Single Input Voice Authentication via Text-First Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In multi-user systems, traditional voice authentication methods require multiple voice inputs, leading to time-consuming and resource-intensive processes, especially when comparing voice inputs against numerous user profiles, which negatively affects user authentication performance.
Innovation Solution
A single input voice authentication system combines speech-to-text, natural language understanding, and text-independent voice authentication to recognize and authenticate users with a single voice input, converting voice audio to text, determining login intent, extracting user IDs, and matching voice inputs against stored voiceprints, effectively transforming a multi-class model into a binary-class model for efficient authentication.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional voice authentication compares voice input with multiple stored voiceprints in multi-user systems, then user identification accuracy is maintained, but authentication time increases and user efficiency deteriorates
Solution Approach 1:
The authentication process is segmented into two distinct stages: first, text-based user identification to locate the specific user profile, and second, voiceprint analysis to verify identity. This segmentation allows the system to handle multiple users efficiently by avoiding comprehensive voiceprint comparison across all users, thus reducing authentication time while maintaining accuracy.
Solution Approach 2:
The system performs preliminary text conversion and user identification before conducting voiceprint analysis. By converting speech to text and identifying the user profile in advance, the system narrows down the comparison scope to a single pre-identified voiceprint, significantly reducing processing time compared to comparing against multiple voiceprints simultaneously.
2Reliability
If multiple voice inputs are required for authentication in multi-user systems, then security is maintained through multiple verification steps, but processing resources are consumed and user efficiency decreases
Solution Approach 1:
The system extracts and utilizes the textual information contained within the voice input to identify the user profile before performing voiceprint verification. This extraction of text-based identification capability from the voice input allows the system to avoid comparing against multiple voiceprints, reducing processing resource consumption while maintaining security through the two-stage verification process.
3Measurement precision
If comprehensive voiceprint comparison is performed against multiple user profiles, then authentication accuracy is ensured, but system scalability is limited due to increased processing complexity
Solution Approach 1:
The authentication process is divided into two independent modules: text-based user identification and voiceprint verification. This segmentation allows each module to operate independently with optimized complexity - the text identification handles user profile location efficiently, while the voiceprint module performs simple binary matching against a single profile, enabling system scalability without compromising authentication accuracy.
Data Source
AI summary
A system and method for authenticating a user via a single voice audio input is disclosed. The method includes obtaining voice audio from a user, converting the voice audio to text, comparing at least a portion of the text to a database, determining whether a user profile exists in the database based on the comparing, the user profile including a voiceprint, in response to determining that the user profile exists in the database, analyzing the voice audio against the voiceprint of the user profile; and authenticating the user based on the voice audio substantially matching the voiceprint of the user profile.