arXiv Open Access 2024

Romanization Encoding For Multilingual ASR

Wen Ding Fei Jia Hainan Xu Yu Xi Junjie Lai +1 lainnya
Lihat Sumber

Abstrak

We introduce romanization encoding for script-heavy languages to optimize multilingual and code-switching Automatic Speech Recognition (ASR) systems. By adopting romanization encoding alongside a balanced concatenated tokenizer within a FastConformer-RNNT framework equipped with a Roman2Char module, we significantly reduce vocabulary and output dimensions, enabling larger training batches and reduced memory consumption. Our method decouples acoustic modeling and language modeling, enhancing the flexibility and adaptability of the system. In our study, applying this method to Mandarin-English ASR resulted in a remarkable 63.51% vocabulary reduction and notable performance gains of 13.72% and 15.03% on SEAME code-switching benchmarks. Ablation studies on Mandarin-Korean and Mandarin-Japanese highlight our method's strong capability to address the complexities of other script-heavy languages, paving the way for more versatile and effective multilingual ASR systems.

Topik & Kata Kunci

Penulis (6)

W

Wen Ding

F

Fei Jia

H

Hainan Xu

Y

Yu Xi

J

Junjie Lai

B

Boris Ginsburg

Format Sitasi

Ding, W., Jia, F., Xu, H., Xi, Y., Lai, J., Ginsburg, B. (2024). Romanization Encoding For Multilingual ASR. https://arxiv.org/abs/2407.04368

Akses Cepat

Lihat di Sumber
Informasi Jurnal
Tahun Terbit
2024
Bahasa
en
Sumber Database
arXiv
Akses
Open Access ✓