Fig. 8From: Single microphone speech separation by diffusion-based HMM estimationDFHMM, illustrative mask for a female-female mixture. Masking results for the female-female mixture using the DFHMM-H scheme. Sonograms of the clean speech of the first female, the second female and the corresponding mixture signal. The resulting hard mask is also depicted with white regions correspond to speaker 2 activeBack to article page