公开(公告)号：US20150248893A1

公开(公告)日：2015-09-03

申请号：US14194192

申请日：2014-02-28

Applicant: Google Inc.

Inventor： Willem Bastiaan KLEIJN ,  Turaj Zakizadeh SHABESTARY

IPC: G10L19/02

CPC classification number: G10L19/02 ,  G10H2210/051 ,  G10H2210/061 ,  G10H2250/235 ,  G10H2250/541 ,  G10L15/14 ,  G10L19/005 ,  G10L19/04 ,  G10L19/06

Abstract: Provided are methods and systems for concealing missing segments and/or discontinuities in an audio signal, thereby restoring the continuity of the signal. The methods and systems are designed for and targeted at audio signals, are based on interpolation and extrapolation operations for sinusoids, and do not rely on the assumption that the sinusoids are harmonic. The methods and systems are improvements over existing audio concealment approaches in that, among other advantages, the methods and systems facilitate asynchronous interpolation, use an interpolation procedure that corresponds to time-domain waveform interpolation if the signal is harmonic, and have a peak selection procedure that is effective for audio signals.

Abstract translation: 提供了用于隐藏音频信号中的丢失段和/或不连续性的方法和系统，从而恢复信号的连续性。 该方法和系统设计用于和针对音频信号，基于正弦曲线的插值和外插操作，并且不依赖于正弦曲线是谐波的假设。 方法和系统是对现有音频隐藏方法的改进，其中除了其他优点之外，方法和系统还促进异步插值，如果信号是谐波，则使用对应于时域波形插值的内插过程，并且具有峰值选择程序 这对音频信号有效。

公开(公告)号：US20170243586A1

公开(公告)日：2017-08-24

申请号：US15052426

申请日：2016-02-24

Applicant: GOOGLE INC.

Inventor： Jay Pierre CIVELLI ,  Mikhal SHEMER ,  Turaj Zakizadeh SHABESTARY ,  David TAPUSKA

IPC: G10L15/30 ,  G10L15/22

CPC classification number: G10L15/30 ,  G10L15/02 ,  G10L15/22 ,  G10L15/32 ,  G10L2015/088 ,  G10L2015/223

Abstract: Provided are methods, systems, and apparatuses for detecting, processing, and responding to audio signals, including speech signals, within a designated area or space. A platform for multiple media devices connected via a network is configured to process speech, such as voice commands, detected at the media devices, and respond to the detected speech by causing the media devices to simultaneously perform one or more requested actions. The platform is capable of scoring the quality of a speech request, handling speech requests from multiple end points of the platform using a centralized processing approach, a de-centralized processing approach, or a combination thereof, and also manipulating partial processing of speech requests from multiple end points into a coherent whole when necessary.

公开(公告)号：US20150201274A1

公开(公告)日：2015-07-16

申请号：US13780728

申请日：2013-02-28

Applicant: Google Inc.

Inventor： Lars Henrik ELLNER ,  Turaj Zakizadeh SHABESTARY ,  Niklas ENBOM

IPC: H04R3/00

CPC classification number: H04R3/00 ,  G10L19/008 ,  G10L25/78 ,  H04M1/60 ,  H04R2499/11

Abstract: Methods and systems are provided for reusing audio streams in an audio mixing environment such that processing resources (e.g., CPU cycles) are not wasted on encoding the same audio streams multiple times (e.g., more than once). Instead, processing resources are conserved without comprising audio quality. The methods described herein are particularly useful when a participant transitions from using a shared encoder to using the participant's own encoder, and when the participant transitions from using the participant's own encoder to using a shared encoder.

Abstract translation: 提供了用于在音频混合环境中重用音频流的方法和系统，使得在对相同音频流多次编码（例如，不止一次）时，处理资源（例如，CPU周期）不被浪费。 相反，处理资源是保守的，而不包括音频质量。 当参与者从使用共享编码器转换到使用参与者自己的编码器以及当参与者从使用参与者自己的编码器转换到使用共享编码器时，这里描述的方法特别有用。