MUST：一种用于低资源语音识别的多语言学生-教师学习方法
MUST: A Multilingual Student-Teacher Learning approach for low-resource speech recognition
AI总结：
本文提出MUST多语言学生-教师学习方法，通过后验概率映射克服知识蒸馏中字符集不一致的限制，利用教师集成训练低资源语音识别模型，使相对字符错误率最多降低9.5%。
摘要：
学生-教师学习或知识蒸馏（KD）此前已被用于解决语音识别（ASR）系统训练中的数据稀缺问题。然而，KD训练的一个局限在于，学生模型的类别必须是教师模型类别的真子集或非真子集。如果字符集不同，即使声学上相似的语言之间也无法进行蒸馏。本文通过提出一种多语言学生-教师（MUST）学习方法来解决上述局限，该方法利用后验概率映射方法。使用一个预训练的映射模型将教师语言的后验概率映射到学生语言的ASR。这些映射后的后验概率被用作KD学习的软标签。实验了多种教师集成方案，以训练低资源语言的ASR模型。与基线单语ASR相比，使用MUST学习训练的模型将相对字符错误率（CER）降低了最多9.5%。
英文摘要：
Student-teacher learning or knowledge distillation (KD) has been previously used to address data scarcity issue for training of speech recognition (ASR) systems. However, a limitation of KD training is that the student model classes must be a proper or improper subset of the teacher model classes. It prevents distillation from even acoustically similar languages if the character sets are not same. In this work, the aforementioned limitation is addressed by proposing a MUltilingual Student-Teacher (MUST) learning which exploits a posteriors mapping approach. A pre-trained mapping model is used to map posteriors from a teacher language to the student language ASR. These mapped posteriors are used as soft labels for KD learning. Various teacher ensemble schemes are experimented to train an ASR model for low-resource languages. A model trained with MUST learning reduces relative character error rate (CER) up to 9.5% in comparison with a baseline monolingual ASR.
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。
