• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文学位 >Some important algorithms for large-aperture microphone arrays: Calibration and determination of talker orientation.
【24h】

Some important algorithms for large-aperture microphone arrays: Calibration and determination of talker orientation.

机译：大口径麦克风阵列的一些重要算法：校准和确定讲话者的方位。

获取原文
获取原文并翻译 | 示例
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Microphone arrays are used to acquire speech data for a variety of signal processing applications. These applications include teleconferencing, video conferencing, speech recognition, source tracking and source identification. Microphone arrays are often used instead of a close-talking microphone (i.e., a head-mounted microphone) since arrays can focus on multiple talkers and do not impede a talker's movement. Not only can a microphone array enhance the signal from an acoustic source but it also can attenuate the signals from competing sources. Microphone arrays are also used to estimate the location of a source which is within the array's environment. The estimated source positions can be used in video conferencing applications to steer a camera or in acoustic enhancement algorithms to focus on a source. Location estimation algorithms and most beamforming algorithms require that the relative locations of the array's microphones be known. A beamforming algorithm is used to "aim" a microphone array at a particular source.; This dissertation presents new research and experimental results for two important algorithmic areas for a large-aperture microphone array. First, an array needs to be calibrated---the microphone positions and relative channel gains need to be known. A new acoustic method is presented here and is shown to be superior to the other methods tried and/or published to date. Second; many array algorithms assume that a source's signal emanates spherically from a point. This assumption, unfortunately, is inaccurate for both a human talker or even a transducer. The radiation varies with the source orientation. Therefore, if the orientation of a source is known, many algorithms can be altered to yield more accurate results. This dissertation presents three entirely acoustic methods for finding talker orientation in a real room environment. Two of the methods derive their algorithmic clues from the time-domain while the third method uses a measure from the frequency-domain. The theory for each method is presented and the results from experiments in a real environment indicate their effectiveness.
机译：麦克风阵列用于为各种信号处理应用程序获取语音数据。这些应用程序包括电话会议，视频会议，语音识别，源跟踪和源标识。由于阵列可以集中在多个通话者上并且不会妨碍通话者的移动，因此通常使用麦克风阵列来代替近距离通话的麦克风（即，头戴式麦克风）。麦克风阵列不仅可以增强来自声源的信号，而且可以衰减来自竞争源的信号。麦克风阵列还用于估计阵列环境中的信号源位置。估计的声源位置可用于视频会议应用中，以操纵摄像机，或用于声学增强算法中，以聚焦于声源上。位置估计算法和大多数波束成形算法要求知道阵列麦克风的相对位置。波束成形算法用于将麦克风阵列“对准”特定的源。本文为大口径麦克风阵列的两个重要算法领域提供了新的研究和实验结果。首先，需要校准阵列-麦克风位置和相对通道增益需要知道。本文介绍了一种新的声学方法，它被证明优于迄今尝试和/或发布的其他方法。第二;许多阵列算法都假定源的信号从一个点发出球形。不幸的是，这个假设对于人类说话者甚至是换能器都是不准确的。辐射随光源方向而变化。因此，如果知道光源的方向，则可以更改许多算法以产生更准确的结果。本文提出了三种在真实房间环境中寻找说话者方位的完全声学方法。其中两种方法是从时域获得算法线索的，而第三种方法是从频域获得度量的。介绍了每种方法的理论，并在实际环境中的实验结果表明了它们的有效性。

著录项

  • 作者

    Sachar, Joshua Marc.;

  • 作者单位

    Brown University.;

  • 授予单位 Brown University.;
  • 学科 Engineering Electronics and Electrical.
  • 学位 Ph.D.
  • 年度 2004
  • 页码 158 p.
  • 总页数 158
  • 原文格式 PDF
  • 正文语种 eng
  • 中图分类 无线电电子学、电信技术;
  • 关键词

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号