Gradient-distributed metal-halide dynamic memristors for adaptive and robust voiceprint recognition

Area of Science:

• Materials Science
• Electrical Engineering
• Signal Processing

Background:

• Voiceprint recognition is crucial for security but struggles with speech variability and noise.
• Current systems face limitations in accurate feature extraction in challenging acoustic environments.

Purpose of the Study:

• To develop an adaptive voiceprint recognition system using a novel memristor.
• To overcome limitations of speech frequency and amplitude variability in noisy conditions.

Main Methods:

• Engineered a large-scale hybrid metal-halide dynamic memristor (MHDM) with a gradient-distributed architecture.
• Utilized the memristor's functional layer to modulate Schottky barriers and interface charges.
• Evaluated the memristor's response time, noise tolerance, and signal processing capabilities.

Main Results:

• Achieved µs-scale response and kHz-scale dynamic signal processing.
• Demonstrated over 20% improvement in signal-to-noise ratio, enhancing noise tolerance.
• Attained 99.3% voiceprint recognition accuracy, maintaining 93.2% in realistic background noise.

Conclusions:

• The developed MHDM offers a scalable solution for secure and efficient voiceprint recognition.
• The gradient architecture provides robust performance in noisy environments.
• This technology shows significant potential for advanced identity authentication systems.