Listen Attentively, And Spell Once: Whole Sentence Generation Via A Non-autoregressive Architecture For Low-latency Speech Recognition

Abstract

Although attention based end-to-end models have achieved promising performance in speech recognition, the multi-pass forward computation in beam-search increases inference time cost, which limits their practical applications. To address this issue, we propose a non-autoregressive end-to-end speech recognition system called LASO (listen attentively, and spell once). Because of the non-autoregressive property, LASO predicts a textual token in the sequence without the dependence on other tokens. Without beam-search, the one-pass propagation much reduces inference time cost of LASO. And because the model is based on the attention based feedforward structure, the computation can be implemented in parallel efficiently. We conduct experiments on publicly available Chinese dataset AISHELL-1. LASO achieves a character error rate of 6.4%, which outperforms the state-of-the-art autoregressive transformer model (6.7%). The average inference latency is 21 ms, which is 1/50 of the autoregressive tra

Stats

• citations21
• S2 citations—
• github stars0
• HF likes0
• heat score10.07
• arxiv keybai2020listen

Related papers

• An Online Attention-based Model For Speech Recognition (2018)9.59
• Fast End-to-end Speech Recognition Via Non-autoregressive Models And Cross-modal Knowledge Transferring From BERT (2021)12.93
• State-of-the-art Speech Recognition With Sequence-to-sequence Models (2017)21.01
• Two-pass End-to-end Speech Recognition (2019)13.97
• Attention-based Sequence-to-sequence Model For Speech Recognition: Development Of State-of-the-art System On Librispeech And Its Application To Non-native English (2018)0.00
• High Quality Streaming Speech Synthesis With Low, Sentence-length-independent Latency (2021)8.60
• Effectiveasr: A Single-step Non-autoregressive Mandarin Speech Recognition Architecture With High Accuracy And Inference Speed (2024)3.58
• Audio-attention Discriminative Language Model For ASR Rescoring (2019)9.23