公开(公告)号：US20160196833A1

公开(公告)日：2016-07-07

申请号：US14591418

申请日：2015-01-07

Applicant: GOOGLE INC.

Inventor： Simon J. GODSILL ,  Herbert BUCHNER ,  Jan SKOGLUND

IPC: G10L21/0208 ,  H04R3/00

Abstract: Provided are methods and systems for enhancing speech when corrupted by transient noise (e.g., keyboard typing noise). The methods and systems utilize a reference microphone input signal for the transient noise in a signal restoration process used for the voice part of the signal. A robust Bayesian statistical model is used to regress the voice microphone on the reference microphone, which allows for direct inference about the desired voice signal while marginalizing the unwanted power spectral values of the voice and transient noise. Also provided is a straightforward and efficient Expectation-maximization (EM) procedure for fast enhancement of the corrupted signal. The methods and systems are designed to operate easily in real-time on standard hardware, and have very low latency so that there is no irritating delay in speaker response.

Abstract translation: 提供了当被瞬态噪声（例如键盘打字噪声）损坏时增强语音的方法和系统。 该方法和系统在用于信号的语音部分的信号恢复过程中利用用于瞬态噪声的参考麦克风输入信号。 鲁棒的贝叶斯统计模型用于回归参考麦克风上的语音麦克风，这允许对所需语音信号的直接推断，同时边缘化语音和瞬态噪声的不需要的功率谱值。 还提供了一种直接有效的期望最大化（EM）程序，用于快速增强损坏的信号。 这些方法和系统被设计为能够在标准硬件上实时地进行操作，并且具有非常低的延迟，使得扬声器响应没有刺激性的延迟。

公开(公告)号：US20160118038A1

公开(公告)日：2016-04-28

申请号：US14521104

申请日：2014-10-22

Applicant: GOOGLE INC.

Inventor： D. James EATON ,  Alastair H. MOORE ,  Patrick A. NAYLOR ,  Jan SKOGLUND

IPC: G10K15/08

CPC classification number: G10K15/08 ,  G10L21/0208 ,  G10L25/03 ,  G10L25/18 ,  G10L25/21 ,  G10L2021/02082 ,  G10L2021/02166 ,  H04R3/005

Abstract: Provided are methods and systems for generating Direct-to-Reverberant Ratio (DRR) estimates. The methods and systems use a null-steered beamformer to produce accurate DRR estimates across a variety of room sizes, reverberation times, and source-receiver distances. The DRR estimation algorithm uses spatial selectivity to separate direct and reverberant energy and account for noise separately. The formulation considers the response of the beamformer to reverberant sound and the effect of noise. The DRR estimation algorithm is more robust to background noise than existing approaches, and is applicable where a signal is recorded with two or more microphones, such as with mobile communications devices, laptop computers, and the like.

Abstract translation: 提供了生成直接到混响比（DRR）估计的方法和系统。 这些方法和系统使用零转向波束形成器来在各种房间尺寸，混响时间和源 - 接收器距离之间产生准确的DRR估计。 DRR估计算法使用空间选择性分离直接和混响能量，分别考虑噪声。 该公式考虑了波束形成器对混响声音和噪声影响的响应。 与现有方法相比，DRR估计算法对背景噪声更加鲁棒，并且适用于使用两个或更多个麦克风记录信号，例如使用移动通信设备，膝上型计算机等。

公开(公告)号：US20160293176A1

公开(公告)日：2016-10-06

申请号：US15182751

申请日：2016-06-15

Applicant: GOOGLE INC.

Inventor： Minyue LI ,  Jan SKOGLUND ,  Willem Bastiaan KLEIJN

IPC: G10L19/24 ,  G10L19/008 ,  G10L19/02 ,  G10L19/035 ,  G10L19/16

CPC classification number: G10L19/24 ,  G10L19/008 ,  G10L19/0212 ,  G10L19/035 ,  G10L19/167

Abstract: Provided are methods, systems, and apparatus for hierarchical decorrelation of multichannel audio. A hierarchical decorrelation algorithm is designed to adapt to possibly changing characteristics of an input signal, and also preserves the energy of the original signal. The algorithm is invertible in that the original signal can be retrieved if needed. Furthermore, the proposed algorithm decomposes the decorrelation process into multiple low-complexity steps. The contribution of these steps is generally in a decreasing order, and thus the complexity of the algorithm can be scaled.

Abstract translation: 提供了用于多声道音频的分层去相关的方法，系统和装置。 分层去相关算法被设计为适应输入信号的可能变化的特性，并且还保留原始信号的能量。 该算法是可逆的，因为如果需要，可以检索原始信号。 此外，所提出的算法将解相关过程分解为多个低复杂度步骤。 这些步骤的贡献通常是递减的，从而可以缩放算法的复杂性。

公开(公告)号：US20150199979A1

公开(公告)日：2015-07-16

申请号：US13899381

申请日：2013-05-21

Applicant: Google, Inc.

Inventor： Andrew J. HINES ,  Jan SKOGLUND ,  Naomi HARTE ,  Anil KOKARAM

IPC: G10L25/78

CPC classification number: G10L25/78 ,  G10L21/0232 ,  G10L25/60

Abstract: Methods and systems are provided for detecting chop in an audio signal. A time-frequency representation, such as a spectrogram, is created for an audio signal and used to calculate a gradient of mean power per frame of the audio signal. Positive and negative gradients are defined for the signal based on the gradient of mean power, and a maximum overlap offset between the positive and negative gradients is determined by calculating a value that maximizes the cross-correlation of the positive and negative gradients. The negative gradient values may be combined (e.g., summed) with the overlap offset, and the combined values then compared with a threshold to estimate the amount of chop present in the audio signal. The chop detection model provided is low-complexity and is applicable to narrowband, wideband, and superwideband speech.

Abstract translation: 提供了用于检测音频信号中的斩波的方法和系统。 为音频信号创建时频表示，如频谱图，用于计算音频信号每帧平均功率的梯度。 基于平均功率梯度的信号定义正和负梯度，通过计算使正和负梯度的互相关最大化的值来确定正梯度和负梯度之间的最大重叠偏移。 负梯度值可以与重叠偏移组合（例如，相加），然后将组合值与阈值进行比较以估计音频信号中存在的斩波量。 提供的斩波检测模型是低复杂度的，适用于窄带，宽带和超宽带语音。

公开(公告)号：US20140244247A1

公开(公告)日：2014-08-28

申请号：US13781262

申请日：2013-02-28

Applicant: GOOGLE INC.

Inventor： Jens Enzo Nyby CHRISTENSEN ,  Simon J. GODSILL ,  Jan SKOGLUND

IPC: G10L21/0232

CPC classification number: G10L25/48 ,  G10L21/02 ,  G10L21/0208 ,  G10L21/0216 ,  G10L25/84 ,  G10L25/93 ,  G10L2025/935

Abstract: Provided are methods and systems for detecting the presence of a transient noise event in an audio stream using primarily or exclusively the incoming audio data. Such an approach offers improved temporal resolution and is computationally efficient. The methods and systems presented utilize some time-frequency representation of an audio signal as the basis in a predictive model in an attempt to find outlying transient noise events and interpret the true detection state as a Hidden Markov Model (HMM) to model temporal and frequency cohesion common amongst transient noise events.

Abstract translation: 提供了用于检测音频流中瞬时噪声事件的存在的方法和系统，其主要或排他地使用输入音频数据。 这种方法提供了改进的时间分辨率，并且在计算上是有效的。 所提出的方法和系统利用音频信号的一些时间频率表示作为预测模型的基础，以试图找出偏离的瞬态噪声事件，并将真实检测状态解释为隐马尔科夫模型（HMM）来建模时间和频率 瞬态噪声事件中共同的凝聚力。

公开(公告)号：US20170194015A1

公开(公告)日：2017-07-06

申请号：US14984373

申请日：2015-12-30

Applicant: GOOGLE INC.

Inventor： Herbert BUCHNER ,  Simon J. GODSILL ,  Jan SKOGLUND

IPC: G10L19/26 ,  G10L21/028 ,  G10L21/0224

CPC classification number: G10L19/26 ,  G10L21/0208 ,  G10L21/0216 ,  G10L21/0224 ,  G10L21/0232 ,  G10L21/028 ,  G10L2021/02161 ,  G10L2021/02165 ,  H04R2410/05

Abstract: Provided are methods and systems for acoustic keystroke transient cancellation/suppression for user communication devices using a semi-blind adaptive filter model. The methods and systems are designed to overcome existing problems in transient noise suppression by taking into account some less-defective signal as side information on the transients and also accounting for acoustic signal propagation, including the reverberation effects, using dynamic models. The methods and systems take advantage of a synchronous reference microphone embedded in the keyboard of the user device, and utilize an adaptive filtering approach exploiting the knowledge of this keybed microphone signal.

公开(公告)号：US20150279386A1

公开(公告)日：2015-10-01

申请号：US14230404

申请日：2014-03-31

Applicant: Google Inc.

Inventor： Jan SKOGLUND ,  Alejandro LUEBS

IPC: G10L21/0208 ,  G10L25/90 ,  G10L25/84

CPC classification number: G10L21/0208 ,  G10L25/78 ,  G10L25/84 ,  G10L25/90

Abstract: Provided are methods and systems for providing situation-dependent transient noise suppression for audio signals. Different strategies (e.g., levels of aggressiveness) of transient suppression and signal restoration are applied to audio signals associated with participants in a video/audio conference depending on whether or not each participant is speaking (e.g., whether a voiced segment or an unvoiced/non-speech segment of audio is present). If no participants are speaking or there is an unvoiced/non-speech sound present, a more aggressive strategy for transient suppression and signal restoration is utilized. On the other hand, where voiced audio is detected (e.g., a participant is speaking), the methods and systems apply a softer, less aggressive suppression and restoration process.

Abstract translation: 提供了用于为音频信号提供与情境相关的瞬态噪声抑制的方法和系统。 瞬态抑制和信号恢复的不同策略（例如，侵略性的水平）被应用于与视频/音频会议中的参与者相关联的音频信号，这取决于每个参与者是否在说话（例如，是否有声音段或无声/非声音 音频片段存在）。 如果没有参与者在说话或存在无声/非语音，则采用更积极的瞬态抑制和信号恢复策略。 另一方面，在检测到有声音频（例如参与者正在说话）的情况下，方法和系统应用更软，更不积极的抑制和恢复过程。

公开(公告)号：US20150199959A1

公开(公告)日：2015-07-16

申请号：US13891978

申请日：2013-05-10

Applicant: Google Inc.

Inventor： Jan SKOGLUND ,  Andrew J. HINES ,  Noami A. HARTE ,  Anil KOKARAM

IPC: G10L15/00

CPC classification number: G10L25/60

Abstract: Methods and systems are provided for using a model of human speech quality perception to provide an objective measure for predicting subjective quality assessments. A Virtual Speech Quality Objective Listener (ViSQOL) model is a signal-based full-reference metric that uses a spectro-temporal measure of similarity between a reference signal and test speech signal. Specifically, the model provides for the ability to detect and predict the level of clock drift, and determine whether such clock drift will impact a listener's quality of experience.

Abstract translation: 提供了使用人类语言质量感知模型的方法和系统来提供用于预测主观质量评估的客观量度。 虚拟语音质量目标监听器（ViSQOL）是一种基于信号的全参考度量，它使用参考信号和测试语音信号之间的相似性的频谱测量。 具体来说，该模型提供了检测和预测时钟漂移水平的能力，并确定这种时钟漂移是否会影响听众的体验质量。

公开(公告)号：US20170221502A1

公开(公告)日：2017-08-03

申请号：US15014481

申请日：2016-02-03

Applicant: Google Inc.

Inventor： Yiteng HUANG ,  Alejandro LUEBS ,  Jan SKOGLUND ,  Willem Bastiaan KLEIJN

IPC: G10L21/0264 ,  G10L25/21 ,  G10L21/0308 ,  G10L21/02

CPC classification number: G10L21/0264 ,  G10L21/0205 ,  G10L21/0216 ,  G10L21/0308 ,  G10L21/0364 ,  G10L25/21 ,  G10L2021/02166 ,  H04R3/005

Abstract: Existing post-filtering methods for microphone array speech enhancement have two common deficiencies. First, they assume that noise is either white or diffuse and cannot deal with point interferers. Second, they estimate the post-filter coefficients using only two microphones at a time, performing averaging over all the microphones pairs, yielding a suboptimal solution. The provided method describes a post-filtering solution that implements signal models which handle white noise, diffuse noise, and point interferers. The method also implements a globally optimized least-squares approach of microphones in a microphone array, providing a more optimal solution than existing conventional methods. Experimental results demonstrate the described method outperforming conventional methods in various acoustic scenarios.

公开(公告)号：US20140112481A1

公开(公告)日：2014-04-24

申请号：US13655225

申请日：2012-10-18

Applicant: Google Inc.

Inventor： Minyue LI ,  Willem Bastiaan KLEIJN ,  Jan SKOGLUND

IPC: H04R5/00

CPC classification number: G10L19/24 ,  G10L19/008 ,  G10L19/0212 ,  G10L19/035 ,  G10L19/167

Abstract: Provided are methods, systems, and apparatus for hierarchical decorrelation of multichannel audio. A hierarchical decorrelation algorithm is designed to adapt to possibly changing characteristics of an input signal, and also preserves the energy of the original signal. The algorithm is invertible in that the original signal can be retrieved if needed. Furthermore, the proposed algorithm decomposes the decorrelation process into multiple low-complexity steps. The contribution of these steps is generally in a decreasing order, and thus the complexity of the algorithm can be scaled.

Abstract translation: 提供了用于多声道音频的分层去相关的方法，系统和装置。 分层去相关算法被设计为适应输入信号的可能变化的特性，并且还保留原始信号的能量。 该算法是可逆的，因为如果需要，可以检索原始信号。 此外，所提出的算法将解相关过程分解为多个低复杂度步骤。 这些步骤的贡献通常是递减的，从而可以缩放算法的复杂性。