Patents by Inventor Ke Hu

Ke Hu 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).

  • Patent number: 11908461
    Abstract: A method of performing speech recognition using a two-pass deliberation architecture includes receiving a first-pass hypothesis and an encoded acoustic frame and encoding the first-pass hypothesis at a hypothesis encoder. The first-pass hypothesis is generated by a recurrent neural network (RNN) decoder model for the encoded acoustic frame. The method also includes generating, using a first attention mechanism attending to the encoded acoustic frame, a first context vector, and generating, using a second attention mechanism attending to the encoded first-pass hypothesis, a second context vector. The method also includes decoding the first context vector and the second context vector at a context vector decoder to form a second-pass hypothesis.
    Type: Grant
    Filed: January 14, 2021
    Date of Patent: February 20, 2024
    Assignee: Google LLC
    Inventors: Ke Hu, Tara N. Sainath, Ruoming Pang, Rohit Prakash Prabhavalkar
  • Publication number: 20230298563
    Abstract: A method of text-only and semi-supervised training for deliberation includes receiving training data including unspoken textual utterances that are each not paired with any corresponding spoken utterance of non-synthetic speech, and training a deliberation model that includes a text encoder and a deliberation decoder on the unspoken textual utterances. The method also includes receiving, at the trained deliberation model, first-pass hypotheses and non-causal acoustic embeddings. The first-pass hypotheses is generated by a recurrent neural network-transducer (RNN-T) decoder for the non-causal acoustic embeddings encoded by a non-causal encoder. The method also includes encoding, using the text encoder, the first-pass hypotheses generated by the RNN-T decoder, and generating, using the deliberation decoder attending to both the first-pass hypotheses and the non-causal acoustic embeddings, second-pass hypotheses.
    Type: Application
    Filed: March 18, 2023
    Publication date: September 21, 2023
    Applicant: Google LLC
    Inventors: Ke Hu, Tara N. Sainath, Yanzhang He, Rohit Prabhavalkar, Sepand Mavandadi, Weiran Wang, Trevor Strohman
  • Publication number: 20230186907
    Abstract: A method of performing speech recognition using a two-pass deliberation architecture includes receiving a first-pass hypothesis and an encoded acoustic frame and encoding the first-pass hypothesis at a hypothesis encoder. The first-pass hypothesis is generated by a recurrent neural network (RNN) decoder model for the encoded acoustic frame. The method also includes generating, using a first attention mechanism attending to the encoded acoustic frame, a first context vector, and generating, using a second attention mechanism attending to the encoded first-pass hypothesis, a second context vector.
    Type: Application
    Filed: February 6, 2023
    Publication date: June 15, 2023
    Applicant: Google LLC
    Inventors: Ke Hu, Tara N. Sainath, Ruoming Pang, Rohit Prakash Prabhavalkar
  • Publication number: 20230107248
    Abstract: A method includes receiving an initial alignment for a candidate hypothesis generated by a transducer decoder model during a first pass. Here, the candidate hypothesis corresponds to a candidate transcription for an utterance and the initial alignment for the candidate hypothesis includes a sequence of output labels. Each output label corresponds to a blank symbol or a hypothesized sub-word unit. The method also include receiving a subsequent sequence of audio encodings characterizing the utterance. During an initial refinement step, the method also includes generating a new alignment for a rescored sequence of output labels using a non-autoregressive decoder. The non-autoregressive decoder is configured to receive the initial alignment for the candidate hypothesis and the subsequent sequence of audio encodings.
    Type: Application
    Filed: September 16, 2022
    Publication date: April 6, 2023
    Applicant: Google LLC
    Inventors: Weiran Wang, Ke Hu, Tara N. Sainath
  • Publication number: 20230109407
    Abstract: A method includes receiving a sequence of acoustic frames and generating, by a first encoder, a first higher order feature representation for a corresponding acoustic frame in the sequence of acoustic frames. The method also includes generating, by a first pass transducer decoder, a first pass speech recognition hypothesis for a corresponding first higher order feature representation and generating, by a text encoder, a text encoding for a corresponding first pass speech recognition hypothesis. The method also includes generating, by a second encoder, a second higher order feature representation for a corresponding first higher order feature representation. The method also includes generating, by a second pass transducer decoder, a second pass speech recognition hypothesis using a corresponding second higher order feature representation and a corresponding text encoding.
    Type: Application
    Filed: September 19, 2022
    Publication date: April 6, 2023
    Applicant: Google LLC
    Inventors: Ke Hu, Tara N. Sainath, Arun Narayanan, Ruoming Pang, Trevor Strohman
  • Patent number: 11610586
    Abstract: A method includes receiving a speech recognition result, and using a confidence estimation module (CEM), for each sub-word unit in a sequence of hypothesized sub-word units for the speech recognition result: obtaining a respective confidence embedding that represents a set of confidence features; generating, using a first attention mechanism, a confidence feature vector; generating, using a second attention mechanism, an acoustic context vector; and generating, as output from an output layer of the CEM, a respective confidence output score for each corresponding sub-word unit based on the confidence feature vector and the acoustic feature vector received as input by the output layer of the CEM. For each of the one or more words formed by the sequence of hypothesized sub-word units, the method also includes determining a respective word-level confidence score for the word. The method also includes determining an utterance-level confidence score by aggregating the word-level confidence scores.
    Type: Grant
    Filed: February 23, 2021
    Date of Patent: March 21, 2023
    Assignee: Google LLC
    Inventors: David Qiu, Qiujia Li, Yanzhang He, Yu Zhang, Bo Li, Liangliang Cao, Rohit Prabhavalkar, Deepti Bhatia, Wei Li, Ke Hu, Tara Sainath, Ian Mcgraw
  • Publication number: 20220270597
    Abstract: A method includes receiving a speech recognition result, and using a confidence estimation module (CEM), for each sub-word unit in a sequence of hypothesized sub-word units for the speech recognition result: obtaining a respective confidence embedding that represents a set of confidence features; generating, using a first attention mechanism, a confidence feature vector; generating, using a second attention mechanism, an acoustic context vector; and generating, as output from an output layer of the CEM, a respective confidence output score for each corresponding sub-word unit based on the confidence feature vector and the acoustic feature vector received as input by the output layer of the CEM. For each of the one or more words formed by the sequence of hypothesized sub-word units, the method also includes determining a respective word-level confidence score for the word. The method also includes determining an utterance-level confidence score by aggregating the word-level confidence scores.
    Type: Application
    Filed: February 23, 2021
    Publication date: August 25, 2022
    Applicant: Google LLC
    Inventors: David Qiu, Qiujia Li, Yanzhang He, Yu Zhang, Bo Li, Liangliang Cao, Rohit Prabhavalkar, Deepti Bhatia, Wei Li, Ke Hu, Tara Sainath, Ian Mcgraw
  • Publication number: 20220172706
    Abstract: A method includes receiving audio data encoding an utterance spoken by a native speaker of a first language, and receiving a biasing term list including one or more terms in a second language different than the first language. The method also includes processing, using a speech recognition model, acoustic features derived from the audio data to generate speech recognition scores for both wordpieces and corresponding phoneme sequences in the first language. The method also includes rescoring the speech recognition scores for the phoneme sequences based on the one or more terms in the biasing term list, and executing, using the speech recognition scores for the wordpieces and the rescored speech recognition scores for the phoneme sequences, a decoding graph to generate a transcription for the utterance.
    Type: Application
    Filed: February 16, 2022
    Publication date: June 2, 2022
    Applicant: Google LLC
    Inventors: Ke Hu, Golan Pundak, Rohit Prakash Prabhavalkar, Antoine Jean Bruguier, Tara N. Sainath
  • Patent number: 11270687
    Abstract: A method includes receiving audio data encoding an utterance spoken by a native speaker of a first language, and receiving a biasing term list including one or more terms in a second language different than the first language. The method also includes processing, using a speech recognition model, acoustic features derived from the audio data to generate speech recognition scores for both wordpieces and corresponding phoneme sequences in the first language. The method also includes rescoring the speech recognition scores for the phoneme sequences based on the one or more terms in the biasing term list, and executing, using the speech recognition scores for the wordpieces and the rescored speech recognition scores for the phoneme sequences, a decoding graph to generate a transcription for the utterance.
    Type: Grant
    Filed: April 28, 2020
    Date of Patent: March 8, 2022
    Assignee: Google LLC
    Inventors: Ke Hu, Antoine Jean Bruguier, Tara N. Sainath, Rohit Prakash Prabhavalkar, Golan Pundak
  • Publication number: 20210225369
    Abstract: A method of performing speech recognition using a two-pass deliberation architecture includes receiving a first-pass hypothesis and an encoded acoustic frame and encoding the first-pass hypothesis at a hypothesis encoder. The first-pass hypothesis is generated by a recurrent neural network (RNN) decoder model for the encoded acoustic frame. The method also includes generating, using a first attention mechanism attending to the encoded acoustic frame, a first context vector, and generating, using a second attention mechanism attending to the encoded first-pass hypothesis, a second context vector.
    Type: Application
    Filed: January 14, 2021
    Publication date: July 22, 2021
    Applicant: Google LLC
    Inventors: Ke Hu, Tara N. Sainath, Ruoming Pang, Rohit Prakash Prabhavalkar
  • Patent number: 10862249
    Abstract: An electrical connector is provided for a mating connector to be inserted thereto. The electrical connector includes an insulating body having a tongue, and multiple terminals. A middle grounding sheet is located between the terminals in upper and lower rows, and has at least one first through hole. Each of left and right sides of the middle grounding sheet has a latch slot and a notch. The insulating body has a first insulating post entering the first through hole and a second insulating post entering the notch. A dimension of the second insulating post is larger than a dimension of the first insulating post. A metal shell frames outside the insulating body to form an insertion space. A first clearance between a first protruding portion and a shielding casing of the mating connector is more than or equal to 0.001 mm and less than or equal to 0.0449 mm.
    Type: Grant
    Filed: August 8, 2019
    Date of Patent: December 8, 2020
    Assignee: LOTES CO., LTD
    Inventors: Ted Ju, Wu Feng, Ya Jun Zeng, Nan Fang He, Jun Fan, Jin Ke Hu, Guo Sheng Zhou, Chin Chi Lin
  • Publication number: 20200349923
    Abstract: A method includes receiving audio data encoding an utterance spoken by a native speaker of a first language, and receiving a biasing term list including one or more terms in a second language different than the first language. The method also includes processing, using a speech recognition model, acoustic features derived from the audio data to generate speech recognition scores for both wordpieces and corresponding phoneme sequences in the first language. The method also includes rescoring the speech recognition scores for the phoneme sequences based on the one or more terms in the biasing term list, and executing, using the speech recognition scores for the wordpieces and the rescored speech recognition scores for the phoneme sequences, a decoding graph to generate a transcription for the utterance.
    Type: Application
    Filed: April 28, 2020
    Publication date: November 5, 2020
    Applicant: Google LLC
    Inventors: Ke Hu, Antoine Jean Bruguier, Tara N. Sainath, Rohit Prakash Prabhavalkar, Golan Pundak
  • Patent number: 10642837
    Abstract: According to embodiments, a derived cache that is derived from a first instance of particular data is used to speed up queries and other operations over a second instance of the particular data. Traditionally, a DBMS generates and uses derived cache data only for the database data from which the derived data was derived. However, according to embodiments, derived cache data associated with a first instance of database data is relocated to the location of a second, newly created, instance of the database data. Since the derived cache data is derived from an identical copy of the database data, the cache data derived for the first instance can successfully be used to speed up applications running over the second instance of the database data.
    Type: Grant
    Filed: January 19, 2017
    Date of Patent: May 5, 2020
    Assignee: Oracle International Corporation
    Inventors: Kothanda Umamageswaran, Krishnan Meiyyappan, Adrian Tsz Him Ng, Vijay Sridharan, Wei Zhang, Ke Hu, Xin Zeng
  • Publication number: 20190363495
    Abstract: An electrical connector is provided for a mating connector to be inserted thereto. The electrical connector includes an insulating body having a tongue, and multiple terminals. A middle grounding sheet is located between the terminals in upper and lower rows, and has at least one first through hole. Each of left and right sides of the middle grounding sheet has a latch slot and a notch. The insulating body has a first insulating post entering the first through hole and a second insulating post entering the notch. A dimension of the second insulating post is larger than a dimension of the first insulating post. A metal shell frames outside the insulating body to form an insertion space. A first clearance between a first protruding portion and a shielding casing of the mating connector is more than or equal to 0.001 mm and less than or equal to 0.0449 mm.
    Type: Application
    Filed: August 8, 2019
    Publication date: November 28, 2019
    Inventors: Ted Ju, Wu Feng, Ya Jun Zeng, Nan Fang He, Jun Fan, Jin Ke Hu, Guo Sheng Zhou, Chin Chi Lin
  • Patent number: 10439332
    Abstract: An electrical connector includes an insulation body, at least one first terminal and at least one second terminal, and a middle grounding plate. The insulation body has a base and a tongue located in a front end of the base. The first terminal and the second terminal are arranged in the base in an upper row and a lower row, and exposed to an upper surface and a lower surface of the tongue. The middle grounding plate has a body portion fixed to the tongue, and located between the first terminal and the second terminal, the first terminal has a first front edge, the second terminal has a second front edge, and the first front edge, the second front edge and an edge in a front end of the body portion are located in a same vertical plane.
    Type: Grant
    Filed: July 13, 2017
    Date of Patent: October 8, 2019
    Assignee: LOTES CO., LTD
    Inventors: Ted Ju, Wu Feng, Ya Jun Zeng, Nan Fang He, Jun Fan, Jin Ke Hu, Guo Sheng Zhou, Chin Chi Lin
  • Patent number: 10381784
    Abstract: A mating connector includes a plastic body, which has a main body and an insertion portion formed by extending forward from the main body. A mating space is concavely formed backward from a front end surface of the insertion portion. The insertion portion has a side wall located at a periphery of the mating space and running through the mating space to form at least two through-holes, and the two through-holes do not communicate with each other. A metal cover surrounds the plastic body and is fastened to the main body. A grounding sheet is located between the metal cover and the side wall. The grounding sheet has a body and at least two elastic portions formed by bending forward from the body. The body covers the side wall, and each of the elastic portions is accommodated in a corresponding through-hole and enters the mating space.
    Type: Grant
    Filed: August 16, 2017
    Date of Patent: August 13, 2019
    Assignee: LOTES CO., LTD.
    Inventors: Ted Ju, Wu Feng, Ya Jun Zeng, Nan Fang He, Jun Fan, Jin Ke Hu, Guo Sheng Zhou, Chin Chi Lin
  • Patent number: 10320126
    Abstract: An electrical connector includes an insulation body. Multiple terminals are fixed on the insulation body in an upper row and a lower row. Each terminal has a soldering portion. The soldering portions are arranged in two rows in a front-rear direction. A middle shielding sheet is fixed on the insulation body, and has a planar plate portion. A baffle extends downward from a rear end of the plate portion and is located between the soldering portions in the two rows. The plate portion extends horizontally in a lateral direction to form a connecting portion, which is in front of the rear end of the plate portion. A rear edge of the connecting portion bends downward to form a leg. In the front-rear direction, a distance between the rear end of the plate portion and the rear edge of the connecting portion is greater than a width of the connecting portion.
    Type: Grant
    Filed: August 16, 2017
    Date of Patent: June 11, 2019
    Assignee: LOTES CO., LTD.
    Inventors: Ted Ju, Wu Feng, Ya Jun Zeng, Nan Fang He, Jun Fan, Jin Ke Hu, Guo Sheng Zhou, Chin Chi Lin
  • Patent number: 10320125
    Abstract: An electrical connector used for mating a mating connector includes an insulation body having a base portion and a tongue extending from the base portion, terminals fixedly disposed in the base portion and grouped into an upper row and a lower row, and a middle shielding sheet embedded in the tongue and located between the upper and lower rows. The tongue has upper and lower surfaces arranged opposite to each other, and a front surface connected to the upper and lower surfaces. The upper and lower rows of terminals are exposed on the upper and lower, respectively. The middle shielding sheet has a front end surface extending out of the front surface, a slot formed on the front end surface, and a first projection portion extending into the slot, overlapping a central line of the middle shielding sheet along a longitudinal direction, and located behind the front surface.
    Type: Grant
    Filed: June 29, 2017
    Date of Patent: June 11, 2019
    Assignee: LOTES CO., LTD.
    Inventors: Ted Ju, Wu Feng, Ya Jun Zeng, Nan Fang He, Jun Fan, Jin Ke Hu, Guo Sheng Zhou, Chin Chi Lin
  • Patent number: 10170867
    Abstract: An electrical connector includes an insulation body having a base and a tongue located in a front end of the base, terminals arranged in upper and lower rows, and a middle grounding plate disposed in the base and extending to the tongue. Terminals in the upper and lower rows are arranged on upper and lower surfaces of the tongue, respectively. The middle grounding plate is located between the upper and lower rows of terminals, and having at least one first hole. An edge of each of two lateral sides of the middle grounding plate is inwardly depressed to form a first groove. The first hole is located between the two first grooves. The insulation body has a first insulation post and a second insulation post, the first insulation post protrudes into the first hole, and the second insulation post protrudes into the first grooves on the lateral sides.
    Type: Grant
    Filed: July 13, 2017
    Date of Patent: January 1, 2019
    Assignee: LOTES CO., LTD
    Inventors: Ted Ju, Wu Feng, Ya Jun Zeng, Nan Fang He, Jun Fan, Jin Ke Hu, Guo Sheng Zhou, Chin Chi Lin
  • Publication number: 20180366850
    Abstract: A test electrical connector includes an insulating body, concavely provided with an insertion slot and multiple accommodating slots. The accommodating slots and an accommodating space are communicated with the insertion slot. Each accommodating slot has a channel penetrating downward through the insulating body. The channel has first and second surfaces. The first surface is obliquely provided facing the second surface. Multiple terminals are respectively accommodated in the accommodating slots. Each terminal has a connecting portion, and a contact portion and a pin respectively extend from the connecting portion. The contact portion is exposed to the insertion slot. The pin is accommodated in the channel, and has third and fourth surfaces. The third surface is obliquely provided facing the fourth surface. A gap is provided between the first surface and the third surface. An elastic body is accommodated in the accommodating space, and abuts the connecting portion.
    Type: Application
    Filed: June 14, 2018
    Publication date: December 20, 2018
    Inventors: Zuo Feng Jin, Jin Ke Hu