Deep neural network-based bottleneck feature and denoising autoencoder-based dereverberation for distant-talking speaker identification

EURASIP Journal on Audio, Speech, and Music Processing

Table 4 Reverberation methods

Reverberation methods
Conventional methods
1	CMN	MFCC with CMN
2	MCLMS-SS	Multichannel least mean squares
		with spectral subtraction
3	MSLP-SS	Multistep linear prediction
		with spectral subtraction
4	BF-MLP	Bottleneck feature extracted
		from multilayer perceptron
DNN-based feature transformation methods
5	BF-DNN	Bottleneck feature extracted
		from deep neural network
6	DAE	Denoising autoencoder-based
		cepstral-domain dereverberation
7	DAE + BF-DNN	Combination of DAE
		and BF-DNN