公开(公告)号：US20150325251A1

公开(公告)日：2015-11-12

申请号：US14274544

申请日：2014-05-09

Applicant: Apple Inc.

Inventor： Sorin V. Dusan ,  Aram M. Lindahl ,  Alexander Kanaris ,  Vasu Iyengar

IPC: G10L21/0208

CPC classification number: G10L21/0216 ,  G10L21/0208 ,  G10L2021/02166

Abstract: Electronic system for audio noise processing and noise reduction comprises: first and second noise estimators, selector and attenuator. First noise estimator processes first audio signal from voice beamformer (VB) and generate first noise estimate. VB generates first audio signal by beamforming audio signals from first and second audio pick-up channels. Second noise estimator processes first and second audio signal from noise beamformer (NB), in parallel with first noise estimator and generates second noise estimate. NB generates second audio signal by beamforming audio signals from first and second audio pick-up channels. First and second audio signals include frequencies in first and second frequency regions. Selector's output noise estimate may be a) second noise estimate in the first frequency region, and b) first noise estimate in the second frequency region. Attenuator attenuates first audio signal in accordance with output noise estimate. Other embodiments are also described.

Abstract translation: 用于音频噪声处理和降噪的电子系统包括：第一和第二噪声估计器，选择器和衰减器。 第一噪声估计器处理来自语音波束形成器（VB）的第一音频信号并产生第一噪声估计。 VB通过来自第一和第二音频拾取通道的波束成形音频信号产生第一音频信号。 第二噪声估计器与第一噪声估计器并行地处理来自噪声波束形成器（NB）的第一和第二音频信号，并产生第二噪声估计。 NB通过波束成形来自第一和第二音频拾取通道的音频信号产生第二音频信号。 第一和第二音频信号包括第一和第二频率区域中的频率。 选择器的输出噪声估计可以是a）第一频率区域中的第二噪声估计，以及b）第二频率区域中的第一噪声估计。 衰减器根据输出噪声估计衰减第一音频信号。 还描述了其它实施例。

公开(公告)号：US10553235B2

公开(公告)日：2020-02-04

申请号：US16256587

申请日：2019-01-24

Applicant: Apple Inc.

Inventor： Nicholas J. Bryan ,  Vasu Iyengar ,  Aram M. Lindahl

IPC: G10L21/028 ,  G10L21/02 ,  G10L21/0232 ,  G10L17/00

Abstract: A method for controlling a speech enhancement process in a far-end device, while engaged in a voice or video telephony communication session over a communication link with a near-end device. A near-end user speech signal is produced, using a microphone to pick up speech of a near-end user, and is analyzed by an automatic speech recognizer (ASR) without being triggered by an ASR trigger phrase or button. The recognized words are compared to a library of phrases to select a matching phrase, where each phrase is associated with a message that represents an audio signal processing operation. The message associated with the matching phrase is sent to the far-end device, which is used to configure the far-end device to adjust the speech enhancement process that produces the far-end speech signal. Other embodiments are also described.

公开(公告)号：US10535362B2

公开(公告)日：2020-01-14

申请号：US15909513

申请日：2018-03-01

Applicant: Apple Inc.

Inventor： Nicholas J. Bryan ,  Vasu Iyengar

IPC: G10L21/0232 ,  G10L21/028 ,  H04R3/00 ,  G10L21/02 ,  G10L21/0216

Abstract: Signals are received from audio pickup channels that contain signals from multiple sound sources. The audio pickup channels may include one or more microphones and one or more accelerometers. Signals representative of multiple sound sources are generated using a blind source separation algorithm. It is then determined which of those signals is deemed to be a voice signal and which is deemed to be a noise signal. The output noise signal may be scaled to match a level of the output voice signal, and a clean speech signal is generated based on the output voice signal and the scaled noise signal. Other aspects are described.

公开(公告)号：US20190272842A1

公开(公告)日：2019-09-05

申请号：US15909513

申请日：2018-03-01

Applicant: Apple Inc.

Inventor： Nicholas J. Bryan ,  Vasu Iyengar

IPC: G10L21/0232 ,  G10L21/028 ,  G10L21/02 ,  H04R3/00

Abstract: Signals are received from audio pickup channels that contain signals from multiple sound sources. The audio pickup channels may include one or more microphones and one or more accelerometers. Signals representative of multiple sound sources are generated using a blind source separation algorithm. It is then determined which of those signals is deemed to be a voice signal and which is deemed to be a noise signal. The output noise signal may be scaled to match a level of the output voice signal, and a clean speech signal is generated based on the output voice signal and the scaled noise signal. Other aspects are described.

公开(公告)号：US10090001B2

公开(公告)日：2018-10-02

申请号：US15225595

申请日：2016-08-01

Applicant: Apple Inc.

Inventor： Lalin S. Theverapperuma ,  Vasu Iyengar ,  Sarmad Aziz Malik ,  Raghavendra Prabhu

IPC: G10L25/30 ,  G10L21/0232 ,  G10L25/72 ,  G10L25/84 ,  G10L21/028

Abstract: Method of speech enhancement using Neural Network-based combined signal starts with training neural network offline which includes: (i) exciting at least one accelerometer and at least one microphone using training accelerometer signal and training acoustic signal, respectively. The training accelerometer signal and the training acoustic signal are correlated during clean speech segments. Training neural network offline further includes(ii) selecting speech included in the training accelerometer signal and in the training acoustic signal, and (iii) spatially localizing the speech by setting a weight parameter in the neural network based on the selected speech included in the training accelerometer signal and in the training acoustic signal. The neural network that is trained offline is then used to generate a speech reference signal based on an accelerometer signal from the at least one accelerometer and an acoustic signal received from the at least one microphone. Other embodiments are described.

公开(公告)号：US20180033449A1

公开(公告)日：2018-02-01

申请号：US15225595

申请日：2016-08-01

Applicant: Apple Inc.

Inventor： Lalin S. Theverapperuma ,  Vasu Iyengar ,  Sarmad Aziz Malik ,  Raghavendra Prabhu

IPC: G10L25/30 ,  G10L25/72 ,  G10L21/0232 ,  G10L25/84 ,  G10L21/028

CPC classification number: G10L25/30 ,  G10L21/0232 ,  G10L21/028 ,  G10L25/72 ,  G10L25/84

Abstract: Method of speech enhancement using Neural Network-based combined signal starts with training neural network offline which includes: (i) exciting at least one accelerometer and at least one microphone using training accelerometer signal and training acoustic signal, respectively. The training accelerometer signal and the training acoustic signal are correlated during clean speech segments. Training neural network offline further includes (ii) selecting speech included in the training accelerometer signal and in the training acoustic signal, and (iii) spatially localizing the speech by setting a weight parameter in the neural network based on the selected speech included in the training accelerometer signal and in the training acoustic signal. The neural network that is trained offline is then used to generate a speech reference signal based on an accelerometer signal from the at least one accelerometer and an acoustic signal received from the at least one microphone. Other embodiments are described.

公开(公告)号：US20130332157A1

公开(公告)日：2013-12-12

申请号：US13911915

申请日：2013-06-06

Applicant: Apple Inc.

Inventor： Vasu Iyengar ,  Sorin V. Dusan

IPC: G10L15/20

CPC classification number: G10L15/20 ,  G10L21/0224 ,  G10L21/0232 ,  G10L2021/02165 ,  G10L2025/783

Abstract: Digital signal processing techniques for automatically reducing audible noise from a sound recording that contains speech. A noise suppression system uses two types of noise estimators, including a more aggressive one and less aggressive one. Decisions are made on how to select or combine their outputs into a usable noise estimate in a different speech and noise conditions. A 2-channel noise estimator is described. Other embodiments are also described and claimed.

Abstract translation: 用于自动降低包含语音的录音的可听噪声的数字信号处理技术。 噪声抑制系统使用两种类型的噪声估计器，包括更具侵略性的一个或更少的噪声估计器。 决定如何在不同的语音和噪声条件下将其输出选择或组合成可用的噪声估计。 描述2通道噪声估计器。 还描述和要求保护其他实施例。