Table 5 Preferred parameter settings for the implementation of the frequency-selective model-based speaker activity detection method
From: A dynamic multi-channel speech enhancement system for distributed microphones in a car environment
Parameter | Value |
|---|---|
γ μ | 0.83 |
γ σ | 0.8 |
Θ p | 0.01 |
Θ SNR4 | 0.25 |