arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2203.15249cs.SDeess.AS
收藏

MFA-Conformer:用于说话人自动确认的多尺度特征聚合方法

MFA-Conformer: Multi-scale Feature Aggregation Conformer for Automatic Speaker Verification

Tsinghua University(清华大学);Tencent Inc(腾讯公司);National Taiwan University(国立台湾大学);The Chinese University of Hong Kong(香港中文大学)

Yang Zhang, Zhiqiang Lv, Haibin Wu, Shanshan Zhang, Pengfei Hu, Zhiyong Wu, Hung-yi Lee, Helen Meng

摘要:

在本文中,我们提出了多尺度特征聚合一致性(MFA-Conformer),一个易于实现的,简单,但有效的骨干自动说话人确认的卷积增强的Transformer(Conformer)的基础上。MFA-Conformer的架构受到了语音识别和说话人验证中最新模型的启发。首先,我们引入卷积子采样层来降低模型的计算成本。其次,我们采用了Conformer模块,它结合了Transformers和卷积神经网络(CNN),以有效地捕获全局和局部特征。最后,来自所有Conformer块的输出特征图在最终池化之前被连接以聚合多尺度表示。我们在广泛使用的基准上评估MFA-Conformer。最佳系统在VoxCeleb 1-O、SITW.Dev和SITW.Eval集上分别获得0.64%、1.29%和1.63%的EER。MFA-Conformer在识别性能和推理速度方面都明显优于流行的ECAPA-TDNN系统。最后,消融研究清楚地表明,全局和局部特征学习的结合可以导致鲁棒和准确的说话人嵌入提取。我们还发布了代码以供将来比较。

英文摘要:

In this paper, we present Multi-scale Feature Aggregation Conformer (MFA-Conformer), an easy-to-implement, simple but effective backbone for automatic speaker verification based on the Convolution-augmented Transformer (Conformer). The architecture of the MFA-Conformer is inspired by recent stateof-the-art models in speech recognition and speaker verification. Firstly, we introduce a convolution subsampling layer to decrease the computational cost of the model. Secondly, we adopt Conformer blocks which combine Transformers and convolution neural networks (CNNs) to capture global and local features effectively. Finally, the output feature maps from all Conformer blocks are concatenated to aggregate multi-scale representations before final pooling. We evaluate the MFA-Conformer on the widely used benchmarks. The best system obtains 0.64%, 1.29% and 1.63% EER on VoxCeleb1-O, SITW.Dev, and SITW.Eval set, respectively. MFA-Conformer significantly outperforms the popular ECAPA-TDNN systems in both recognition performance and inference speed. Last but not the least, the ablation studies clearly demonstrate that the combination of global and local feature learning can lead to robust and accurate speaker embedding extraction. We have also released the code for future comparison.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑