Audio-attention discriminative language model for ASR rescoring
This work addresses domain adaptation for automatic speech recognition systems, offering a practical solution for production environments, though it is incremental as it builds on existing rescoring techniques.
The paper tackles the challenge of adapting end-to-end ASR systems to new domains with limited data by proposing an attention-based discriminative language model for rescoring first-pass ASR outputs, resulting in an 8% improvement in word error rate using only a fraction of the training data.
End-to-end approaches for automatic speech recognition (ASR) benefit from directly modeling the probability of the word sequence given the input audio stream in a single neural network. However, compared to conventional ASR systems, these models typically require more data to achieve comparable results. Well-known model adaptation techniques, to account for domain and style adaptation, are not easily applicable to end-to-end systems. Conventional HMM-based systems, on the other hand, have been optimized for various production environments and use cases. In this work, we propose to combine the benefits of end-to-end approaches with a conventional system using an attention-based discriminative language model that learns to rescore the output of a first-pass ASR system. We show that learning to rescore a list of potential ASR outputs is much simpler than learning to generate the hypothesis. The proposed model results in 8% improvement in word error rate even when the amount of training data is a fraction of data used for training the first-pass system.