Speaker-Independent Microphone Identification in Noisy Conditions

TL;DR

This paper introduces a neural-network-based denoising approach to improve microphone identification accuracy in noisy speech recordings, demonstrating significant performance gains and robustness against noise and counter-forensics attacks.

Contribution

It presents a novel framework combining denoising with microphone classification, enhancing identification accuracy in noisy conditions and validating the effectiveness of denoising as a preprocessing step.

Findings

Denoising significantly improves microphone classification accuracy in noisy environments.

The method enhances the discriminating power of state-of-the-art features.

Denoising prior to classification is validated as effective for noisy recordings.

Abstract

This work proposes a method for source device identification from speech recordings that applies neural-network-based denoising, to mitigate the impact of counter-forensics attacks using noise injection. The method is evaluated by comparing the impact of denoising on three state-of-the-art features for microphone classification, determining their discriminating power with and without denoising being applied. The proposed framework achieves a significant performance increase for noisy material, and more generally, validates the usefulness of applying denoising prior to device identification for noisy recordings.