EURASIP Journal on Advances in Signal Processing

Table 1 Amount of data for the Dev. and Eval. sets of SimData and RealData and for the training dataset¹

From: Environment-dependent denoising autoencoder for distant-talking speech recognition

	SimData		RealData		Training Data
	Dev.	Eval.	Dev.	Eval.
# of sentences	1484	2176	179	372	7861
	(∼ 3 h)	(∼ 4.8 h)	(∼ 0.3 h)	(∼ 0.6 h)	(∼ 17.5 h)
# of speakers	10	28	5	10	92

¹The clean and multi-condition training datasets are the same size

Back to article page