公开(公告)号：US10861479B2

公开(公告)日：2020-12-08

申请号：US16598462

申请日：2019-10-10

Applicant: GOOGLE LLC

Inventor： Turaj Zakizadeh Shabestary ,  Willem Bastiaan Kleijn ,  Jan Skoglund

IPC: G10L21/0208 ,  G10L21/0232 ,  G10L15/08 ,  H04R3/04 ,  H04M9/08

Abstract: Techniques of performing linear acoustic echo cancellation performing a phase correction operation on the estimate of the echo signal based on a clock drift between a capture of an input microphone signal and a playout of a loudspeaker signal. Along these lines, the existence of the clock drift, i.e., a small difference in the sampling rates of the input microphone signal and the loudspeaker signal, can cause processing circuitry in a device configured to perform LAEC operations to generate a filter based on the magnitudes of the short-term Fourier transforms (STFTs) of the input microphone signal and the loudspeaker signal. Such a filter is real-valued and results in a positive estimate of the acoustic echo signal included in the input microphone signal. The phase of this estimate may then be aligned with the phase of the input microphone signal.

公开(公告)号：US20200176004A1

公开(公告)日：2020-06-04

申请号：US16206823

申请日：2018-11-30

Applicant: Google LLC

Inventor： Willem Bastiaan Kleijn ,  Jan K. Skoglund ,  Alejandro Luebs ,  Sze Chie Lim

IPC: G10L19/02 ,  G10L25/30

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for coding speech using neural networks. One of the methods includes obtaining a bitstream of parametric coder parameters characterizing spoken speech; generating, from the parametric coder parameters, a conditioning sequence; generating a reconstruction of the spoken speech that includes a respective speech sample at each of a plurality of decoder time steps, comprising, at each decoder time step: processing a current reconstruction sequence using an auto-regressive generative neural network, wherein the auto-regressive generative neural network is configured to process the current reconstruction to compute a score distribution over possible speech sample values, and wherein the processing comprises conditioning the auto-regressive generative neural network on at least a portion of the conditioning sequence; and sampling a speech sample from the possible speech sample values.

公开(公告)号：US20190392853A1

公开(公告)日：2019-12-26

申请号：US16019402

申请日：2018-06-26

Applicant: GOOGLE LLC

Inventor： Willem Bastiaan Kleijn ,  Turaj Zakizadeh Shabestary

IPC: G10L21/0208 ,  H04M9/08 ,  G06F3/16

Abstract: According to an aspect, a method for multi-channel echo cancellation includes receiving a microphone signal and a multi-channel loudspeaker driving signal. The multi-channel loudspeaker driving signal includes a first driving signal that drives a first loudspeaker, and a second driving signal that drives a second loudspeaker. The first driving signal is substantially the same as second driving signal. The microphone signal includes a near-end signal with echo. The method includes determining a unique solution for acoustic transfer functions for a present acoustic scenario based on the microphone signal and the multi-channel loudspeaker driving signal. The acoustic transfer functions include first and second acoustic transfer function. The unique solution is determined based on time-frequency transforms of observations from the present acoustic scenario and at least one previous acoustic scenario. The method includes removing the echo from the microphone signal based on the first and second acoustic transfer function.

公开(公告)号：US10264386B1

公开(公告)日：2019-04-16

申请号：US15893138

申请日：2018-02-09

Applicant: Google LLC

Inventor： Willem Bastiaan Kleijn

IPC: H04R25/00 ,  H04S7/00 ,  H04S3/00

Abstract: Techniques of rendering high-order ambisonics (HOAs) involve adjusting the weights of a spherical harmonic (SH) expansion of a sound field based on weights of a SH expansion of a direction emphasis function that multiplies a monopole density that, when its product with a Green's function is integrated over the unit sphere, produces the sound field. An advantage of the improved techniques lies in the ability to better reproduce directionality of a given sound field in a computationally manner, whether the sound field is a temporal function or a time-frequency function.

公开(公告)号：US20180174598A1

公开(公告)日：2018-06-21

申请号：US15846049

申请日：2017-12-18

Applicant: GOOGLE LLC

Inventor： Turaj Zakizadeh Shabestary ,  Willem Bastiaan Kleijn ,  Jan Skoglund

IPC: G10L21/0232 ,  G10L15/08 ,  H04R3/04

CPC classification number: G10L21/0232 ,  G10L15/08 ,  G10L21/0208 ,  G10L2015/088 ,  G10L2021/02082 ,  H04M9/082 ,  H04R3/04

Abstract: Techniques of performing linear acoustic echo cancellation performing a phase correction operation on the estimate of the echo signal based on a clock drift between a capture of an input microphone signal and a playout of a loudspeaker signal. Along these lines, the existence of the clock drift, i.e., a small difference in the sampling rates of the input microphone signal and the loudspeaker signal, can cause processing circuitry in a device configured to perform LAEC operations to generate a filter based on the magnitudes of the short-term Fourier transforms (STFTs) of the input microphone signal and the loudspeaker signal. Such a filter is real-valued and results in a positive estimate of the acoustic echo signal included in the input microphone signal. The phase of this estimate may then be aligned with the phase of the input microphone signal.

公开(公告)号：US20190259397A1

公开(公告)日：2019-08-22

申请号：US16404076

申请日：2019-05-06

Applicant: Google LLC

Inventor： Willem Bastiaan Kleijn ,  Jan Skoglund ,  Sze Chie Lim

IPC: G10L19/008 ,  G10L19/20 ,  H04S3/00 ,  G10L19/24

Abstract: A method includes: receiving a representation of a soundfield, the representation characterizing the soundfield around a point in space; decomposing the received representation into independent signals; and encoding the independent signals, wherein a quantization noise for any of the independent signals has a common spatial profile with the independent signal.

公开(公告)号：US20180212690A1

公开(公告)日：2018-07-26

申请号：US15926808

申请日：2018-03-20

Applicant: GOOGLE LLC

Inventor： Willem Bastiaan Kleijn ,  Andrew Allen

IPC: H04B15/00 ,  G10L21/0364

CPC classification number: H04B15/00 ,  G10L21/0364

Abstract: Provided are methods and systems for improving the intelligibility of speech in a noisy environment. A communication model is developed that includes noise inherent in the message production and message interpretation processes, and considers that these noises have fixed signal-to-noise ratios. The communication model forms the basis of an algorithm designed to optimize the intelligibility of speech in a noisy environment. The intelligibility optimization algorithm only does something (e.g., manipulates the audio signal) when needed, and thus if no noise is present the algorithm does not alter or otherwise interfere with the audio signals, thereby preventing any speech distortion. The algorithm is also very fast and efficient in comparison to most existing approaches for speech intelligibility enhancement, and therefore the algorithm lends itself to easy implementation in an appropriate device (e.g., cellular phone or smartphone).

公开(公告)号：US12062380B2

公开(公告)日：2024-08-13

申请号：US18144413

申请日：2023-05-08

Applicant: Google LLC

Inventor： Willem Bastiaan Kleijn ,  Jan K. Skoglund ,  Alejandro Luebs ,  Sze Chie Lim

IPC: G10L19/02 ,  G10L25/30

CPC classification number: G10L19/0204 ,  G10L25/30

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for coding speech using neural networks. One of the methods includes obtaining a bitstream of parametric coder parameters characterizing spoken speech; generating, from the parametric coder parameters, a conditioning sequence; generating a reconstruction of the spoken speech that includes a respective speech sample at each of a plurality of decoder time steps, comprising, at each decoder time step: processing a current reconstruction sequence using an auto-regressive generative neural network, wherein the auto-regressive generative neural network is configured to process the current reconstruction to compute a score distribution over possible speech sample values, and wherein the processing comprises conditioning the auto-regressive generative neural network on at least a portion of the conditioning sequence; and sampling a speech sample from the possible speech sample values.

公开(公告)号：US11676613B2

公开(公告)日：2023-06-13

申请号：US17332898

申请日：2021-05-27

Applicant: Google LLC

Inventor： Willem Bastiaan Kleijn ,  Jan K. Skoglund ,  Alejandro Luebs ,  Sze Chie Lim

IPC: G10L19/02 ,  G10L25/30

CPC classification number: G10L19/0204 ,  G10L25/30

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for coding speech using neural networks. One of the methods includes obtaining a bitstream of parametric coder parameters characterizing spoken speech; generating, from the parametric coder parameters, a conditioning sequence; generating a reconstruction of the spoken speech that includes a respective speech sample at each of a plurality of decoder time steps, comprising, at each decoder time step: processing a current reconstruction sequence using an auto-regressive generative neural network, wherein the auto-regressive generative neural network is configured to process the current reconstruction to compute a score distribution over possible speech sample values, and wherein the processing comprises conditioning the auto-regressive generative neural network on at least a portion of the conditioning sequence; and sampling a speech sample from the possible speech sample values.

公开(公告)号：US11024321B2

公开(公告)日：2021-06-01

申请号：US16206823

申请日：2018-11-30

Applicant: Google LLC

Inventor： Willem Bastiaan Kleijn ,  Jan K. Skoglund ,  Alejandro Luebs ,  Sze Chie Lim

IPC: G10L19/02 ,  G10L25/30

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for coding speech using neural networks. One of the methods includes obtaining a bitstream of parametric coder parameters characterizing spoken speech; generating, from the parametric coder parameters, a conditioning sequence; generating a reconstruction of the spoken speech that includes a respective speech sample at each of a plurality of decoder time steps, comprising, at each decoder time step: processing a current reconstruction sequence using an auto-regressive generative neural network, wherein the auto-regressive generative neural network is configured to process the current reconstruction to compute a score distribution over possible speech sample values, and wherein the processing comprises conditioning the auto-regressive generative neural network on at least a portion of the conditioning sequence; and sampling a speech sample from the possible speech sample values.