Generating and using text-to-speech data for speech recognition models

US12205596B2 · kind B2 · utility

Assignee

• MICROSOFT TECHNOLOGY LICENSING, LLC · US

Inventors

• Guoli Ye · Sammamish, US
• Yan Huang · Hsinchu, TW
• Wenning Wei · Beijing, CN
• Lei He · Moraga, US
• Eva Sharma · Vancouver, CA
• Jian Wu · Bellevue, US
• Yao Tian · Huzhou City, CN
• Edward C. Lin · Beijing, CN
• Yifan Gong · Sammamish, US
• Rui Zhao · Beijing, CN
• Jinyu Li · Beijing, CN
• William Maxwell Gale · Sunnyvale, US

Key dates

• Filing date: Feb 10, 2023
• Grant date: Jan 21, 2025
• Expiry date: Jul 13, 2043

Classification

• Technology area (CPC G): Physics
• CPC primary: G10L2015/088
• WIPO field: Computer technology
• WIPO sector: Electrical engineering

Abstract

Systems, methods, and devices are provided for generating and using text-to-speech (TTS) data for improved speech recognition models. A main model is trained with keyword independent baseline training data. In some instances, acoustic and language model sub-components of the main model are modified with new TTS training data. In some instances, the new TTS training is obtained from a multi-speaker neural TTS system for a keyword that is underrepresented in the baseline training data. In some instances, the new TTS training data is used for pronunciation learning and normalization of keyword dependent confidence scores in keyword spotting (KWS) applications. In some instances, the new TTS training data is used for rapid speaker adaptation in speech recognition models.