研究人员开发了一种新颖的方法，将轻量级语音识别模型（如Moonshine）应用于孟加拉语等形态丰富的语言。识别出的核心问题是面向英语的tokenizer导致了自回归崩溃。通过将BanglaBERT WordPiece词汇表移植到解码器中，模型的token肥沃度显著降低，缓解了解码不稳定性。该方法在Lipi-Ghor数据集上实现了21.54%的具有竞争力的词错误率，为ASR模型在不进行大量重新训练的情况下进行跨脚本适应提供了可扩展的解决方案。 AI
影响 使得语音识别在边缘设备上更有效地部署到非拉丁语种。
排序理由 该集群包含一篇学术论文，详细介绍了改进特定语言ASR模型性能的新颖技术方法。
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
arXiv:2607.09598v1 Announce Type: new Abstract: 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 …
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 b…