18485069. Neural Networks For Speaker Verification simplified abstract (GOOGLE LLC)
Contents
Neural Networks For Speaker Verification
Organization Name
Inventor(s)
Georg Heigold of Mountain View CA (US)
Samuel Bengio of Los Altos CA (US)
Ignacio Lopez Moreno of New York NY (US)
Neural Networks For Speaker Verification - A simplified explanation of the abstract
This abstract first appeared for US patent application 18485069 titled 'Neural Networks For Speaker Verification
Simplified Explanation
The patent application describes systems and methods for speaker verification using neural networks. Here are the key points:
- The method involves training a neural network for a speaker verification model.
- Users are enrolled at a client device, and their identities are verified based on characteristics of their voices.
- The method includes receiving data that characterizes an utterance of a user.
- A speaker representation is generated for the utterance using a neural network on a computing device.
- The neural network is trained based on a plurality of training samples that include data characterizing utterances labeled as matching or non-matching speakers.
Potential applications of this technology:
- Speaker verification for secure access control systems.
- Voice-based authentication for mobile devices and online platforms.
- Fraud detection in call centers and financial transactions.
- Personalized voice assistants that can identify individual users.
Problems solved by this technology:
- Enhances security by verifying the identity of users based on their voices.
- Reduces the risk of unauthorized access to sensitive information.
- Provides a convenient and efficient method for user authentication.
Benefits of this technology:
- Improved accuracy in speaker verification compared to traditional methods.
- Scalable and adaptable to different languages and dialects.
- Can be integrated into existing systems and devices.
- Reduces the need for passwords or PINs, enhancing user experience.
Original Abstract Submitted
This document generally describes systems, methods, devices, and other techniques related to speaker verification, including (i) training a neural network for a speaker verification model, (ii) enrolling users at a client device, and (iii) verifying identities of users based on characteristics of the users' voices. Some implementations include a computer-implemented method. The method can include receiving, at a computing device, data that characterizes an utterance of a user of the computing device. A speaker representation can be generated, at the computing device, for the utterance using a neural network on the computing device. The neural network can be trained based on a plurality of training samples that each: (i) include data that characterizes a first utterance and data that characterizes one or more second utterances, and (ii) are labeled as a matching speakers sample or a non-matching speakers sample.
