M-CIF: Multi-Scale Alignment For CIF-Based Non-Autoregressive ASR
Abstract
The Continuous Integrate-and-Fire (CIF) mechanism provides effective alignment for non-autoregressive (NAR) speech recognition. This mechanism creates a smooth and monotonic mapping from acoustic features to target tokens, achieving performance on Mandarin competitive with other NAR approaches. However, without finer-grained guidance, its stability degrades in some languages such as English and French. In this paper, we propose Multi-scale CIF (M-CIF), which performs multi-level alignment by integrating character and phoneme level supervision progressively distilled into subword representations, thereby enhancing robust acoustic-text alignment. Experiments show that M-CIF reduces WER compared to the Paraformer baseline, especially on CommonVoice by 4.21% in German and 3.05% in French. To further investigate these gains, we define phonetic confusion errors (PE) and space-related segmentation errors (SE) as evaluation metrics. Analysis of these metrics across different M-CIF settings reveals that the phoneme and character layers are essential for enhancing progressive CIF alignment.
Keywords
Cite
@article{arxiv.2510.22172,
title = {M-CIF: Multi-Scale Alignment For CIF-Based Non-Autoregressive ASR},
author = {Ruixiang Mao and Xiangnan Ma and Qing Yang and Ziming Zhu and Yucheng Qiao and Yuan Ge and Tong Xiao and Shengxiang Gao and Zhengtao Yu and Jingbo Zhu},
journal= {arXiv preprint arXiv:2510.22172},
year = {2025}
}