Recently, there has been an increasing interest in unifying streaming and non-streaming speech recognition models to reduce development, training and deployment cost. The best-known approaches rely on either window-based or dynamic chunk-based attention strategy and causal convolutions to minimize the degradation due to streaming. However, the performance gap still remains relatively large between non-streaming and a full-contextual model trained independently. To address this, we propose a dynamic chunk-based convolution replacing the causal convolution in a hybrid Connectionist Temporal Classification (CTC)-Attention Conformer architecture. Additionally, we demonstrate further improvements through initialization of weights from a full-contextual model and parallelization of the convolution and self-attention modules. We evaluate our models on the open-source Voxpopuli, LibriSpeech and in-house conversational datasets. Overall, our proposed model reduces the degradation of the streaming mode over the non-streaming full-contextual model from 41.7% and 45.7% to 16.7% and 26.2% on the LibriSpeech test-clean and test-other datasets respectively, while improving by a relative 15.5% WER over the previous state-of-the-art unified model.

翻译：近年来，统一流式与非流式语音识别模型以减少开发、训练和部署成本的研究日益受到关注。最知名的方法依赖于基于窗口或动态分块的注意力策略以及因果卷积来最小化流式化带来的性能退化。然而，非流式模型与独立训练的全上下文模型之间仍存在较大的性能差距。为解决这一问题，我们提出在混合连接时序分类（CTC）-注意力Conformer架构中，用动态分块卷积替代因果卷积。此外，我们通过从全上下文模型初始化权重以及并行化卷积与自注意力模块，进一步提升了性能。我们在开源Voxpopuli、LibriSpeech以及内部对话数据集上评估了模型。总体而言，我们所提出的模型将流式模式相较于非流式全上下文模型的性能退化，在LibriSpeech test-clean和test-other数据集上分别从41.7%和45.7%降至16.7%和26.2%，同时相较于此前最优的统一模型实现了15.5%的词错误率（WER）相对降低。