20240038218. SPEECH MODEL PERSONALIZATION VIA AMBIENT CONTEXT HARVESTING simplified abstract (Intel Corporation)
Contents
SPEECH MODEL PERSONALIZATION VIA AMBIENT CONTEXT HARVESTING
Organization Name
Inventor(s)
Gabriel Amores of Sevilla (ES)
Guillermo Perez of Sevilla (ES)
Moshe Wasserblat of Maccabim (IL)
Michael Deisher of Hillsboro OR (US)
Loic Dufrensne De Virel of Hillsboro OR (US)
SPEECH MODEL PERSONALIZATION VIA AMBIENT CONTEXT HARVESTING - A simplified explanation of the abstract
This abstract first appeared for US patent application 20240038218 titled 'SPEECH MODEL PERSONALIZATION VIA AMBIENT CONTEXT HARVESTING
Simplified Explanation
The patent application describes an apparatus for speech modeling with personalization through ambient context harvesting. The apparatus consists of a microphone, context harvesting module, confidence module, and training module.
- The context harvesting module determines the context associated with the captured audio signals.
- The confidence module determines the confidence level of the context applied to the audio signals.
- The training module trains a neural network based on the confidence level being above a predetermined threshold.
Potential Applications:
- Speech recognition and transcription systems
- Personalized voice assistants
- Ambient context-aware applications
Problems Solved:
- Improving speech recognition accuracy by incorporating contextual information
- Enhancing personalization in voice-based applications
- Addressing the challenge of adapting speech models to different contexts
Benefits:
- Improved accuracy and reliability of speech recognition
- Enhanced user experience through personalized voice interactions
- Adaptability to different ambient contexts for better performance
Original Abstract Submitted
an apparatus for speech model with personalization via ambient context harvesting, is described herein. the apparatus includes a microphone, context harvesting module, confidence module, and training module. the context harvesting module is to determine a context associated with the captured audio signals. a confidence module is to determine a confidence of the context as applied to the audio signals. a training module is to train a neural network in response to the confidence being above a predetermined threshold.
