arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 1712.04276cs.SDeess.ASstat.ML
收藏

基于噪声训练卷积神经网络的多说话人定位

Multi-Speaker Localization Using Convolutional Neural Network Trained with Noise

Friedrich-Alexander-University Erlangen-Nürnberg(埃尔朗根-纽伦堡弗里德里希-亚历山大大学);Fraunhofer Institute for Integrated Circuits(弗劳恩霍夫集成电路研究所);International Audio Laboratories(国际音频实验室)

Soumitro Chakrabarty, Emanuël A. P. Habets

摘要:

将多说话人定位问题转化为多类多标签分类问题,并采用基于卷积神经网络(CNN)的声源定位方法进行求解。利用不相交的说话人活动的共同假设,我们提出了一种新的方法来训练的CNN使用合成的噪声信号。所提出的定位方法进行评估,为两个扬声器和一个众所周知的转向响应功率法相比。

英文摘要:

The problem of multi-speaker localization is formulated as a multi-class multi-label classification problem, which is solved using a convolutional neural network (CNN) based source localization method. Utilizing the common assumption of disjoint speaker activities, we propose a novel method to train the CNN using synthesized noise signals. The proposed localization method is evaluated for two speakers and compared to a well-known steered response power method.

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。

↑