Deep recurrent de-noising auto-encoder and blind de-reverberation for reverberated speech recognition

By Felix Weninger, Shinji Watanabe, Yuuki Tachioka, Björn Schuller. 2014

89 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Time-Frequency Masking in the Complex Domain for Speech Dereverberation and Denoising (2017, positive): “[40] use deep bidirectional Long ShortTerm Memory (LSTM) recurrent neural networks (RNNs) to dereverberate features that are inputed to an ASR system.”
Text-informed speech enhancement with deep neural networks (2015, background): “Note that other types of deep neural networks such as bidirectional recurrent neural networks [17] can be adopted to this framework.”
Speech Dereverberation Using Fully Convolutional Networks (2018, methodology): “[15] used bi-directional long short-term memory (bi-LSTM) recurrent neural networks”
Deep neural network based spectral feature mapping for robust speech recognition (2015, methodology): “In [15], spectral subtraction and a RNN based denoising auto-encoder are employed to learn Mel filterbank from noise.”
Investigating Generative Adversarial Networks based Speech Dereverberation for Robust Speech Recognition (2018, background): “, feedforward [13], recurrent [14] and convolutional [15], have been explored in the field.”
Dual autoencoders features for imbalance classification problem (2016, methodology): “On the other hand, both the autoencoder [26,27] and the denoising autoencoder [28] are used to learn features for speech recognitions.”
The MERL/MELCO/TUM system for the REVERB Challenge using Deep Recurrent Neural Network Feature Enhancement (2014, methodology): “In this study we use our spectral enhancement method based on deep neural networks introduced for de-reverberation in [13].”
Deep Learning for Environmentally Robust Speech Recognition (2017, background): “here the deepness comes from layers through time. Furthermore, similar to DBM and SAE, stacking RNN blocks 3 into a deep structure has attracted increased attention, leading to a Deep RNN (DRNN) [41]–[43]. Recently, the Gated Recurrent Unit (GRU) has also emerged as a computationally simpler alterna”