Invention Grant
- Patent Title: Full-sequence training of deep structures for speech recognition
- Patent Title (中): 用于语音识别的深层结构的全序训练
-
Application No.: US12886568Application Date: 2010-09-21
-
Publication No.: US09031844B2Publication Date: 2015-05-12
- Inventor: Dong Yu , Li Deng , Abdel-rahman Samir Abdel-rahman Mohamed
- Applicant: Dong Yu , Li Deng , Abdel-rahman Samir Abdel-rahman Mohamed
- Applicant Address: US WA Redmond
- Assignee: Microsoft Technology Licensing, LLC
- Current Assignee: Microsoft Technology Licensing, LLC
- Current Assignee Address: US WA Redmond
- Agent Sandy Swain; Judy Yee; Micky Minhas
- Main IPC: G10L15/14
- IPC: G10L15/14 ; G06K9/62 ; G06N3/04 ; G06N3/08

Abstract:
A method includes an act of causing a processor to access a deep-structured model retained in a computer-readable medium, the deep-structured model includes a plurality of layers with respective weights assigned to the plurality of layers, transition probabilities between states, and language model scores. The method further includes the act of jointly substantially optimizing the weights, the transition probabilities, and the language model scores of the deep-structured model using the optimization criterion based on a sequence rather than a set of unrelated frames.
Public/Granted literature
- US20120072215A1 FULL-SEQUENCE TRAINING OF DEEP STRUCTURES FOR SPEECH RECOGNITION Public/Granted day:2012-03-22
Information query