用本地词表替换英文分词器，解决小模型语音识别崩溃问题
Tokenizer Transplantation: Mitigating Autoregressive Collapse in Edge-Efficient Bengali ASR
- 将英文化分词器换成孟加拉语专用词表，降低分词碎片化
- 分词频率从9.16降至1.30，推理序列长度减少85.8%
- 在882小时数据上实现21.54%错误率，适合边缘设备部署
轻量级语音识别模型对边缘部署至关重要，但如Moonshine等高度优化的架构在语法复杂的非拉丁语言（如孟加拉语）上表现不佳。本研究发现其根源在于模型采用英语中心的字节级分词器，导致孟加拉语单词被切分为高密度字节链，引发推理时灾难性自回归崩溃。为此，提出一种新型词汇移植流程：将解码器词汇表替换为本地脚本的BanglaBERT WordPiece词汇，并调整对应词嵌入矩阵。实验显示，词频从9.16降至1.30；自回归序列长度减少85.8%，完全消除解码不稳定性。在882小时的Lipi-Ghor数据集上，改进后的模型达到21.54%的词错误率（WER），实时因子（RTF）为0.0053。该研究提供了一种无需资源密集型预训练的可扩展、可复现的跨语言适配小型语音识别模型的方法。
原文摘要 · Abstract (English)
Lightweight speech recognition models are critical for edge deployment, yet highly optimized architectures like Moonshine often fail on morphologically rich, non-Latin languages such as Bengali. This study identifies the root cause of this failure as the model's English-centric byte-level tokenizer, which fragments Bengali words into high-fertility byte chains and triggers catastrophic autoregressive collapse during inference. To resolve this, a novel vocabulary transplantation pipeline is proposed to replace the decoder vocabulary with the native-script BanglaBERT WordPiece vocabulary and resize the corresponding token embedding matrix. Experimental results demonstrate a reduction in token fertility from 9.16 to 1.30. By decreasing autoregressive sequence length by 85.8%, decoding instability is entirely mitigated. When evaluated on the 882-hour Lipi-Ghor dataset, the modified architecture achieves a competitive 21.54% Word Error Rate (WER) and a Real-Time Factor (RTF) of 0.0053. Ultimately, this research provides a scalable, reproducible blueprint for cross-script adaptation of compact ASR models without the need for resource-intensive pre-training.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。