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. 发明授权

US5664059A Self-learning speaker adaptation based on spectral variation source decomposition 失效
标题翻译：基于光谱变化源分解的自学习扬声器适应
公开(公告)号：US5664059A
公开(公告)日：1997-09-02
申请号：US710361
申请日：1996-09-16
申请人： Yunxin Zhao
发明人： Yunxin Zhao
IPC分类号： G10L5/06 , G10L7/08
CPC分类号： G10L15/07 , G10L15/144
摘要： A self-learning speaker adaptation method for automatic speech recognition is provided. The method includes building a plurality of Gaussian mixture density phone models for use in recognizing speech. The Gaussian mixture density phone models are used to recognize a first utterance of speech from a given speaker. After the first utterance of speech has been recognized, the recognized first utterance of speech is used to adapt the Gaussian mixture density hone models for use in recognizing a subsequent utterance of speech from that same speaker, whereby the Gaussian mixture density phone models are automatically adapted to that speaker in self-learning fashion to thereby produce a plurality of adapted Gaussian mixture density phone models.
摘要翻译：提供了一种用于自动语音识别的自学习扬声器适配方法。该方法包括构建用于识别语音的多个高斯混合密度电话模型。高斯混合密度手机模型用于识别来自给定说话者的第一话语。在第一语言发音被识别之后，所识别的第一话语被用于使高斯混合密度磨练模型适应于用于识别来自同一说话者的语音的后续话语，由此高斯混合密度手机模型被自动适应以自学习的方式向该扬声器发送，从而产生多个适应的高斯混合密度手机模型。

2. 外观设计

USD1014491S1 Mobile phone stand 有权
公开(公告)号：USD1014491S1
公开(公告)日：2024-02-13
申请号：US29833138
申请日：2022-04-01
申请人： Yunxin Zhao
设计人： Yunxin Zhao
摘要： FIG. 1 is a perspective view of a mobile phone stand, showing my new design;
FIG. 2 is another perspective view thereef;
FIG. 3 is a front view thereof;
FIG. 4 is a rear view thereof;
FIG. 5 is a left side view thereof;
FIG. 6 is a right side view thereof;
FIG. 7 is a top plan view thereof; and,
FIG. 8 is a bottom plan view thereof.

3. 发明授权

US5450523A Training module for estimating mixture Gaussian densities for speech unit models in speech recognition systems 失效
标题翻译：用于在语音识别系统中为语音单元模型估计混合高斯密度的训练模块
公开(公告)号：US5450523A
公开(公告)日：1995-09-12
申请号：US71334
申请日：1993-06-01
申请人： Yunxin Zhao
发明人： Yunxin Zhao
IPC分类号： G10L9/00
摘要： A model-training module generates mixture Gaussian density models from speech training data for continuous, or isolated word speech recognition systems. Speech feature sequences are labeled into segments of states of speech units using Viterbi-decoding based optimized segmentation algorithm. Each segment is modeled by a Gaussian density, and the parameters are estimated by sample mean and sample covariance. A mixture Gaussian density is generated for each state of each speech unit by merging the Gaussian densities of all the segments with the same corresponding label. The resulting number of mixture components is proportional to the dispersion and sample size of the training data. A single, fully merged, Gaussian density is also generated for each state of each speech unit. The covariance matrices of the mixture components are selectively smoothed by a measure of relative sharpness of the Gaussian density and the smoothing can also be done blockwise. The weights of the mixture components are set uniformly initially, and are reestimated using a segmental-average procedure. The weighting coefficients, together with the Gaussian densities, then become the models of speech units for use in speech recognition.
摘要翻译：模型训练模块从用于连续或隔离的单词语音识别系统的语音训练数据生成混合高斯密度模型。语音特征序列使用基于维特比解码的优化分割算法被标记成语音单元的状态段。每个段由高斯密度建模，参数通过样本均值和样本协方差估计。通过将所有段的高斯密度与相同的相应标号进行合并，为每个语音单元的每个状态生成混合高斯密度。所得到的混合物组分数量与训练数据的色散和样品量成比例。也为每个语音单元的每个状态生成单个，完全合并的高斯密度。混合分量的协方差矩阵通过高斯密度的相对清晰度的度量来选择性地平滑，并且平滑也可以是块状的。混合组分的重量最初均匀地设定，并使用分段平均程序进行重新估计。加权系数与高斯密度一起成为用于语音识别的语音单元的模型。

4. 发明授权

US5349645A Word hypothesizer for continuous speech decoding using stressed-vowel centered bidirectional tree searches 失效
标题翻译：用于连续语音解码的Word假设，使用以紧密元音为中心的双向树搜索
公开(公告)号：US5349645A
公开(公告)日：1994-09-20
申请号：US807255
申请日：1991-12-31
申请人： Yunxin Zhao
发明人： Yunxin Zhao
IPC分类号： G10L15/14 , G10L9/00
CPC分类号： G10L15/142
摘要： A word hypothesis module for speech decoding consists of four submodules: vowel center detection, bidirectional tree searches around each vowel center, forward-backward pruning, and additional short words hypotheses. By detecting the strong energy vowel centers, a vowel-centered lexicon tree can be placed at each vowel center and searches can be performed in both the left and right directions, where only simple phone models are used for fast acoustic match. A stage-wise forward-backward technique computes the word-beginning and word-ending likelihood scores over the generated half-word lattice for further pruning of the lattice. To avoid potential miss of short words with weak energy vowel centers, a lexicon tree is compiled for these words and tree searches are performed between each pair of adjacent vowel centers. The integration of the word hypothesizer with a top-down Viterbi beam search in continuous speech decoding provides two-pass decoding which significantly reduces computation time.
摘要翻译：用于语音解码的单词假设模块由四个子模块组成：元音中心检测，每个元音中心周围的双向树搜索，前向后向修剪和其他短字假设。通过检测强能元音中心，可以在每个元音中心放置一个以元音为中心的词典树，并且可以在左右两个方向进行搜索，仅使用简单的手机模型进行快速声匹配。逐级向前 - 反向技术计算生成的半字格阵上的词开始和词结合似然分数，用于进一步修剪格。为了避免潜在的缺乏弱能量元音中心的短语，为这些单词编译了一个词典树，并且在每对相邻的元音中心之间执行树搜索。在连续语音解码中，单词假设与自顶向下维特比波束搜索的集成提供了双遍解码，这显着减少了计算时间。

5. 发明授权

US5696878A Speaker normalization using constrained spectra shifts in auditory filter domain 失效
标题翻译：在听觉滤波器域中使用约束光谱位移的扬声器归一化
公开(公告)号：US5696878A
公开(公告)日：1997-12-09
申请号：US123637
申请日：1993-09-17
申请人： Yoshio Ono , Yunxin Zhao , Hisashi Wakita
发明人： Yoshio Ono , Yunxin Zhao , Hisashi Wakita
IPC分类号： G10L11/00 , G10L15/00 , G10L5/06
CPC分类号： G10L25/48 , G10L15/00 , G10L25/18 , G10L25/24
摘要： A speaker normalization method is described based on spectral shifts in the auditory filter domain. The method is characterized by using an estimated vocal tract length as a criterion to determine the spectral shift value. Certain constraints are found to be necessary for the shift in the auditory filter domain, and two techniques based on these constraints, the One-Bark shift and the refined Bark-scale shift, are introduced. When tested in vowel classification experiments, significant performance improvement was obtained for both techniques. The method is useful for speaker normalization in speaker-independent speech recognition.
摘要翻译：基于听觉滤波器域中的频谱偏移来描述扬声器归一化方法。该方法的特征在于使用估计的声道长度作为确定光谱偏移值的标准。发现某些约束对于听觉滤波器域中的移位是必要的，并且基于这些约束的两种技术，即单树皮移位和精细的巴克尺度位移被引入。当在元音分类实验中进行测试时，两种技术都获得了显着的性能改进。该方法对于与说话者无关的语音识别中的扬声器归一化是有用的。

6. 外观设计

USD1035635S1 Mobile phone stand 有权
公开(公告)号：USD1035635S1
公开(公告)日：2024-07-16
申请号：US29791337
申请日：2022-01-08
申请人： Yunxin Zhao
设计人： Yunxin Zhao
摘要： FIG. 1 is a perspective view of a mobile phone stand, showing my new design in an open configuration;
FIG. 2 is another perspective view showing the design in an closed configuration;
FIG. 3 is a front view thereof;
FIG. 4 is a rear view thereof;
FIG. 5 is a left side view thereof;
FIG. 6 is a right side view thereof;
FIG. 7 is a top plan view thereof; and,
FIG. 8 is a bottom plan view thereof.
The broken lines in the drawings illustrate portions of the mobile phone stand that form no part of the claimed design.

7. 外观设计

USD1035634S1 Mobile phone stand 有权
公开(公告)号：USD1035634S1
公开(公告)日：2024-07-16
申请号：US29791336
申请日：2022-01-08
申请人： Yunxin Zhao
设计人： Yunxin Zhao
摘要： FIG. 1 is a perspective view of a mobile phone stand, showing my new design in an open configuration;
FIG. 2 is another perspective view showing the design in an closed configuration;
FIG. 3 is a front view thereof;
FIG. 4 is a rear view thereof;
FIG. 5 is a left side view thereof;
FIG. 6 is a right side view thereof;
FIG. 7 is a top plan view thereof; and,
FIG. 8 is a bottom plan view thereof.
The broken lines in the drawings illustrate portions of the mobile phone stand that form no part of the claimed design.

8. 发明授权

US5794192A Self-learning speaker adaptation based on spectral bias source decomposition, using very short calibration speech 失效
标题翻译：基于光谱偏置源分解的自学习扬声器适配，使用非常短的校准语音
公开(公告)号：US5794192A
公开(公告)日：1998-08-11
申请号：US712802
申请日：1996-09-12
申请人： Yunxin Zhao
发明人： Yunxin Zhao
IPC分类号： G10L15/10 , G10L15/20 , G10L21/02 , G10L5/06 , G10L7/08
CPC分类号： G10L15/10 , G10L15/20 , G10L21/02
摘要： A speaker adaptation technique based on the separation of speech spectra variation sources is developed for improving speaker-independent continuous speech recognition. The variation sources include speaker acoustic characteristics, and contextual dependency of allophones. Statistical methods are formulated to normalize speech spectra based on speaker acoustic characteristics and then adapt mixture Gaussian density phone models based on speaker phonologic characteristics. Adaptation experiments using short calibration speech (5 sec./speaker) have shown substantial performance improvement over the baseline recognition system.
摘要翻译：开发了基于语音频谱变化源分离的扬声器适应技术，用于改善与扬声器无关的连续语音识别。变化源包括扬声器的声学特性，以及变体的语境依赖性。根据扬声器的声学特性，制定统计方法，对语音频谱进行归一化，然后根据扬声器的语音特征对混合高斯密度手机模型进行调整。使用短校准语音（5秒/扬声器）的适应实验显示出比基线识别系统显着的性能改进。

9. 发明授权

US5193142A Training module for estimating mixture gaussian densities for speech-unit models in speech recognition systems 失效
标题翻译：用于语音识别系统中语音单元模型的混合高斯密度的训练模块
公开(公告)号：US5193142A
公开(公告)日：1993-03-09
申请号：US613352
申请日：1990-11-15
申请人： Yunxin Zhao
发明人： Yunxin Zhao
IPC分类号： G10L15/14
CPC分类号： G10L15/144
摘要： A model-training module generates mixture Gaussian density models from speech training data for continuous, or isolated word HMM-based speech recognition systems. Speech feature sequences are labeled into segments of states of speech units using Viterbi-decoding based optimized segmentation algorithm. Each segment is modeled by a Gaussian density, and the parameters are estimated by sample mean and sample covariance. A mixture Gaussian density is generated for each state of each speech unit by merging the Gaussian densities of all the segments with the same corresponding label. The resulting number of mixture components is proportional to the dispension and sample size of the training data. A single, fully merged, Gaussian density is also generated for each state of each speech unit. The covariance matrices of the mixture components are selectively smoothed by a measure of relative sharpness of the Gaussian density. The weights of the mixture components are set uniformly initially, and are reestimated using a segmental-average procedure. The weighting coefficients, together with the Gaussian densities, then become the models of speech units for use in speech recognition.
摘要翻译：模型训练模块从语音训练数据生成用于连续或隔离词HMM的语音识别系统的混合高斯密度模型。语音特征序列使用基于维特比解码的优化分割算法被标记成语音单元的状态段。每个段由高斯密度建模，参数通过样本均值和样本协方差估计。通过将所有段的高斯密度与相同的相应标号进行合并，为每个语音单元的每个状态生成混合高斯密度。所得到的混合物组分数量与培训数据的分布和样本量成比例。也为每个语音单元的每个状态生成单个，完全合并的高斯密度。通过测量高斯密度的相对锐度来选择性地平滑混合分量的协方差矩阵。混合组分的重量最初均匀地设定，并使用分段平均程序进行重新估计。加权系数与高斯密度一起成为用于语音识别的语音单元的模型。

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