公开(公告)号：US20160247518A1

公开(公告)日：2016-08-25

申请号：US15147549

申请日：2016-05-05

Applicant: Huawei Technologies Co., Ltd.

Inventor： Bjoern Schuller ,  Felix Weninger ,  Christian Kirst ,  Peter Grosche

IPC: G10L21/0272 ,  G10L21/0216 ,  G10L25/84 ,  H04S5/00 ,  H04S7/00

CPC classification number: G10L21/0272 ,  G10L21/0216 ,  G10L25/84 ,  H04S5/00 ,  H04S7/30 ,  H04S2420/01

Abstract: The present invention relates to an apparatus for improving a perception of a sound signal, the apparatus comprising: a separation unit configured to separate the sound signal into at least one speech component and at least one noise component; and a spatial rendering unit configured to generate an auditory impression of the at least one speech component at a first virtual position with respect to a user, when output via a transducer unit, and of the at least one noise component at a second virtual position with respect to the user, when output via the transducer unit.

Abstract translation: 本发明涉及一种用于改善声音信号感知的装置，该装置包括：分离单元，被配置为将声音信号分离成至少一个语音分量和至少一个噪声分量; 以及空间绘制单元，被配置为当经由换能器单元输出时，在相对于用户的第一虚拟位置处生成所述至少一个语音分量的听觉印象，以及在第二虚拟位置处产生所述至少一个噪声分量的听觉印象， 当通过换能器单元输出时，它们与用户相关。

公开(公告)号：US09536538B2

公开(公告)日：2017-01-03

申请号：US14716289

申请日：2015-05-19

Applicant: Huawei Technologies Co., Ltd.

Inventor： Cyril Joder ,  Felix Weninger ,  Bjoern Schuller ,  David Virette

IPC: G10L21/0208 ,  G10L21/0216 ,  G10L21/0232

CPC classification number: G10L21/0208 ,  G10L21/0216 ,  G10L21/0232

Abstract: A method for reconstructing at least one target signal comprises determining a first set of feature vectors from the input signal, the first set of feature vectors forming a non-negative input matrix; determining a second set of feature vectors, the second set of feature vectors forming a non-negative noise matrix; decomposing the input matrix into a sum of a first matrix and a second matrix, the first matrix representing a product of a non-negative bases matrix and a non-negative weight matrix, and the second matrix representing a combination of the noise matrix and a noise weight vector; and reconstructing the at least one target signal based on the non-negative bases matrix and the non-negative weight matrix.

Abstract translation: 一种用于重构至少一个目标信号的方法包括从所述输入信号确定第一组特征向量，所述第一组特征向量形成非负输入矩阵; 确定第二组特征向量，所述第二组特征向量形成非负噪声矩阵; 将输入矩阵分解为第一矩阵和第二矩阵的和，第一矩阵表示非负基数矩阵和非负权重矩阵的乘积，第二矩阵表示噪声矩阵和 噪声权重向量; 以及基于所述非负基数矩阵和所述非负权重矩阵来重构所述至少一个目标信号。

公开(公告)号：US11889292B2

公开(公告)日：2024-01-30

申请号：US17581527

申请日：2022-01-21

Applicant: HUAWEI TECHNOLOGIES CO., LTD.

Inventor： Mohammad Taghizadeh ,  Gil Keren ,  Shuo Liu ,  Bjoern Schuller

IPC: H04S7/00 ,  G10L21/0208 ,  G10L25/18 ,  G10L25/30 ,  G10L21/10

CPC classification number: H04S7/307 ,  G10L21/0208 ,  G10L21/10 ,  G10L25/18 ,  G10L25/30 ,  H04S2400/03

Abstract: The disclosure relates to an audio processing apparatus, comprising: a plurality of audio sensors, each audio sensor configured to receive a respective plurality of audio frames of an audio signal from an audio source, wherein the respective plurality of audio frames defines an audio channel of the audio signal; and a processing circuitry configured to: determine a respective feature set having at least one feature for each audio frame of each of the plurality of audio frames, wherein the plurality of features define a three-dimensional feature array; process the three-dimensional feature array using a neural network, wherein the neural network comprises a self-attention layer configured to process a plurality of two-dimensional sub-arrays of the three-dimensional feature array; and generate an output signal on the basis of the plurality of processed two-dimensional sub-arrays. Moreover, the disclosure relates to a corresponding audio processing method.

公开(公告)号：US20150262590A1

公开(公告)日：2015-09-17

申请号：US14716289

申请日：2015-05-19

Applicant: Huawei Technologies Co., Ltd.

Inventor： Cyril Joder ,  Felix Weninger ,  Bjoern Schuller ,  David Virette

IPC: G10L21/0208

CPC classification number: G10L21/0208 ,  G10L21/0216 ,  G10L21/0232

Abstract: A method for reconstructing at least one target signal comprises determining a first set of feature vectors from the input signal, the first set of feature vectors forming a non-negative input matrix; determining a second set of feature vectors, the second set of feature vectors forming a non-negative noise matrix; decomposing the input matrix into a sum of a first matrix and a second matrix, the first matrix representing a product of a non-negative bases matrix and a non-negative weight matrix, and the second matrix representing a combination of the noise matrix and a noise weight vector; and reconstructing the at least one target signal based on the non-negative bases matrix and the non-negative weight matrix.

Abstract translation: 一种用于重构至少一个目标信号的方法包括从所述输入信号确定第一组特征向量，所述第一组特征向量形成非负输入矩阵; 确定第二组特征向量，所述第二组特征向量形成非负噪声矩阵; 将输入矩阵分解为第一矩阵和第二矩阵的和，第一矩阵表示非负基数矩阵和非负权重矩阵的乘积，第二矩阵表示噪声矩阵和 噪声权重向量; 以及基于所述非负基数矩阵和所述非负权重矩阵来重构所述至少一个目标信号。