Spot keywords from very noisy and mixed speech
This addresses a challenging keyword spotting task for noisy and mixed speech conditions, though it appears incremental as it builds on existing architectures like EfficientNet.
The paper tackled the problem of detecting keywords in speech with strong interfering speech and mixed keywords, achieving high effectiveness with a novel Mix Training strategy that outperformed standard data augmentation and mixup training.
Most existing keyword spotting research focuses on conditions with slight or moderate noise. In this paper, we try to tackle a more challenging task: detecting keywords buried under strong interfering speech (10 times higher than the keyword in amplitude), and even worse, mixed with other keywords. We propose a novel Mix Training (MT) strategy that encourages the model to discover low-energy keywords from noisy and mixed speech. Experiments were conducted with a vanilla CNN and two EfficientNet (B0/B2) architectures. The results evaluated with the Google Speech Command dataset demonstrated that the proposed mix training approach is highly effective and outperforms standard data augmentation and mixup training.