利索-全球专利检索

  1. 登录
  2. 注册

搜索结果

116 条记录 (耗时 0.026 秒)

    AUDIO DEVICE WITH UNCERTAINTY QUANTIFICATION AND RELATED METHODS
    4.
    发明公开
    AUDIO DEVICE WITH UNCERTAINTY QUANTIFICATION AND RELATED METHODS 审中-公开

    公开(公告)号：EP4435781A1

    公开(公告)日：2024-09-25

    申请号：EP23163841.2

    申请日：2023-03-23

    申请人： GN Audio A/S

    发明人： LAROCHE, Clément ,  NOZAL, Diego Caviedes

    IPC分类号： G10L21/02 ,  G10L25/60 ,  G10L25/30

    CPC分类号： G10L25/60 ,  G10L25/30 ,  G10L21/02

    摘要： An audio device comprising memory, an interface, and one or more processors, wherein the one or more processors are configured to obtain audio data; process the audio data for provision of an audio output; process the audio data for provision of one or more audio parameters indicative of one or more characteristics of the audio data; map the one or more audio parameters to a first latent space of a first neural network for provision of a mapping parameter indicative of whether the one or more audio parameters belong to a training manifold of the first latent space ; determine, based on the mapping parameter, an uncertainty parameter indicative of an uncertainty of processing quality; and control the processing of the audio data for provision of the audio output based on the uncertainty parameter.

    A METHOD FOR TRAINING A NEURAL NETWORK AND A DATA PROCESSING DEVICE
    6.
    发明公开
    A METHOD FOR TRAINING A NEURAL NETWORK AND A DATA PROCESSING DEVICE 审中-公开

    公开(公告)号：EP4390922A1

    公开(公告)日：2024-06-26

    申请号：EP22216102.8

    申请日：2022-12-22

    申请人： GN Audio A/S

    发明人： HADDAD, Karim ,  MOWLAEE, Pejman ,  OLSSON, Rasmus Kongsgaard

    IPC分类号： G10L25/30 ,  G10L21/0208

    CPC分类号： G10L2021/0208220130101 ,  G10L25/30 ,  G10L21/0264

    摘要： A computer-implemented method (600) for training a neural network (100) by using a data processing device (102) is presented. The method (600) comprises providing (602) pairs (110) of target audio data sets (106) and distorted audio data sets (108) to the neural network (100), providing (604) values (112) of user preference settings, wherein the values (112) are associated to the pairs (110) of target and distorted audio data sets (106, 108), such that the neural network (100) can be trained by a combination of the pairs (110) of the target audio data (106) and the distorted audio data (108) in combination with the values (112) of the user preference settings, wherein the distorted audio data (108) and the values (112) of the user preference settings are used as inputs to the neural network (100) during training, and wherein a loss function (114) used for training the neural network is a function of the values (112) of the user preference settings, the target audio data (106), and the distorted audio data (108) thereby providing for that a variety of user preference settings can be met with one and the same trained neural network.

    HEADSET WITH ACTIVE NOISE CANCELLATION
    10.
    发明公开
    HEADSET WITH ACTIVE NOISE CANCELLATION 审中-公开

    公开(公告)号：EP3720144A1

    公开(公告)日：2020-10-07

    申请号：EP20167783.8

    申请日：2020-04-02

    申请人： GN Audio A/S

    发明人： Hoby, Michael ,  Bovbjerg, Christoffer

    IPC分类号： H04R3/00

    摘要： The present invention relates to an earphone with active noise cancellation and to a headset with such an earphone.
    The earphone (1) is adapted to be arranged in an operating position at a first ear (5) of a user (4) and, in the operating position, provide a first acoustic output signal (So) to the first ear (5). The earphone comprises: a first input unit (31) adapted to receive a first audio input signal (S I ); a first noise cancellation signal path (37) with a first microphone (26) and a first noise cancellation filter (32), wherein the first microphone (26) is arranged to receive ambient sound (S A ) from ambient space (7) with the earphone in the operating position and is adapted to provide a corresponding first reference signal (S R ), and wherein the first noise cancellation filter (32) is adapted to apply a first transfer function (H) to the first reference signal (S R ) to provide a first noise cancellation signal (S C ); a first output unit (33) adapted to provide a first audio output signal (S D ) by combining the first audio input signal (S I ) and the first noise cancellation signal (Sc); a first electroacoustic transducer (23) adapted to provide the first acoustic output signal (So) in dependence on the first audio output signal (S D ); and a first noise cancellation controller (34) adapted to adaptively control the first transfer function (H) of the first noise cancellation filter (32) to cause the first acoustic output signal (S O ) to counteract ambient sound (S A ), such that, with the earphone in the operating position, the level of ambient sound (S A ) arriving at the first ear (5) is reduced, while still allowing desired sound from the first audio input signal (S I ), to pass through.
    The earphone (1) is characterized in that it further comprises a level analyzer (35) adapted to provide, based on an analysis of the first audio input signal (S I ) and/or the first reference signal (S R ), a sound level estimate (L S ) indicating a total sound level at the first ear (5) and to compare the sound level estimate (L S ) with a predetermined threshold (L T ) indicating a noise floor level; and in that the first noise cancellation controller (34) further is adapted to, in dependence on the comparison, control the wide-band gain (G) of the first noise cancellation signal path (37) to cause a decrease (ΔG) of the wide-band gain (G) in time periods wherein the total sound level at the first ear (5) is below the noise floor level compared to time periods wherein the total sound level at the first ear (5) is above the noise floor level.
    The invention may e.g. be used to reduce noise perceived by a user while listening to speech or music through an active noise cancellation earphone and/or to improve the comfort perceived by the user when using an active noise cancellation headset in low-noise environments.