Revisiting Recurrent Neural Networks for robust ASR

By Oriol Vinyals, Suman V. Ravuri, Daniel Povey. 2012

169 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Model Architectures, Speech & Audio, Signal Processing

How researchers see this paper

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

In Artificial Intelligence

In Model Architectures

In Speech & Audio

In Signal Processing

Citations in the corpus

Long-term recurrent convolutional networks for visual recognition and description (2014, background): “Though RNNs have proven successful on tasks such as speech recognition [43] and text generation [38], it can be difficult to train them to learn long-term dynamics, likely due in part to the vanishing and exploding gradients problem [12] that can result from propagating the gradients down through th”
Recurrent Neural Networks for Noise Reduction in Robust ASR (2012, methodology): “The tandem network trained on clean and noisy data outperforms a tandem approach trained on clean data only as reported in [10].”
Long-Term Recurrent Convolutional Networks for Visual Recognition and Description (2017, background): “Though RNNs have proven successful on tasks such as speech recognition [15] and text generation [16], it can be difficult to train them to learn long-term dynamics, likely due in part to the vanishing and exploding gradients problem [7] that can result from propagating the gradients down through…”
Dynamic Recursive Neural Network (2019, unknown): “Different from the way of shar ing weights along the sequence in Recurrent Neural Net works (RNN) [40], recursive network shares weights at ev ery node, which could be considered as a generalization of RNN. [7] tries recursive layers on image recognition but gets worse performance than a single…”
Towards End-to-End Speech Recognition with Deep Convolutional Neural Networks (2016, methodology): “Notice that the CNN model could take much less time to train in comparison with the LSTM model when keeping roughly the same number of parameters.”
Performance Evaluation of Deep Neural Networks Applied to Speech Recognition: RNN, LSTM and GRU (2019, methodology): “HMM-RNN frameworks were used in the past [18, 19], however, they do not perform well with DNN. Graves et al. [8] came up with a very eﬃcient Connectionist Temporal Classiﬁcation (CTC) method of sequence labeling to train RNNs end-to-end.”
Recurrent deep neural networks for robust speech recognition (2014, background): “Meanwhile, recurrent neural networks (RNNs) have been also explored for robust ASR in [20, 21, 22, 23].”
Modeling temporal dependencies in data using a DBN-LSTM (2015, background): “However, these models do not perform as well as deep networks ([3]).”
A Spectral Masking Approach to Noise-Robust Speech Recognition Using Deep Neural Networks (2014, methodology): “This has also been observed for DNN-based ASR systems in [4]–[6].”
Robust speech recognition using long short-term memory recurrent neural networks for hybrid acoustic modelling (2014, methodology): “and using the predictions as additional input features for an HMM: in [11], RNNs were compared to other NN architectures.”
Deep Neural Networks for Single-Channel Multi-Talker Speech Recognition (2015, background): “Our best system achieves 18.8% overall word error rate (WER), which is 2.8% absolute better than that obtained with the state-of-the-art IBM system with fewer assumptions.”
Deep neural network based spectral feature mapping for robust speech recognition (2015, methodology): “Recurrent neural networks (RNNs) have been applied in a tandem system [8] or in a hybrid system with deep structure [9] to model temporal dependency of speech and noise for robust ASR.”
A Primal-Dual Method for Training Recurrent Neural Networks Constrained by the Echo-State Property (2013, background): “Since deep learning became popular in recent years, much more research has been devoted to the RNN (Graves et al., 2006; Graves, 2012; Maas et al., 2012; Mikolov, 2012; Vinyals et al., 2012; Graves et al., 2013).”
An Overview of Noise-Robust Automatic Speech Recognition (2014, methodology): “In addition to using the standard feed-forward structure of DNNs, recurrent neural networks (RNN) that model temporal signal dependence in an explicit way have also been exploited for noise-robust ASR, either in modeling the posterior [79] or predicting clean speech from noisy speech [80].”