知搜-全球专利检索

  1. Login
  2. Sign Up

Search Results

9 records (took 0.017 seconds)

    INITIALIZATION OF CTC SPEECH RECOGNITION WITH STANDARD HMM
    2.
    发明申请
    INITIALIZATION OF CTC SPEECH RECOGNITION WITH STANDARD HMM 审中-公开

    公开(公告)号：US20190013015A1

    公开(公告)日：2019-01-10

    申请号：US15645985

    申请日：2017-07-10

    Applicant: Sony Interactive Entertainment Inc.

    Inventor： Xavier Menendez-Pidal ,  Ruxin Chen

    IPC: G10L15/14 ,  G10L15/16 ,  G10L15/02

    CPC classification number: G10L15/144 ,  G10L15/02 ,  G10L15/063 ,  G10L15/16 ,  G10L2015/025

    Abstract: A method for improved initialization of speech recognition system comprises mapping a trained hidden markov model based recognition node network (HMM) to a Connectionist Temporal Classification (CTC) based node label scheme. The central state of each frame in the HMM are mapped to CTC-labeled output nodes and the non-central states of each frame are mapped to CTC-blank nodes to generate a CTC-labeled HMM and each central state represents a phoneme from human speech detected and extracted by a computing device. Next the CTC-labeled HMM is trained using a cost function, wherein the cost function is not part of a CTC cost function. Finally the CTC-labeled HMM is trained using a CTC cost function to produce a CTC node network. The CTC node network may be iteratively trained by repeating the initialization steps.

    AUDIO, VIDEO, SIMULATION, AND USER INTERFACE PARADIGMS
    3.
    发明申请
    AUDIO, VIDEO, SIMULATION, AND USER INTERFACE PARADIGMS 审中-公开
    音频，视频，模拟和用户界面参数

    公开(公告)号：US20160310847A1

    公开(公告)日：2016-10-27

    申请号：US15199384

    申请日：2016-06-30

    Applicant: Sony Interactive Entertainment Inc.

    Inventor： Gustavo Hernandez-Abrego ,  Xavier Menendez-Pidal ,  Steven Osman ,  Ruxin Chen ,  Rishi Deshpande ,  Care Michaud-Wideman ,  Richard Marks ,  Eric J. Larsen ,  Xiaodong Mao

    IPC: A63F13/428 ,  A63F13/213 ,  G06F3/00 ,  G06F3/01 ,  G10L17/04

    CPC classification number: A63F13/428 ,  A63F13/213 ,  A63F13/217 ,  G06F3/005 ,  G06F3/012 ,  G06F3/015 ,  G06F2203/011 ,  G10L17/005 ,  G10L17/04

    Abstract: Consumer electronic devices have been developed with enormous information processing capabilities, high quality audio and video outputs, large amounts of memory, and may also include wired and/or wireless networking capabilities. Additionally, relatively unsophisticated and inexpensive sensors, such as microphones, video camera, GPS or other position sensors, when coupled with devices having these enhanced capabilities, can be used to detect subtle features about users and their environments. A variety of audio, video, simulation and user interface paradigms have been developed to utilize the enhanced capabilities of these devices. These paradigms can be used separately or together in any combination. One paradigm automatically creating user identities using speaker identification. Another paradigm includes a control button with 3-axis pressure sensitivity for use with game controllers and other input devices.

    Abstract translation: 消费电子设备已经开发出了巨大的信息处理能力，高质量的音频和视频输出，大量的存储器，还可以包括有线和/或无线网络功能。 此外，当与具有这些增强能力的设备耦合时，相对不成熟和便宜的传感器（例如麦克风，摄像机，GPS或其他位置传感器）可用于检测关于用户及其环境的微妙特征。 已经开发出各种音频，视频，仿真和用户界面范例来利用这些设备的增强功能。 这些范例可以单独使用或以任何组合一起使用。 一个范例使用说话人识别自动创建用户身份。 另一种范例包括一个具有3轴压力敏感度的控制按钮，用于游戏控制器和其他输入设备。

    Initialization of CTC speech recognition with standard HMM
    6.
    发明授权
    Initialization of CTC speech recognition with standard HMM 有权

    公开(公告)号：US10714076B2

    公开(公告)日：2020-07-14

    申请号：US15645985

    申请日：2017-07-10

    Applicant: Sony Interactive Entertainment Inc.

    Inventor： Xavier Menendez-Pidal ,  Ruxin Chen

    IPC: G10L15/00 ,  G10L15/16 ,  G10L15/14 ,  G10L15/02 ,  G10L15/06

    Abstract: A method for improved initialization of speech recognition system comprises mapping a trained hidden markov model based recognition node network (HMM) to a Connectionist Temporal Classification (CTC) based node label scheme. The central state of each frame in the HMM are mapped to CTC-labeled output nodes and the non-central states of each frame are mapped to CTC-blank nodes to generate a CTC-labeled HMM and each central state represents a phoneme from human speech detected and extracted by a computing device. Next the CTC-labeled HMM is trained using a cost function, wherein the cost function is not part of a CTC cost function. Finally the CTC-labeled HMM is trained using a CTC cost function to produce a CTC node network. The CTC node network may be iteratively trained by repeating the initialization steps.

Patent Agency Ranking