Saved in:
Bibliographic Details
Main Authors: Kim, Hyunsik, Kim, Haeri, Lee, Munhak, Lee, Kyungmin
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2602.01717
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917241379880960
author Kim, Hyunsik
Kim, Haeri
Lee, Munhak
Lee, Kyungmin
author_facet Kim, Hyunsik
Kim, Haeri
Lee, Munhak
Lee, Kyungmin
contents Multilingual automatic speech recognition (ASR) requires tokenization that efficiently covers many writing systems. Byte-level BPE (BBPE) using UTF-8 is widely adopted for its language-agnostic design and full Unicode coverage, but its variable-length encoding inflates token sequences for non-Latin scripts, such as Chinese, Japanese, and Korean (CJK). Longer sequences increase computational load and memory use. We propose BBPE16, a UTF-16-based BBPE tokenizer that represents most modern scripts with a uniform 2-byte code unit. BBPE16 preserves BBPE's language-agnostic properties while substantially improving cross-lingual token sharing. Across monolingual, bilingual, and trilingual ASR, and in a multilingual continual-learning setup, BBPE16 attains comparable or better accuracy; for Chinese, it reduces token counts by up to 10.4% and lowers decoding iterations by up to 10.3%. These reductions speed up fine-tuning and inference and decrease memory usage, making BBPE16 a practical tokenization choice for multilingual ASR.
format Preprint
id arxiv_https___arxiv_org_abs_2602_01717
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle BBPE16: UTF-16-based byte-level byte-pair encoding for improved multilingual speech recognition
Kim, Hyunsik
Kim, Haeri
Lee, Munhak
Lee, Kyungmin
Computation and Language
Artificial Intelligence
Multilingual automatic speech recognition (ASR) requires tokenization that efficiently covers many writing systems. Byte-level BPE (BBPE) using UTF-8 is widely adopted for its language-agnostic design and full Unicode coverage, but its variable-length encoding inflates token sequences for non-Latin scripts, such as Chinese, Japanese, and Korean (CJK). Longer sequences increase computational load and memory use. We propose BBPE16, a UTF-16-based BBPE tokenizer that represents most modern scripts with a uniform 2-byte code unit. BBPE16 preserves BBPE's language-agnostic properties while substantially improving cross-lingual token sharing. Across monolingual, bilingual, and trilingual ASR, and in a multilingual continual-learning setup, BBPE16 attains comparable or better accuracy; for Chinese, it reduces token counts by up to 10.4% and lowers decoding iterations by up to 10.3%. These reductions speed up fine-tuning and inference and decrease memory usage, making BBPE16 a practical tokenization choice for multilingual ASR.
title BBPE16: UTF-16-based byte-level byte-pair encoding for improved multilingual speech recognition
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2602.01717