公开(公告)号：US11587556B2

公开(公告)日：2023-02-21

申请号：US16594624

申请日：2019-10-07

申请人： Audio Analytic Ltd

发明人： Christopher James Mitchell ,  Sacha Krstulovic ,  Cagdas Bilen ,  Juan Azcarreta Ortiz ,  Giacomo Ferroni ,  Arnoldas Jasonas ,  Francesco Tuveri

IPC分类号： G10L15/16 ,  G10L15/02 ,  G10L15/08 ,  G06N20/20 ,  G06K9/62 ,  G06N3/04 ,  G06N3/084

摘要： A method for recognising at least one of a non-verbal sound event and a scene in an audio signal comprising a sequence of frames of audio data, the method comprising: for each frame of the sequence: processing the frame of audio data to extract multiple acoustic features for the frame of audio data; and classifying the acoustic features to classify the frame by determining, for each of a set of sound classes, a score that the frame represents the sound class; processing the sound class scores for multiple frames of the sequence of frames to generate, for each frame, a sound class decision for each frame; and processing the sound class decisions for the sequence of frames to recognise the at least one of a non-verbal sound event and a scene.

公开(公告)号：US10783434B1

公开(公告)日：2020-09-22

申请号：US16594605

申请日：2019-10-07

申请人： Audio Analytic Ltd

发明人： Christopher James Mitchell ,  Sacha Krstulovic ,  Cagdas Bilen ,  Juan Azcarreta Ortiz ,  Giacomo Ferroni ,  Arnoldas Jasonas ,  Francesco Tuveri

IPC分类号： G10L25/30 ,  G06N3/08 ,  G06N3/04 ,  G10L25/51

摘要： A method of training a non-verbal sound class detection machine learning system, the non-verbal sound class detection machine learning system comprising a machine learning model configured to: receive data for each frame of a sequence of frames of audio data obtained from an audio signal; for each frame of the sequence of frames: process the data for multiple frames; and output data for at least one sound class score representative of a degree of affiliation of the frame with at least one sound class of a plurality of sound classes, wherein the plurality of sound classes comprises: one or more target sound classes; and a non-target sound class representative of an absence of each of the one or more target sound classes; wherein the method comprises: training the machine learning model using a loss function.

公开(公告)号：US20210104230A1

公开(公告)日：2021-04-08

申请号：US16594624

申请日：2019-10-07

申请人： Audio Analytic Ltd.

发明人： Christopher James Mitchell ,  Sacha Krstulovic ,  Cagdas Bilen ,  Juan Azcarreta Ortiz ,  Giacomo Ferroni ,  Amoldas Jasonas ,  Francesco Tuveri

IPC分类号： G10L15/16 ,  G10L15/02 ,  G06K9/62 ,  G06N3/08 ,  G06N3/04 ,  G06N20/20

摘要： A method for recognising at least one of a non-verbal sound event and a scene in an audio signal comprising a sequence of frames of audio data, the method comprising: for each frame of the sequence: processing the frame of audio data to extract multiple acoustic features for the frame of audio data; and classifying the acoustic features to classify the frame by determining, for each of a set of sound classes, a score that the frame represents the sound class; processing the sound class scores for multiple frames of the sequence of frames to generate, for each frame, a sound class decision for each frame; and processing the sound class decisions for the sequence of frames to recognise the at least one of a non-verbal sound event and a scene.