专利快速检索-快速检索全球专利，免费商用专利数据库-IPRDB

1. 发明申请

WO2012154798A1 SPEAKER LIVENESS DETECTION 审中-公开
标题翻译：扬声器生活检测
公开(公告)号：WO2012154798A1
公开(公告)日：2012-11-15
申请号：PCT/US2012/037040
申请日：2012-05-09
申请人： INTERNATIONAL BUSINESS MACHINES CORPORATION , BAUGHMAN, Aaron, K. , PELECANOS, Jason, W.
发明人： BAUGHMAN, Aaron, K. , PELECANOS, Jason, W.
IPC分类号： G10L17/00
CPC分类号： G10L17/22 , G10L17/26
摘要： A signal representative of an unpredictable audio stimulus is provided to a putative live speaker within a putative live recording environment. A second signal purportedly emanating from the putative live speaker and/or the environment is received. This second signal is examined for influence of the unpredictable audio stimulus on the putative live speaker and/or the putative live recording environment. The examining includes at least one of audio feedback analysis, Lombard analysis, and evoked otoacoustic response analysis. Based on the examining, a determination is made as to whether the putative live speaker is an actual live speaker and/or whether the putative live recording environment is an actual live recording environment.
摘要翻译：将代表无法预测的音频刺激的信号提供给推定的现场录音环境中的推定的现场演讲者。据称是从推定的现场演讲者和/或环境中发出的第二个信号。对第二个信号进行检查，以了解不可预知的音频刺激对推定的现场演讲者和/或推定的现场录制环境的影响。检查包括音频反馈分析，伦巴第分析和诱发耳声响应分析中的至少一个。根据审查，确定推定的现场演讲者是否是实际的现场演讲者，和/或推定的现场录制环境是否是实际的录音环境。

2. 发明申请

WO2005055200A1 MODEL ADAPTATION SYSTEM AND METHOD FOR SPEAKER RECOGNITION 审中-公开
标题翻译：用于语音识别的模型适应系统和方法
公开(公告)号：WO2005055200A1
公开(公告)日：2005-06-16
申请号：PCT/AU2004/001718
申请日：2004-12-03
申请人： QUEENSLAND UNIVERSITY OF TECHNOLOGY , PELECANOS, Jason , VOGT, Robert , SRIDHARAN, Subramanian
发明人： PELECANOS, Jason , VOGT, Robert , SRIDHARAN, Subramanian
IPC分类号： G10L17/00
CPC分类号： G10L17/04
摘要： A system and method for speaker recognition speaker modelling whereby prior speaker information is incorporated into the modelling process, utilising the maximum a posteriori (MAP) algorithm and extending it to contain prior Gaussian component correlation information. Firstly a background model (10) is estimated. Pooled acoustic reference data (11) relating to a specific demographic of speakers (population of interest) from a given total population is then trained via the Expectation Maximization (EM) algorithm (12) to produce a background model (13). The background model (13) is adapted utilising information from a plurality of reference speakers (21) in accordance with the Maximum A Posteriori (MAP) criterion (22). Utilizing MAP estimation technique, the reference speaker data and prior information obtained from the background model parameters are combined to produce a library of adapted speaker models, namely Gaussian Mixture Models (23).
摘要翻译：一种用于说话者识别扬声器建模的系统和方法，其中先前的说话者信息被并入到建模过程中，利用最大后验（MAP）算法并将其扩展为包含先前的高斯分量相关信息。首先估计一个背景模型（10）。然后通过期望最大化（EM）算法（12）训练与给定总人口的特定人群（兴趣人群）有关的汇集的声学参考数据（11）以产生背景模型（13）。背景模型（13）根据最大后验（最大后验）（MAP）标准（22）利用来自多个参考扬声器（21）的信息。利用MAP估计技术，将从背景模型参数获得的参考说话者数据和先验信息相结合，以产生适应的说话者模型库，即高斯混合模型（23）。

你已经成功收藏专利！

检索式保存成功!

IPRDB

热门服务

关于我们

友情链接

联系方式