English(EN) Identifying and typifying demographic unfairness in phoneme-level embeddings of self-supervised speech recognition models

新框架识别语音识别模型中的人口统计学不公平性

作者 PulseAugur 编辑部 · [2 个来源] · 2026-04-24 15:03

一项新的研究论文识别了两种导致语音识别模型人口统计学不公平性的错误类型——随机方差和系统性偏差。研究发现，虽然这两种错误类型都存在，但随机错误似乎是公平性更重要的障碍。有趣的是，使用增强公平性的算法对模型进行微调，并未改变域内探针训练的益处或随机嵌入错误的测量水平。 AI

影响识别出ASR系统中的关键错误类型，可能指导未来研究朝着更公平的语音技术发展。

排序理由学术论文，详细介绍了分析语音识别模型中人口统计学不公平性的框架。

AI 生成摘要 · Google Gemini · 来自 2 个来源。我们如何撰写摘要 →

报道来源 [2]

arXiv cs.CL TIER_1 English(EN) · Felix Herron, Solange Rossato, Alexandre Allauzen, Fran\c{c}ois Portet · 2026-04-27 04:00

Identifying and typifying demographic unfairness in phoneme-level embeddings of self-supervised speech recognition models

arXiv:2604.22631v1 Announce Type: new Abstract: Modern automatic speech recognition (ASR) systems have been observed to function better for certain speaker groups (SGs) than others, despite recent gains in overall performance. One potential impediment to progress towards fairer A…
arXiv cs.CL TIER_1 English(EN) · François Portet · 2026-04-24 15:03

Identifying and typifying demographic unfairness in phoneme-level embeddings of self-supervised speech recognition models

Modern automatic speech recognition (ASR) systems have been observed to function better for certain speaker groups (SGs) than others, despite recent gains in overall performance. One potential impediment to progress towards fairer ASR is a more nuanced understanding of the types …