Improving accuracy of rare words for RNN-Transducer through unigram shallow fusion
This is an incremental improvement for ASR systems, specifically benefiting users who encounter rare words in speech recognition tasks.
This paper addresses the challenge of rare word recognition in end-to-end automatic speech recognition (ASR) systems, specifically RNN-Transducers. The authors propose unigram shallow fusion (USF), which applies a fixed reward to rare words during decoding, resulting in a 3.7% relative improvement in word error rate (WER) for rare words without degrading overall performance.
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.