Discriminative adaptation for speaker verification
Discriminative adaptation for speaker verification
复制标题
说话人验证的判别性适应
DOI：
10.1109/icslp.1996.607965
复制
发表时间：
1996
期刊：
影响因子：
--
通讯作者：
B. Juang
中科院分区：
文献类型：
--
作者：
Filipp Korkmazskiy;B. Juang
描述了一种说话者验证系统，其中说话者和冒名顶替者模型经过调整以实现最大区分度或同等的最小验证错误。这一目标是通过将最小分类误差（MCE）标准和广义概率下降（GPD）算法扩展到将说话者模型参数和相应的反说话者模型参数适应测试环境的任务来实现的，从而最小化验证错误率的经验估计。我们在当前的研究中解决了两类参数的适应问题：模型参数和决策阈值。通过应用涉及简化 MAP（最大后验）方法和 GPD 算法的组合技术，我们在等错误率方面获得了实质性改进。包含 43 个说话者（每个人有 5 个适应话语）的数据库的等错误率从之前报告的最佳结果 5.41% 降低到了 2.17%。我们讨论了本工作中研究的几种替代方法，以便为在说话者验证任务中使用判别性方法提供比较见解。
Describes a speaker verification system in which the talker and imposter models are adapted to achieve maximum discrimination or, equivalently, minimum verification error. This goal is accomplished by extending the minimum classification error (MCE) criterion and the generalized probabilistic descent (GPD) algorithm to the task of adapting talker model parameters and the corresponding anti-talker model parameters to the test environments so as to minimize an empirical estimate of the verification error rate. We address in the current study adaptation of two types of parameters: the model parameters and the decision threshold. We have obtained substantial improvements in the equal error rate by applying combined techniques involving a simplified MAP (maximum a posteriori) method and the GPD algorithm. The equal error rate for a database of 43 talkers with 5 adaptation utterances each was reduced from the previously reported best result of 5.41% to 2.17%. We discuss several alternative methods that have been investigated in this work to provide comparative insights for the use of discriminative methods in speaker verification tasks.