LT-LM: a novel non-autoregressive language model for single-shot lattice rescoring
This addresses the speed bottleneck in ASR systems for real-time applications, though it is incremental as it focuses on efficiency rather than accuracy breakthroughs.
The paper tackles the computational expense of autoregressive language models in ASR rescoring by proposing LT-LM, a non-autoregressive model that processes entire lattices in a single call, achieving over 300 times faster rescoring than existing methods with a slight increase in WER.
Neural network-based language models are commonly used in rescoring approaches to improve the quality of modern automatic speech recognition (ASR) systems. Most of the existing methods are computationally expensive since they use autoregressive language models. We propose a novel rescoring approach, which processes the entire lattice in a single call to the model. The key feature of our rescoring policy is a novel non-autoregressive Lattice Transformer Language Model (LT-LM). This model takes the whole lattice as an input and predicts a new language score for each arc. Additionally, we propose the artificial lattices generation approach to incorporate a large amount of text data in the LT-LM training process. Our single-shot rescoring performs orders of magnitude faster than other rescoring methods in our experiments. It is more than 300 times faster than pruned RNNLM lattice rescoring and N-best rescoring while slightly inferior in terms of WER.