面向汉语普通话发音错误检测与诊断的音高感知 RNN-T
Pitch-Aware RNN-T for Mandarin Chinese Mispronunciation Detection and Diagnosis
AI总结：
本文提出一种音高感知的无状态 RNN-T 模型用于汉语普通话发音错误检测与诊断，通过音高融合模块结合 HuBERT 特征，仅用母语者数据训练即在非母语场景下显著优于基线。
摘要：
利用自动语音识别（ASR）的发音错误检测与诊断（MDD）系统在汉语普通话中面临两大挑战：1）两阶段模型在音素或声调分类阶段与 MDD 阶段之间造成信息鸿沟。2）汉语 MDD 数据集的稀缺限制了模型训练。本文提出一种用于汉语 MDD 的无状态 RNN-T 模型，通过音高融合模块（Pitch Fusion Block）利用带有音高嵌入的 HuBERT 特征。该模型仅使用母语者数据训练，在非母语场景下与最先进的基线相比，音素错误率改善了 3%，错误接受率提高了 7%。
英文摘要：
Mispronunciation Detection and Diagnosis (MDD) systems, leveraging Automatic Speech Recognition (ASR), face two main challenges in Mandarin Chinese: 1) The two-stage models create an information gap between the phoneme or tone classification stage and the MDD stage. 2) The scarcity of Mandarin MDD datasets limits model training. In this paper, we introduce a stateless RNN-T model for Mandarin MDD, utilizing HuBERT features with pitch embedding through a Pitch Fusion Block. Our model, trained solely on native speaker data, shows a 3% improvement in Phone Error Rate and a 7% increase in False Acceptance Rate over the state-of-the-art baseline in non-native scenarios
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
