专利快速检索-快速检索全球专利，免费商用专利数据库-IPRDB

61. 发明申请

WO00060579A1 A FREQUENCY DOMAIN INTERPOLATIVE SPEECH CODEC SYSTEM 审中-公开
标题翻译：频域插播语音编解码系统
公开(公告)号：WO00060579A1
公开(公告)日：2000-10-12
申请号：PCT/US2000/008790
申请日：2000-04-04
IPC分类号： G10L11/02 , G10L11/04 , G10L19/00 , G10L19/02 , G10L19/04 , G10L19/08 , G10L19/14 , G10L19/12
CPC分类号： G10L19/18 , G10L19/005 , G10L19/02 , G10L19/0204 , G10L19/04 , G10L19/083 , G10L19/09 , G10L25/27 , G10L25/30 , G10L25/78 , G10L25/90 , G10L2019/0012 , G10L2025/783
摘要： A prototype word PW gain (34) is vector quantized using a vector quantizer (VQ) that explicitly populates the codebook by representative steady state and transient vectors of PW gain for tracking the abrupt variations in speech levels during onsets and other non-stationary events, while maintaining the accuracy of the speech level during stationary conditions. The rapidly evolving waveform REW (42) and slowly evolving waveform SEW (40) component vectors are converted to magnitude-phase. The variable dimension SEW magnitude vector is quantized using a hierarchical approach. The REW magnitude vector sequence is normalized to unity RMS value. The normalized REW magnitude vectors are modeled by a multi-band sub-band model which converts the variable dimension REW magnitude shape vectors, e.g., six dimensional REW sub-band vectors. The sub-band vectors are averaged over time, resulting in a single average REW sub-band vector for each frame.
摘要翻译：使用矢量量化器（VQ）对原型字PW增益（34）进行矢量量化，该矢量量化器（VQ）通过代表稳定状态和PW增益的瞬态矢量显式地填充码本，用于跟踪在开始和其他非平稳事件期间语音电平的突然变化，同时在静止状态下保持语音水平的准确性。快速演变的波形REW（42）和缓慢演变的波形SEW（40）分量矢量被转换为幅度相位。可变尺度SEW幅度矢量使用分层方法进行量化。 REW幅度矢量序列归一化为单位RMS值。归一化的REW幅度矢量由多频带子带模型建模，该多频带子频带模型转换可变维REW幅度形状矢量，例如六维REW子带矢量。子带向量随时间平均，导致每个帧的单个平均REW子带向量。

62. 发明申请

WO00060576A1 SPECTRAL PHASE MODELING OF THE PROTOTYPE WAVEFORM COMPONENTS FOR A FREQUENCY DOMAIN INTERPOLATIVE SPEECH CODEC SYSTEM 审中-公开
标题翻译：频域语音语音编解码系统的原型波形组件的频谱相位建模
公开(公告)号：WO00060576A1
公开(公告)日：2000-10-12
申请号：PCT/US2000/009073
申请日：2000-04-04
IPC分类号： G10L11/02 , G10L11/04 , G10L19/00 , G10L19/02 , G10L19/04 , G10L19/08 , G10L19/14 , G10L19/06
CPC分类号： G10L19/18 , G10L19/005 , G10L19/02 , G10L19/0204 , G10L19/04 , G10L19/083 , G10L19/09 , G10L25/27 , G10L25/30 , G10L25/78 , G10L25/90 , G10L2019/0012 , G10L2025/783
摘要： Encoding of prototype waveform (PW) components applicable to GeoMobile and Telephony Earth Station (TES), provides improved voice quality, enabling a dual-channel mode of operation which permits more users to communicate over the same physical channel. A PW gain (34) is quantized using a vector quantizer (VQ) having a codebook representative of steady-state and transient vectors for tracking abrupt speech level variations during onsets and other non-stationary events, while maintaining accuracy during stationary conditions. The rapidly evolving waveform (REW) and slowly evolving waveform (SEW) component vectors, (42) and (40), respectively, are converted to magnitude-phase. A voicing measure, characterizing the degree of signal periodicity, and an interpolated pitch countour, based on estimated open loop pitch, are determined for each speech frame. Only the SEW and REW spectral magnitude information, the voicing measure, and pitch frequency countour are transmitted. The SEW phase component is reconstructed at the decoder from this information. The REW phase component is reconstructed at the decoder from the reconstructed SEW component, the voicing measure, REW magnitudes, and the pitch frequency contour information.
摘要翻译：适用于地理移动和电话地球站（TES）的原型波形（PW）组件的编码提供了改进的语音质量，实现了双通道操作模式，允许更多用户通过相同的物理通道进行通信。使用具有代表稳态和瞬态向量的码本的矢量量化器（VQ）来量化PW增益（34），用于跟踪在开启和其他非平稳事件期间的突发语音电平变化，同时在稳定条件期间保持精度。快速演变的波形（REW）和缓慢演变的波形（SEW）分量矢量（42）和（40）分别被转换为幅度相位。对于每个语音帧，确定基于估计的开环间距的表征度量，表征信号周期的程度和内插音调countour。仅发送SEW和REW频谱幅度信息，发声测量和音调频率countour。在该信息的解码器上重构SEW相位分量。 REW相位分量在解码器处从重建的SEW分量，发音测量，REW幅度和音调频率轮廓信息重建。

63. 发明申请

WO99041844A1 A NEURAL NETWORK BASED METHOD FOR EXPONENT CODING IN A TRANSFORM CODER FOR HIGH QUALITY AUDIO 审中-公开
标题翻译：一种基于神经网络的高质量音频转换编码器编码方法
公开(公告)号：WO99041844A1
公开(公告)日：1999-08-19
申请号：PCT/SG1998/000009
申请日：1998-02-12
IPC分类号： G10L25/30 , H04B1/66
CPC分类号： H04B1/665 , G10L25/30
摘要： A method and apparatus for assigning an exponent coding strategy in a digital audio transform coder. Different coding strategies having different differential coding limits may be assigned to different set of transform exponents according to the frequency domain characteristics of the audio signal. A neural network processing system is utilised to perform an efficient mapping of each exponent set to an appropriate coding strategy.
摘要翻译：一种用于在数字音频变换编码器中分配指数编码策略的方法和装置。根据音频信号的频域特性，可以将具有不同差分编码限制的不同编码策略分配给不同组的变换指数。使用神经网络处理系统来执行每个指数集合到适当编码策略的有效映射。

64. 发明申请

WO1997032428A1 TRAINING PROCESS 审中-公开
标题翻译：培训流程
公开(公告)号：WO1997032428A1
公开(公告)日：1997-09-04
申请号：PCT/GB1997000265
申请日：1997-01-30
申请人： BRITISH TELECOMMUNICATIONS PUBLIC LIMITED COMPANY , HOLLIER, Michael, Peter , GRAY, Philip
发明人： BRITISH TELECOMMUNICATIONS PUBLIC LIMITED COMPANY
IPC分类号： H04M03/22
CPC分类号： G10L25/69 , G10L19/018 , G10L25/30 , H04M3/22 , H04M3/2236 , H04M2201/40
摘要： Training apparatus for establishing the network definition function of a trainable processing apparatus (5) for analysing a signal, comprises means (8) for providing a training sequence comprising a first signal and a distorted version of the first signal, analysis means (9) for receiving the training sequence and generating a distortion perception measure for indicating the extent to which the distortion would be perceptible to a human observer, and means for applying the distortion perception measure to the trainable processing apparatus (5) to determine the network definition function.
摘要翻译：用于建立用于分析信号的可训练处理装置（5）的网络定义功能的训练装置包括用于提供包括第一信号和第一信号的失真版本的训练序列的装置（8），用于接收所述训练序列并产生用于指示所述失真对人类观察者可感知的程度的失真感知度量，以及用于将所述失真感知度量应用于所述可训练处理装置（5）以确定所述网络定义功能的装置。

65. 发明申请

WO2019096149A1 基于记忆和注意力模型的听觉选择方法和装置审中-公开
公开(公告)号：WO2019096149A1
公开(公告)日：2019-05-23
申请号：PCT/CN2018/115351
申请日：2018-11-14
申请人：中国科学院自动化研究所
发明人：许家铭 , 石晶 , 徐波
IPC分类号： G10L15/22
CPC分类号： G10L15/22 , G10L19/0017 , G10L21/0208 , G10L21/0272 , G10L25/30
摘要：基于记忆和注意力模型的听觉选择方法和装置，属于语音分离技术领域。旨在解决现有技术中监督标签的排列、说话人混叠数目不确定以及记忆单元维度固定的问题。提供一种基于记忆和注意力模型的听觉选择方法，包括将原始语音信号编码为时频矩阵(步骤S1)，对时频矩阵进行编码和变换，将其转化为语音向量(步骤S2)，利用长时记忆单元存储说话人以及其对应的语音向量(步骤S3)，获取目标说话人的语音向量，通过注意力选择模型从原始语音信号中分离出目标语音(步骤S4)。能够不需要固定或者指定说话人的数目，从原始语音信号中分离出目标语音。

66. 发明申请

WO2019078567A1 소리 신호를 처리하는 반도체 장치 및 이를 포함하는 마이크 장치 审中-公开
公开(公告)号：WO2019078567A1
公开(公告)日：2019-04-25
申请号：PCT/KR2018/012142
申请日：2018-10-16
申请人： 서울대학교 산학협력단
发明人： 김수환 , 조준수
IPC分类号： G10L19/02 , G10L25/30 , G10L21/0272 , H04R19/00
CPC分类号： G10L19/02 , G10L21/0272 , G10L25/30 , H04R19/00
摘要： 본 기술에 의한 반도체 장치는 입력 신호를 디지털 신호로 변환하는 아날로그 디지털 변환기; 디지털 신호를 외부에서 입력된 동작 파라미터의 영향을 받아 처리하는 신호 처리부; 및 아날로그 디지털 변환기의 출력 또는 신호 처리부의 출력을 외부에 제공하는 인터페이스를 포함한다.

67. 发明申请

WO2019063547A1 METHOD AND ELECTRONIC DEVICE FOR FORMANT ATTENUATION/AMPLIFICATION 审中-公开
公开(公告)号：WO2019063547A1
公开(公告)日：2019-04-04
申请号：PCT/EP2018/075965
申请日：2018-09-25
申请人： SONY EUROPE LIMITED
发明人： DERUTY, Emmanuel
IPC分类号： G10L21/003 , G10L25/30 , H03G3/20 , H04H60/04 , G10L25/15
CPC分类号： G10L21/003 , G10L25/15 , G10L25/30 , H03G9/005 , H03G9/025 , H04H60/04
摘要： A method comprising determining feature values of an input audio window and determining a formant attenuation/amplification coefficient for the input audio window based on the processing of the feature values by a neural network.

68. 发明申请

WO2018204708A1 BIOMIMETIC CODECS AND BIOMIMETIC CODING TECHNIQUES 审中-公开
公开(公告)号：WO2018204708A1
公开(公告)日：2018-11-08
申请号：PCT/US2018/030973
申请日：2018-05-03
申请人： UNIVERSITY OF WASHINGTON
发明人： SARIKAYA, Mehmet , USTUNDAG, Burak Berk
IPC分类号： H04B1/66
CPC分类号： H04B1/66 , G10L19/02 , G10L25/30 , H04N19/42
摘要： Examples of biomimetic codecs and biomimetic coding techniques are described herein. Morphologically-adaptive coding networks can be developed in accordance with energy dissipation driven "heat" generated by application of training data. The morphologically-adaptive coding networks may be representative of common features expected in an input signal or data stream. Decoding may proceed using the morphologically-adaptive coding network. Morphologically-adaptive coding networks may be used as a cortex that can be shared for boosting multimedia data compression rates and/or increasing the encode-decode fidelity of information content while the features remain queryable in encoded form. Examples of the biomimetic codecs and biomimetic coding techniques provide a broad-based technology platform that can be used in context-IDed multimedia storage, pattern recognition, and high-performance computing/big data management, the hallmarks of web- and cloud-based systems.

69. 发明申请

WO2017191249A1 SPEECH ENHANCEMENT AND AUDIO EVENT DETECTION FOR AN ENVIRONMENT WITH NON-STATIONARY NOISE 审中-公开
标题翻译：非定常噪声环境下的语音增强和音频事件检测
公开(公告)号：WO2017191249A1
公开(公告)日：2017-11-09
申请号：PCT/EP2017/060637
申请日：2017-05-04
申请人： ROBERT BOSCH GMBH
发明人： HASAN AL BANNA, Taufiq
IPC分类号： G10L21/0216 , G10L25/30 , G10L15/26
CPC分类号： G10L21/0216 , G10L15/20 , G10L15/26 , G10L21/0205 , G10L25/30
摘要： A computer-implemented system and method provide an audio label for a noise signal. The computer-implemented system and method include receiving an audio input and obtaining the noise signal from the audio input. The computer-implemented system and method include extracting audio features of the noise signal. The computer-implemented system and method include determining and outputting an audio label for the extracted audio features of the noise signal based on machine learning data.
摘要翻译：计算机实现的系统和方法为噪声信号提供音频标签。计算机实现的系统和方法包括接收音频输入并从音频输入获得噪声信号。计算机实现的系统和方法包括提取噪声信号的音频特征。计算机实现的系统和方法包括基于机器学习数据确定并输出用于噪声信号的提取的音频特征的音频标签。

70. 发明申请

WO2017172632A2 CHARACTERIZING, SELECTING AND ADAPTING AUDIO AND ACOUSTIC TRAINING DATA FOR AUTOMATIC SPEECH RECOGNITION SYSTEMS 审中-公开
标题翻译：自动语音识别系统的特点，选择和适应音频和声学培训数据
公开(公告)号：WO2017172632A2
公开(公告)日：2017-10-05
申请号：PCT/US2017/024343
申请日：2017-03-27
申请人： NUANCE COMMUNICATIONS, INC.
发明人： SHARMA, Dushyant , NAYLOR, Patrick , JOST, Uwe, Helmut
IPC分类号： G10L15/065 , G10L21/003 , G10L15/20
CPC分类号： G10L25/30 , G10L15/063 , G10L15/065 , G10L15/20 , G10L19/0018 , G10L19/06 , G10L21/003 , G10L2015/0631
摘要： A system for and method of characterizing a target application acoustic domain analyzes one or more speech data samples from the target application acoustic domain to determine one or more target acoustic characteristics, including a CODEC type and bit-rate associated with the speech data samples. The determined target acoustic characteristics may also include other aspects of the target speech data samples such as sampling frequency, active bandwidth, noise level, reverberation level, clipping level, and speaking rate. The determined target acoustic characteristics are stored in a memory as a target acoustic data profile. The data profile may be used to select and/or modify one or more out of domain speech samples based on the one or more target acoustic characteristics.
摘要翻译：表征目标应用声学域的系统和方法分析来自目标应用声学域的一个或多个语音数据样本以确定一个或多个目标声学特性，包括CODEC类型和比特率与语音数据样本相关联。所确定的目标声学特性还可以包括目标语音数据采样的其他方面，诸如采样频率，有效带宽，噪声水平，混响水平，限幅水平和讲话速率。确定的目标声学特性作为目标声学数据轮廓存储在存储器中。数据简档可以用于基于一个或多个目标声学特性选择和/或修改一个或多个域外语音样本。

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