# Paraformer-v2: An improved non-autoregressive transformer for noise-robust speech recognition

arXiv: https://arxiv.org/abs/2409.17746 · HTML: https://mlsanity.com/paper/2409.17746/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 26 Sep 2024.

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

- [Fun-ASR Technical Report](https://mlsanity.com/paper/2509.12508/) (2025)
- [CleanS2S: Single-file Framework for Proactive Speech-to-Speech Interaction](https://mlsanity.com/paper/2506.01268/) (2025)
- [Qwen-Audio-3.0-ASR Technical Report](https://mlsanity.com/paper/2609.07549/) (2026)
- [UMA-Split: unimodal aggregation for both English and Mandarin non-autoregressive speech recognition](https://mlsanity.com/paper/2509.14653/) (2025)
- [Text-Utilization for Encoder-dominated Speech Recognition Models](https://mlsanity.com/paper/2604.26514/) (2026)

## Abstract (arXiv)

Attention-based encoder-decoder, e.g. transformer and its variants, generates the output sequence in an autoregressive (AR) manner. Despite its superior performance, AR model is computationally inefficient as its generation requires as many iterations as the output length. In this paper, we propose Paraformer-v2, an improved version of Paraformer, for fast, accurate, and noise-robust non-autoregressive speech recognition. In Paraformer-v2, we use a CTC module to extract the token embeddings, as the alternative to the continuous integrate-and-fire module in Paraformer. Extensive experiments demonstrate that Paraformer-v2 outperforms Paraformer on multiple datasets, especially on the English datasets (over 14% improvement on WER), and is more robust in noisy environments.