# Non-Autoregressive Transformer ASR with CTC-Enhanced Decoder Input

arXiv: https://arxiv.org/abs/2010.15025 · HTML: https://mlsanity.com/paper/2010.15025/

## Key facts

- Cited by 24 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 28 Oct 2020.

## Built on by (most cited first, 24 of 24)

- [Paraformer: Fast and Accurate Parallel Transformer for Non-autoregressive End-to-End Speech Recognition](https://mlsanity.com/paper/2206.08317/) (2022)
- [Recent Advances in End-to-End Automatic Speech Recognition](https://mlsanity.com/paper/2111.01690/) (2021)
- [A Survey on Non-Autoregressive Generation for Neural Machine Translation and Beyond](https://mlsanity.com/paper/2204.09269/) (2022)
- [Wav-BERT: Cooperative Acoustic and Linguistic Representation Learning for Low-Resource Speech Recognition](https://mlsanity.com/paper/2109.09161/) (2021)
- [Non-autoregressive Transformer-based End-to-end ASR using BERT](https://mlsanity.com/paper/2104.04805/) (2021)
- [A context-aware knowledge transferring strategy for CTC-based ASR](https://mlsanity.com/paper/2210.06244/) (2022)
- [Fast-MD: Fast Multi-Decoder End-to-End Speech Translation with Non-Autoregressive Hidden Intermediates](https://mlsanity.com/paper/2109.12804/) (2021)
- [An Improved Single Step Non-autoregressive Transformer for Automatic Speech Recognition](https://mlsanity.com/paper/2106.09885/) (2021)
- [Consistent Training and Decoding For End-to-end Speech Recognition Using Lattice-free MMI](https://mlsanity.com/paper/2112.02498/) (2021)
- [Automatic Speech Recognition with BERT and CTC Transformers: A Review](https://mlsanity.com/paper/2410.09456/) (2024)
- [TSNAT: Two-Step Non-Autoregressvie Transformer Models for Speech Recognition](https://mlsanity.com/paper/2104.01522/) (2021)
- [4D ASR: Joint modeling of CTC, Attention, Transducer, and Mask-Predict decoders](https://mlsanity.com/paper/2212.10818/) (2022)
- [Decoupling recognition and transcription in Mandarin ASR](https://mlsanity.com/paper/2108.01129/) (2021)
- [Non-autoregressive Transformer with Unified Bidirectional Decoder for Automatic Speech Recognition](https://mlsanity.com/paper/2109.06684/) (2021)
- [Sub-8-Bit Quantization Aware Training for 8-Bit Neural Network Accelerator with On-Device Speech Recognition](https://mlsanity.com/paper/2206.15408/) (2022)
- [Fingerspelling PoseNet: Enhancing Fingerspelling Translation with Pose-Based Transformer Models](https://mlsanity.com/paper/2311.12128/) (2023)
- [Joint Beam Search Integrating CTC, Attention, and Transducer Decoders](https://mlsanity.com/paper/2406.02950/) (2024)
- [Adaptive Sparse and Monotonic Attention for Transformer-based Automatic Speech Recognition](https://mlsanity.com/paper/2209.15176/) (2022)
- [Dynamic Alignment Mask CTC: Improved Mask-CTC with Aligned Cross Entropy](https://mlsanity.com/paper/2303.07687/) (2023)
- [Towards Effective and Efficient Non-autoregressive Decoding Using Block-based Attention Mask](https://mlsanity.com/paper/2406.10034/) (2024)
- [Enhancing the Robustness of Contextual ASR to Varying Biasing Information Volumes Through Purified Semantic Correlation Joint Modeling](https://mlsanity.com/paper/2509.05908/) (2025)
- [Audio-Conditioned Diffusion LLMs for ASR and Deliberation Processing](https://mlsanity.com/paper/2509.16622/) (2025)
- [Pay Attention to CTC: Fast and Robust Pseudo-Labelling for Unified Speech Recognition](https://mlsanity.com/paper/2602.19316/) (2026)
- [A Lexical-aware Non-autoregressive Transformer-based ASR Model](https://mlsanity.com/paper/2305.10839/) (2023)

## Abstract (arXiv)

Non-autoregressive (NAR) transformer models have achieved significantly inference speedup but at the cost of inferior accuracy compared to autoregressive (AR) models in automatic speech recognition (ASR). Most of the NAR transformers take a fixed-length sequence filled with MASK tokens or a redundant sequence copied from encoder states as decoder input, they cannot provide efficient target-side information thus leading to accuracy degradation. To address this problem, we propose a CTC-enhanced NAR transformer, which generates target sequence by refining predictions of the CTC module. Experimental results show that our method outperforms all previous NAR counterparts and achieves 50x faster decoding speed than a strong AR baseline with only 0.0 ~ 0.3 absolute CER degradation on Aishell-1 and Aishell-2 datasets.