 3.7%, 2.4% for test clean, test other relatively to the shallow fusion baseline, with multilevel decoding. In transfer learning from an MLASR base model to the IARPA Babel Swahili model, the best scheme improved the transferred model on eval set by 9.9%, 9.8% in CER, WER relatively to the 2-stage transfer baseline.
KW - Automatic speech recognition (ASR)
KW - cold fusion
KW - deep fusion
KW - language model
KW - sequence to sequence
KW - shallow fusion
UR - http://www.scopus.com/inward/record.url?scp=85069000387&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85069000387&partnerID=8YFLogxK
U2 - 10.1109/ICASSP.2019.8683380
DO - 10.1109/ICASSP.2019.8683380
M3 - Conference contribution
AN - SCOPUS:85069000387
T3 - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
SP - 6191
EP - 6195
BT - 2019 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Proceedings
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 44th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019
Y2 - 12 May 2019 through 17 May 2019
ER -