Enhancing the Unified Streaming and Non-streaming Model with Contrastive Learning
This work addresses accuracy improvements for speech recognition systems, but it is incremental as it builds on existing unified models.
The paper tackles the problem of improving accuracy in unified streaming and non-streaming speech recognition models by using contrastive learning to bridge the representation gap between modes, achieving a CER of 4.66% in streaming mode and 4.31% in non-streaming mode on the AISHELL-1 benchmark.
The unified streaming and non-streaming speech recognition model has achieved great success due to its comprehensive capabilities. In this paper, we propose to improve the accuracy of the unified model by bridging the inherent representation gap between the streaming and non-streaming modes with a contrastive objective. Specifically, the top-layer hidden representation at the same frame of the streaming and non-streaming modes are regarded as a positive pair, encouraging the representation of the streaming mode close to its non-streaming counterpart. The multiple negative samples are randomly selected from the rest frames of the same sample under the non-streaming mode. Experimental results demonstrate that the proposed method achieves consistent improvements toward the unified model in both streaming and non-streaming modes. Our method achieves CER of 4.66% in the streaming mode and CER of 4.31% in the non-streaming mode, which sets a new state-of-the-art on the AISHELL-1 benchmark.