论文标题
学习基于唇唇的音频扬声器嵌入与Av-Hubert
Learning Lip-Based Audio-Visual Speaker Embeddings with AV-HuBERT
论文作者
论文摘要
本文研究了视听扬声器表示的自我监督的预训练，其中显示了视觉流，显示说话者的口腔区域与语音一起用作输入。我们的研究重点是视听隐藏单元BERT（AV-HUBERT）方法，该方法是最近开发的通用音频语音训练前训练框架。我们进行了广泛的实验，以探测预训练和视觉方式的有效性。实验结果表明，AV-Hubert概括地概括了与说话者相关的下游任务，从而使标签效率提高了大约10倍的范围，仅有音频和视听扬声器验证。我们还表明，结合视觉信息，甚至仅仅是唇部区域，都大大提高了性能和噪声稳健性，在清洁条件下将EER降低了38％，在嘈杂的条件下将EER降低了75％。
This paper investigates self-supervised pre-training for audio-visual speaker representation learning where a visual stream showing the speaker's mouth area is used alongside speech as inputs. Our study focuses on the Audio-Visual Hidden Unit BERT (AV-HuBERT) approach, a recently developed general-purpose audio-visual speech pre-training framework. We conducted extensive experiments probing the effectiveness of pre-training and visual modality. Experimental results suggest that AV-HuBERT generalizes decently to speaker related downstream tasks, improving label efficiency by roughly ten fold for both audio-only and audio-visual speaker verification. We also show that incorporating visual information, even just the lip area, greatly improves the performance and noise robustness, reducing EER by 38% in the clean condition and 75% in noisy conditions.
