论文标题
通过Umigram浅融合提高RNN-TransDucer的稀有单词的准确性
Improving accuracy of rare words for RNN-Transducer through unigram shallow fusion
论文作者
论文摘要
端到端的自动语音识别（ASR）系统，例如复发性神经网络传感器（RNN-T），已经流行，但罕见的词仍然是一个挑战。在本文中，我们提出了一种简单而有效的方法，称为Umigram浅融合（USF），以改善RNN-T的稀有词。在USF中，我们根据umigram数量从RNN-T训练数据中提取稀有单词，并在解码过程中遇到单词时应用固定的奖励。我们表明，这种简单的方法可以将稀有单词的性能提高3.7％的亲戚，而不会在一般测试集上降低，而USF的改进是基于任何基于语言的撤销的任何其他基于语言模型的添加剂。然后，我们表明同一USF在常规混合系统上不起作用。最后，我们认为，由于基于子字的RNN-T在解码过程中使用的viterbi搜索，USF通过修复概率估算中的概率估算中的错误来工作。
End-to-end automatic speech recognition (ASR) systems, such as recurrent neural network transducer (RNN-T), have become popular, but rare word remains a challenge. In this paper, we propose a simple, yet effective method called unigram shallow fusion (USF) to improve rare words for RNN-T. In USF, we extract rare words from RNN-T training data based on unigram count, and apply a fixed reward when the word is encountered during decoding. We show that this simple method can improve performance on rare words by 3.7% WER relative without degradation on general test set, and the improvement from USF is additive to any additional language model based rescoring. Then, we show that the same USF does not work on conventional hybrid system. Finally, we reason that USF works by fixing errors in probability estimates of words due to Viterbi search used during decoding with subword-based RNN-T.
