No abstract

No abstract

With the rapid development of speech assistants, adapting server-intended automatic speech recognition (ASR) solutions to a direct device has become crucial. For on-device speech recognition tasks, researchers and industry prefer end-to-end ASR systems as they can be made resource-efficient while maintaining a higher quality compared to hybrid systems. However, building end-to-end models requires a significant amount of speech data. Personalization, which is mainly handling out-of-vocabulary (OOV) words, is another challenging task associated with speech assistants. In this work, we consider building an effective end-to-end ASR system in low-resource setups with a high OOV rate, embodied in Babel Turkish and Babel Georgian tasks. We propose a method of dynamic acoustic unit augmentation based on the Byte Pair Encoding with dropout (BPE-dropout) technique. The method non-deterministically tokenizes utterances to extend the token’s contexts and to regularize their distribution for the model’s recognition of unseen words. It also reduces the need for optimal subword vocabulary size search. The technique provides a steady improvement in regular and personalized (OOV-oriented) speech recognition tasks (at least 6% relative word error rate (WER) and 25% relative F-score) at no additional computational cost. Owing to the BPE-dropout use, our monolingual Turkish Conformer has achieved a competitive result with 22.2% character error rate (CER) and 38.9% WER, which is close to the best published multilingual system.

No abstract

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.