Patents by Inventor Che ZHAO

Che ZHAO has filed for patents to protect the following inventions. This listing includes patent applications that are pending as well as patents that have already been granted by the United States Patent and Trademark Office (USPTO).

  • Publication number: 20250006178
    Abstract: Systems and methods are provided for identifying targeted datasets that are configured to facilitate an improvement in the accuracy of an acoustic model included in the automatic speech recognition system. Systems obtain a obtain a test dataset comprising (i) audio data having natural speech utterances and (ii) a transcription of the natural speech utterances. Systems generate a text-to-speech dataset comprising audio data having synthesized speech utterances based on the transcription of the natural speech utterances. Systems apply the test dataset and the text-to-speech dataset to the acoustic model to obtain a first acoustic model output and a second acoustic model output, respectively. Systems identify a first set of errors in the first acoustic model output and a second set of errors in the second acoustic model output. Finally, based on comparing the first set of errors and the second set of errors, an acoustic model error ratio is generated.
    Type: Application
    Filed: November 15, 2021
    Publication date: January 2, 2025
    Inventors: Haoxuan LI, Rui JIANG, Yang LIU, Edward C LIN, Lei SUN, Che ZHAO
  • Publication number: 20240403539
    Abstract: Solutions for custom display post processing (DPP) in speech recognition (SR) use a customized multi-stage DPP pipeline that transforms a stream of SR tokens from lexical form to display form. A first transformation stage of the DPP pipeline receives the stream of tokens, in turn, by an upstream filter, a base model stage, and a downstream filter, and transforms a first aspect of the stream of tokens (e.g., disfluency, inverse text normalization (ITN), capitalization, etc.) from lexical form into display form. The upstream filter and/or the downstream filter alter the stream of tokens to change the default behavior of the DPP pipeline into custom behavior. Additional transformation stages of the DPP pipeline perform further transforms, allowing for outputting final text in a display format that is customized for a specific user. This permits each user to efficiently leverage a common baseline DPP pipeline to produce a custom output.
    Type: Application
    Filed: July 3, 2024
    Publication date: December 5, 2024
    Inventors: Wei LIU, Padma VARADHARAJAN, Piyush BEHRE, Nicholas KIBRE, Edward C. LIN, Shuangyu CHANG, Che ZHAO, Khuram SHAHID, Heiko Willy RAHMEL
  • Patent number: 12061861
    Abstract: Solutions for custom display post processing (DPP) in speech recognition (SR) use a customized multi-stage DPP pipeline that transforms a stream of SR tokens from lexical form to display form. A first transformation stage of the DPP pipeline receives the stream of tokens, in turn, by an upstream filter, a base model stage, and a downstream filter, and transforms a first aspect of the stream of tokens (e.g., disfluency, inverse text normalization (ITN), capitalization, etc.) from lexical form into display form. The upstream filter and/or the downstream filter alter the stream of tokens to change the default behavior of the DPP pipeline into custom behavior. Additional transformation stages of the DPP pipeline perform further transforms, allowing for outputting final text in a display format that is customized for a specific user. This permits each user to efficiently leverage a common baseline DPP pipeline to produce a custom output.
    Type: Grant
    Filed: July 26, 2022
    Date of Patent: August 13, 2024
    Assignee: Microsoft Technology Licensing, LLC.
    Inventors: Wei Liu, Padma Varadharajan, Piyush Behre, Nicholas Kibre, Edward C. Lin, Shuangyu Chang, Che Zhao, Khuram Shahid, Heiko Willy Rahmel
  • Publication number: 20240157032
    Abstract: A high-strength medical fiber composite material includes a sodium alginate hydrogel matrix and a fiber framework. The fiber framework is completely embedded in the sodium alginate hydrogel matrix and formed by compounding supporting layer fibers and reinforcing layer fibers. The reinforcing layer fibers are located above the supporting layer fibers. The reinforcing layer fibers and the supporting layer fibers are orthogonal to each other. According to the high-strength medical fiber composite material prepared in the present invention, the stiffness is improved by 3-4 orders of magnitude, the tensile strength is improved by 2-3 orders of magnitude, and the high-strength medical fiber composite material has high biocompatibility and safety and a great application prospect.
    Type: Application
    Filed: November 29, 2021
    Publication date: May 16, 2024
    Applicant: CHANGZHOU INSTITUTE OF TECHNOLOGY
    Inventors: Che ZHAO, Songxue LIU, Chun FENG, Zhiwei WU, Yiwei ZHANG, Wenbiao JIANG, Xiaozhen LI
  • Publication number: 20230351098
    Abstract: Solutions for custom display post processing (DPP) in speech recognition (SR) use a customized multi-stage DPP pipeline that transforms a stream of SR tokens from lexical form to display form. A first transformation stage of the DPP pipeline receives the stream of tokens, in turn, by an upstream filter, a base model stage, and a downstream filter, and transforms a first aspect of the stream of tokens (e.g., disfluency, inverse text normalization (ITN), capitalization, etc.) from lexical form into display form. The upstream filter and/or the downstream filter alter the stream of tokens to change the default behavior of the DPP pipeline into custom behavior. Additional transformation stages of the DPP pipeline perform further transforms, allowing for outputting final text in a display format that is customized for a specific user. This permits each user to efficiently leverage a common baseline DPP pipeline to produce a custom output.
    Type: Application
    Filed: July 26, 2022
    Publication date: November 2, 2023
    Inventors: Wei LIU, Padma VARADHARAJAN, Piyush BEHRE, Nicholas KIBRE, Edward C. LIN, Shuangyu CHANG, Che ZHAO, Khuram SHAHID, Heiko Willy RAHMEL