论文标题
E2E模型的基于似然比的域适应方法
A Likelihood Ratio based Domain Adaptation Method for E2E Models
论文作者
论文摘要
端到端（E2E）自动语音识别模型（例如经常性神经网络传感器（RNN-T））已成为流媒体ASR应用程序等流媒体应用程序的流行选择。尽管E2E模型非常有效地学习培训的培训数据，但它们对看不见的领域的准确性仍然是一个具有挑战性的问题。此外，这些模型需要配对的音频和文本培训数据，在计算上很昂贵，并且很难适应会话演讲的快速发展性质。在这项工作中，我们使用可能比率的可能性比例来探索一种上下文偏见方法，该方法利用文本数据源将RNN-T模型调整到新的域和实体中。我们表明，该方法有效地改善了稀有单词的识别，并且在多个多域外数据集中，在1-最佳单词错误率（WER）中相对提高了10％，而N-t-test Oracle WER（n = 8）中的相对提高为10％，而在一般数据集中则没有任何降解。我们还表明，通过第二频繁撤退模型的适应上下文偏见适应的补充可带来增强的改进。
End-to-end (E2E) automatic speech recognition models like Recurrent Neural Networks Transducer (RNN-T) are becoming a popular choice for streaming ASR applications like voice assistants. While E2E models are very effective at learning representation of the training data they are trained on, their accuracy on unseen domains remains a challenging problem. Additionally, these models require paired audio and text training data, are computationally expensive and are difficult to adapt towards the fast evolving nature of conversational speech. In this work, we explore a contextual biasing approach using likelihood-ratio that leverages text data sources to adapt RNN-T model to new domains and entities. We show that this method is effective in improving rare words recognition, and results in a relative improvement of 10% in 1-best word error rate (WER) and 10% in n-best Oracle WER (n=8) on multiple out-of-domain datasets without any degradation on a general dataset. We also show that complementing the contextual biasing adaptation with adaptation of a second-pass rescoring model gives additive WER improvements.
