首页> 外文学位 >Discriminative and generative approaches for long- and short-term speaker characteristics modeling: Application to speaker verification.

【24h】

Discriminative and generative approaches for long- and short-term speaker characteristics modeling: Application to speaker verification.

机译：长期和短期说话者特征建模的判别和生成方法：在说话者验证中的应用。

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

The speaker verification problem can be stated as follows: given two speech recordings, determine whether or not they have been uttered by the same speaker. Most current speaker verification systems are based on Gaussian mixture models. This probabilistic representation allows to adequately model the complex distribution of the speech frames. It however represents an inadequate basis for discriminating between speakers, which is the key issue in the area of speaker verification. In the first part of this thesis, we attempt to overcome these difficulties by proposing to combine support vector machines with two generative approaches based on Gaussian mixture models. In the second part of this thesis, we present a new approach to modeling the speaker's long-term prosodic and spectral characteristics. This novel approach is based on continuous approximations of the prosodic and cepstral contours. Finally, we perform a scores fusion between systems based on long- and short-term speaker features.

机译：说话者验证问题可以描述如下：给定两个语音记录，确定它们是否由同一说话者发出。当前大多数说话人验证系统都是基于高斯混合模型。这种概率表示允许对语音帧的复杂分布进行充分建模。然而，这代表了不充分的区分说话者的基础，这是说话者验证领域的关键问题。在本文的第一部分中，我们试图通过将支持向量机与基于高斯混合模型的两种生成方法相结合来克服这些困难。在本文的第二部分，我们提出了一种对说话者的长期韵律和频谱特征进行建模的新方法。这种新颖的方法是基于韵律和倒谱轮廓的连续近似。最后，我们基于长期和短期说话者功能在系统之间进行分数融合。

著录项

作者
Dehak, Najim.;
展开▼
作者单位

Ecole de Technologie Superieure (Canada).;

展开▼
授予单位 Ecole de Technologie Superieure (Canada).;
学科 Artificial Intelligence.
学位 Ph.D.
年度 2009
页码 164 p.
总页数 164
原文格式 PDF
正文语种 eng
中图分类人工智能理论;
关键词
入库时间 2022-08-17 11:38:27

相似文献

外文文献
中文文献
专利

1. Speaker Model Clustering for Efficient Speaker Identification in Large Population Applications [J] . Apsingekar V.R.De Leon P.L. Audio, Speech, and Language Processing, IEEE Transactions on . 2009,第4期

机译：扬声器模型聚类，可在大量应用中实现高效的说话人识别
2. Application of Native Speaker Models for Identifying Deviations in Rhetorical Moves in Non-Native Speaker Manuscripts [J] . Assef Khalili, Farahman Farrokhi, Majid Aslanabadi Research and Development in Medical Education . 2016,第1期

机译：本机说话人模型在识别非本国说话人手稿修辞动作中的偏差中的应用
3. Automatic Speaker Recognition Dependency on Both the Shape of Auditory Critical Bands and Speaker Discriminative MFCCs [J] . JOKIC I, DELIC V, JOKIC S, Advances in Electrical and Computer Engineering . 2015,第4期

机译：听觉关键带的形状和说话人区分性MFCC的自动说话人识别依赖性
4. Evaluation of the generative and discriminative text-independent speaker verification approaches on handheld devices [C] . Curelaru Florin International Conference on Speech Technology and Human-Computer Dialogue . 2015

机译：在手持设备上评估生成性和区分性文本无关的说话者验证方法
5. Speaker Characteristic-based Acoustic Model Adaptation Method for Speaker Recognition Systems [D] . Millington, Daniel S. 2011

机译：基于说话者特征的说话人识别系统声学模型自适应方法
6. Ordered short-term memory differs in signers and speakers: Implications for models of short-term memory [O] . Daphne Bavelier, Elissa L. Newport, Matt Hall, -1

机译：签名者和说话者的有序短期记忆有所不同：短期记忆模型的含义
7. Discriminative and generative approaches for long- and short-term speaker characteristics modeling : application to speaker verification [O] . Dehak Najim 2009

机译：用于长期建模的判别和生成方法
8. Fusing Discriminative and Generative Methods for Speaker Recognition: Experiments on Switchboard and NFI/TNO Field Data. [R] . Campbell, W. M., Reynolds, D. A., Campbell, J. P. 2016

机译：融合说话人识别的判别和生成方法：交换机和NFI / TNO现场数据的实验。

Discriminative and generative approaches for long- and short-term speaker characteristics modeling: Application to speaker verification.

摘要

著录项

相似文献

相关主题

期刊订阅