Recurrent Neural Networks for Noise Reduction in Robust ASR

By Andrew L. Maas, Quoc V. Le, Tyler M. O'Neil, Oriol Vinyals, Patrick Nguyen, Andrew Y. Ng. 2012

354 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Natural Language Processing, 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 Natural Language Processing

In Speech & Audio

In Signal Processing

Citations in the corpus

Speech recognition with deep recurrent neural networks (2013, background): “HMM-RNN systems [5] have also seen a recent revival [6, 7], but do not currently perform as well as deep networks.”
Increasing the Interpretability of Recurrent Neural Networks Using Hidden Markov Models (2016, methodology): “M L ] 1 6 Ju n 20 16 ically train the RNN to predict HMM states (Bourlard & Morgan, 1994) or posteriors (Maas et al., 2012), referred to as hybrid or tandem methods respectively.”
Deep Learning: Methods and Applications (2014, methodology): “Learning algorithms for the RNN have been dramatically improved since then, and much better results have been obtained recently using the RNN [48, 134, 235], especially when the bi-directional LSTM (long short-term memory) is used [135, 136].”
Exploring Speech Enhancement with Generative Adversarial Networks for Robust Speech Recognition (2017, methodology): “While deep learning has previously been applied to SFM for ASR [17, 18, 19], our work is the first to use GANs for this task.”
EESEN: End-to-end speech recognition using deep RNN models and WFST-based decoding (2015, methodology): “Acoustic modeling in Eesen is viewed as a sequence-to-sequence learning problem.”
A Hybrid DSP/Deep Learning Approach to Real-Time Full-Band Speech Enhancement (2017, methodology): “Deep learning techniques are already being used for noise suppression [5], [6], [7], [8], [9].”
A tutorial survey of architectures, algorithms, and applications for deep learning (2014, positive): “It achieves stability and fast convergence in training, helped by capping the growing gradient in training RNNs.”
Speech Enhancement in Multiple-Noise Conditions Using Deep Neural Networks (2016, background): “The success of deep neural networks (DNNs) in automatic speech recognition led to investigation of deep neural networks for noise suppression for ASR [13] and speech enhancement [14] [15] [16].”
A Primal-Dual Method for Training Recurrent Neural Networks Constrained by the Echo-State Property (2013, background): “, 2011), the way to solve the exploding gradient problem is to empirically clip the gradient if the norm of the gradient exceeds a certain threshold, and the way to avoid the vanishing gradient is also empirical — either to add a regularization term to push up the gradient or exploit the information”
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].”
An investigation of deep neural networks for noise robust speech recognition (2013, background): “Index Terms— noise robustness, deep neural network, adaptive training, Aurora 4”
Joint noise adaptive training for robust automatic speech recognition (2014, background): “But most such techniques assume Gaussian mixture model (GMM) based acoustic models (GMM-AMs); it is currently unclear how to extend them to DNN based systems.”
A deep auto-encoder based low-dimensional feature extraction from FFT spectral envelopes for statistical parametric speech synthesis (2016, methodology): “Deep auto-encoder based bottleneck features have been used by several groups for ASR [15, 16] and a deep denoising autoencoder has also verified for noise-robust ASR [17] or reverberant ASR tasks [18, 19].”
A Regression Approach to Speech Enhancement Based on Deep Neural Networks (2015, background): “However they often fail to track non-stationary noise for real-world scenarios in unexpected acoustic conditions.”
Denoising Gravitational Waves with Enhanced Deep Recurrent Denoising Auto-encoders (2017, methodology): “Apart from the timestep augmentation [2], we also add random shifts to the training data, to make the model more resilient to variations in the location of the signal.”
Speaker Adaptive Training of Deep Neural Network Acoustic Models Using I-Vectors (2015, methodology): “Also, in addition to DNNs, our framework can be applied readily to other types of acoustic models such as convolutional neural networks (CNNs) [22], [23], [24], [25] and recurrent neural networks (RNNs) [26], [27], [28], [29], [30].”
Deep Learning for Environmentally Robust Speech Recognition (2017, methodology): “In this research, data-driven approaches based on a supervised machine-learning paradigm have received increasing attention, and have emerged as viable methods for enhancing robustness of ASR systems (Maas et al. 2012).”
Large-Scale Domain Adaptation via Teacher-Student Learning (2017, background): “This improvement is much larger than what can be obtained from the traditional feature mapping [28][29] and mask learning [30][31] methods.”
Investigation of Speech Separation as a Front-End for Noise Robust Speech Recognition (2014, methodology): “When stereo data is available, feature mapping methods like SPLICE [7] and recurrent neural networks [25] have been used.”