1.add data augmentor class for DeepSpeech2(inclued speed_change, resample, online bayesias_noline) 2.NosieAugmentor and ImpulseResaponseAugmentor will be added later