Jian K Wu

US Patent:

7693713, Apr 6, 2010

Filed:

Jun 17, 2005

Appl. No.:

11/156106

Inventors:

Xiaodong He - Issaquah WA, US
Jian Wu - Redmond WA, US

Assignee:

Microsoft Corporation - Redmond WA

International Classification:

G10L 15/06

US Classification:

704243, 704245

Abstract:

Speech models are trained using one or more of three different training systems. They include competitive training which reduces a distance between a recognized result and a true result, data boosting which divides and weights training data, and asymmetric training which trains different model components differently.

US Patent:

8532991, Sep 10, 2013

Filed:

Mar 10, 2010

Appl. No.:

12/720968

Inventors:

Xiaodong He - Issaquah WA, US
Jian Wu - Redmond WA, US

Assignee:

Microsoft Corporation - Redmond WA

International Classification:

G10L 15/06
G10L 15/00

US Classification:

704243, 704255

Abstract:

Speech models are trained using one or more of three different training systems. They include competitive training which reduces a distance between a recognized result and a true result, data boosting which divides and weights training data, and asymmetric training which trains different model components differently.

US Patent:

20090325154, Dec 31, 2009

Filed:

Jun 20, 2006

Appl. No.:

11/922385

Inventors:

Jingyue Ju - Englewood Cliffs NJ, US
Jian Wu - New York NY, US
Dae H. Kim - New York NY, US

Assignee:

The Trustees of Columbia University in the City of New York - New York NY

International Classification:

C12Q 1/68
C07H 21/04

US Classification:

435 6, 536 221

Abstract:

This invention provides methods for pyrosequencing and compositions comprising 3′-O— modified deoxynucleoside triphosphates.

US Patent:

62967371, Oct 2, 2001

Filed:

Jun 12, 2000

Appl. No.:

9/592358

Inventors:

Jian Wu - Seattle WA
Hugh West - Seattle WA
Terry M. Grant - Auburn WA

Assignee:

Weyerhaeuser Company - Federal Way WA

International Classification:

D21H 1733
D21H 1763
D21H 1768
D21H 1769
D21H 1755

US Classification:

1621641

Abstract:

The invention relates to cellulose fluff pulp products that are debondable into fluff with markedly lower energy input, to a process for making the products, and to absorbent products using the fluff Most of the pulp products show no reduction in liquid absorbency rate from that of untreated fiber and significantly higher rates than pulps treated with the usual debonding agents. The products are made by adhering fine non-cellulosic particles to the fiber surfaces using a retention aid. The fiber is preferably treated with the retention aid in an aqueous suspension for a sufficient time so that the retention aid is substantively bonded with little or none left free in the water. The fine particulate additive is then added and becomes attached and uniformly distributed over the fiber surfaces with very little particle agglomeration occurring. The fiber is most usually not refined or only very lightly refined before sheeting. However it may be significantly refined to produce a product having a very high surface area.

US Patent:

6074524, Jun 13, 2000

Filed:

Aug 29, 1997

Appl. No.:

8/920689

Inventors:

Jian Wu - Seattle WA
Hugh West - Seattle WA
Terry M. Grant - Auburn WA

Assignee:

Weyerhaeuser Company - Tacoma WA

International Classification:

D21H 1763
D21H 1767
D21H 1768

US Classification:

162100

Abstract:

The invention relates to cellulose fluff pulp products that are debondable into fluff with markedly lower energy input, to a process for making the products, and to absorbent products using the fluff. Most of the pulp products show no reduction in liquid absorbency rate from that of untreated fiber and significantly higher rates than pulps treated with the usual debonding agents. The products are made by adhering fine non-cellulosic particles to the fiber surfaces using a retention aid. The fiber is preferably treated with the retention aid in an aqueous suspension for a sufficient time so that the retention aid is substantively bonded with little or none left free in the water. The fine particulate additive is then added and becomes attached and uniformly distributed over the fiber surfaces with very little particle agglomeration occurring. The fiber is most usually not refined or only very lightly refined before sheeting. However, it may be significantly refined to produce a product having a very high surface area.

US Patent:

20220189467, Jun 16, 2022

Filed:

Dec 15, 2020

Appl. No.:

17/123087

Inventors:

- Redmond WA, US
Emilian STOIMENOV - Bellevue WA, US
Christopher Hakan BASOGLU - Everett WA, US
Kshitiz KUMAR - Redmond WA, US
Jian WU - Bellevue WA, US

International Classification:

G10L 15/16
G10L 25/51
G10L 19/16

Abstract:

Disclosed speech recognition techniques improve user-perceived latency while maintaining accuracy by: receiving an audio stream, in parallel, by a primary (e.g., accurate) speech recognition engine (SRE) and a secondary (e.g., fast) SRE; generating, with the primary SRE, a primary result; generating, with the secondary SRE, a secondary result; appending the secondary result to a word list; and merging the primary result into the secondary result in the word list. Combining output from the primary and secondary SREs into a single decoder as described herein improves user-perceived latency while maintaining or improving accuracy, among other advantages.

US Patent:

20230102295, Mar 30, 2023

Filed:

Dec 1, 2022

Appl. No.:

18/060949

Inventors:

- Redmond WA, US
Emilian STOIMENOV - Bellevue WA, US
Christopher Hakan BASOGLU - Everett WA, US
Kshitiz KUMAR - Redmond WA, US
Jian WU - Bellevue WA, US

International Classification:

G10L 15/30
G10L 15/16
G10L 19/16
G10L 25/51

Abstract:

Disclosed speech recognition techniques improve user-perceived latency while maintaining accuracy by: receiving an audio stream, in parallel, by a primary (e.g., accurate) speech recognition engine (SRE) and a secondary (e.g., fast) SRE; generating, with the primary SRE, a primary result; generating, with the secondary SRE, a secondary result; appending the secondary result to a word list; and merging the primary result into the secondary result in the word list. Combining output from the primary and secondary SREs into a single decoder as described herein improves user-perceived latency while maintaining or improving accuracy, among other advantages.

US Patent:

20210082438, Mar 18, 2021

Filed:

Nov 13, 2019

Appl. No.:

16/682921

Inventors:

- Redmond WA, US
Tianyan ZHOU - Bellevue WA, US
Jinyu LI - Redmond WA, US
Yifan GONG - Sammamish WA, US
Jian WU - Bellevue WA, US
Zhuo CHEN - Woodinville WA, US

International Classification:

G10L 17/18
G10L 17/02
G06N 3/08

Abstract:

Embodiments may include reception of a plurality of speech frames, determination of a multi-dimensional acoustic feature associated with each of the plurality of speech frames, determination of a plurality of multi-dimensional phonetic features, each of the plurality of multi-dimensional phonetic features determined based on a respective one of the plurality of speech frames, generation of a plurality of two-dimensional feature maps based on the phonetic features, input of the feature maps and the plurality of acoustic features to a convolutional neural network, the convolutional neural network to generate a plurality of speaker embeddings based on the plurality of feature maps and the plurality of acoustic features, aggregation of the plurality of speaker embeddings into a first speaker embedding based on respective weights determined for each of the plurality of speaker embeddings, and determination of a speaker associated with the plurality of speech frames based on the first speaker embedding.