# Non-Autoregressive ASR with Self-Conditioned Folded Encoders

arXiv: https://arxiv.org/abs/2202.08474 · HTML: https://mlsanity.com/paper/2202.08474/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 17 Feb 2022.

## Built on by (most cited first, 5 of 5)

- [A Survey on Non-Autoregressive Generation for Neural Machine Translation and Beyond](https://mlsanity.com/paper/2204.09269/) (2022)
- [Latent Iterative Refinement for Modular Source Separation](https://mlsanity.com/paper/2211.11917/) (2022)
- [Parameter-Efficient Conformers via Sharing Sparsely-Gated Experts for End-to-End Speech Recognition](https://mlsanity.com/paper/2209.08326/) (2022)
- [Improving CTC-AED model with integrated-CTC and auxiliary loss regularization](https://mlsanity.com/paper/2308.08449/) (2023)
- [Overcoming Decoder Inconsistencies in Whisper for Dravidian and Low-Resource Languages](https://mlsanity.com/paper/2606.09535/) (2026)

## Abstract (arXiv)

This paper proposes CTC-based non-autoregressive ASR with self-conditioned folded encoders. The proposed method realizes non-autoregressive ASR with fewer parameters by folding the conventional stack of encoders into only two blocks; base encoders and folded encoders. The base encoders convert the input audio features into a neural representation suitable for recognition. This is followed by the folded encoders applied repeatedly for further refinement. Applying the CTC loss to the outputs of all encoders enforces the consistency of the input-output relationship. Thus, folded encoders learn to perform the same operations as an encoder with deeper distinct layers. In experiments, we investigate how to set the number of layers and the number of iterations for the base and folded encoders. The results show that the proposed method achieves a performance comparable to that of the conventional method using only 38% as many parameters. Furthermore, it outperforms the conventional method when increasing the number of iterations.