首页> 外国专利> METHOD FOR SEARCHING A NOISE MODEL IN NOISY SOUND SIGNALS

METHOD FOR SEARCHING A NOISE MODEL IN NOISY SOUND SIGNALS

机译:噪声信号中噪声模型的搜索方法

摘要

More precisely it is related to noise model in voice signal the invention mainly relates in the noisy environment to noise reduction voice signal, such as in the cockpit of pilot or any other vehicle. This method is to digitize input and handles these signals from noise model to eliminate corresponding noise model as much as possible. Input gates each sampling of continuous pattern P, the noise model of iterative search for good and all carries out input signal itself, search for the noise that N number of continuous pattern (the minimum value N1 and maximum value N2 between N) has expected characteristics, it can be used for noise reduction process input signal by the corresponding samples of storage NxP to constitute noise model, and a new noise model and the storage new noise model found by search to replace previous one or keep previous two models according to reapective features. The model obtains finding what energy content by N number of mode (ratio between energy content is two value S 1/S) closer to each other.
机译:更确切地说,它与语音信号中的噪声模型有关,本发明主要在嘈杂的环境中与噪声降低的语音信号有关,例如在飞行员或任何其他车辆的驾驶舱中。这种方法是将输入数字化并处理来自噪声模型的这些信号,以尽可能消除相应的噪声模型。输入门对连续模式P进行每次采样,迭代搜索好噪声模型并全部执行输入信号本身,搜索N个连续模式(N之间的最小值N1和最大值N2)具有预期特性的噪声,它可用于存储NxP的相应样本的降噪处理输入信号,以构成噪声模型,并根据搜索结果找到一个新的噪声模型和通过搜索找到的存储新噪声模型来替换前一个或保留前两个模型。该模型通过N个模式(能量含量之间的比率是两个值S 1 / S)彼此更接近来找到什么能量含量。

著录项

  • 公开/公告号EP0993671A1

    专利类型

  • 公开/公告日2000-04-19

    原文格式PDF

  • 申请/专利权人 THOMSON-CSF SEXTANT;

    申请/专利号EP19980935094

  • 发明设计人 PASTOR DOMINIQUE;REYNAUD GERARD;

    申请日1998-07-03

  • 分类号G10L3/02;

  • 国家 EP

  • 入库时间 2022-08-22 01:47:42

相似文献

  • 专利
  • 外文文献
  • 中文文献
获取专利

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