Hybrid speech recognition with Deep Bidirectional LSTM

By Alex Graves, Navdeep Jaitly, Abdel-rahman Mohamed. 2013

1,724 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Natural Language Processing, 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 Signal Processing

Citations in the corpus

Neural Machine Translation by Jointly Learning to Align and Translate (2014, methodology): “A similar approach of aligning an output symbol with an input symbol was proposed recently by Graves (2013) in the context of handwriting synthesis.”
Unifying Visual-Semantic Embeddings with Multimodal Neural Language Models (2014, methodology): “ding, writing and reseting information. LSTMs have been used to achieve state-of-the-art performance in several tasks such as handwriting recognition [31], sequence generation [32] speech recognition [33] and machine translation [11] among others. Dropout [34] strategies have also been proposed to p”
RWTH ASR Systems for LibriSpeech: Hybrid vs Attention - w/o Data Augmentation (2019, background): “The NN acoustic model architecture is a bi-directional LSTM [11,12].”
Transformers with convolutional context for ASR (2019, neutral): “Early CTC experiments showed no gains for WSJ [19] while it was later proved to be one of the current best large scale loss functions [6].”
Light Gated Recurrent Units for Speech Recognition (2018, methodology): “These recurrent layers were bidirectional RNNs [35], which were obtained by concatenating the forward hidden states (collected by processing the sequence from the beginning to the end) with backward hidden states (gathered by scanning the speech in the reverse time order).”
Improved Semantic Representations From Tree-Structured Long Short-Term Memory Networks (2015, methodology): “, 2014), speech recognition (Graves et al., 2013), image caption generation (Vinyals et al.”
Generating Images from Captions with Attention (2015, background): “Deep Neural Networks have achieved signiﬁcant success in various tasks such as image recognition (Krizhevsky et al., 2012), speech transcription (Graves et al., 2013), and machine translation (Bah-danau et al., 2015).”
NTU RGB+D 120: A Large-Scale Benchmark for 3D Human Activity Understanding (2019, methodology): “To produce a more discriminative set of features for each part, we introduce bidirectional context passing (similar to bidirectional LSTM [95]) for our feature generation network, as illustrated in Fig.”
Residual LSTM: Design of a Deep Recurrent Architecture for Distant Speech Recognition (2017, methodology): “Recurrent neural networks (RNN) were successfully applied to learn long term dependencies of sequential data [11, 12, 13, 14].”
Forecasting Economics and Financial Time Series: ARIMA vs. LSTM (2018, methodology): “…Eyben 2009; Graves et al. 2013; Sak et al. 2014], time-series prediction [Hochreiter and Schmidhuber, 1997; Gers, 2000; Yim, 2002; Zhang, 2005; Graves et al., 2009, 2013; Schmidhuber, 2015; Brownlee, 2016; Gamboa, 2017; Roondiwala, 2017] as well as its applications in economics and finance…”
Deep Learning: Methods and Applications (2014, methodology): “The effectiveness of this method has been demonstrated in handwriting recognition tasks and in a small speech task [135, 136] to be discussed in more detail in Section 7 of this monograph.”
Baseline Needs More Love: On Simple Word-Embedding-Based Models and Associated Pooling Mechanisms (2018, methodology): “We omit the details of the LSTM and refer the interested readers to the work by Graves et al. (2013) for further explanation.”
Action Recognition using Visual Attention (2015, methodology): “LSTMs have been recently shown to perform well in the domain of speech recognition (Graves et al., 2013), machine translation (Sutskever et al., 2014), image description (Xu et al., 2015; Vinyals et al., 2015) and video description (Yao et al., 2015; Venugopalan et al., 2014).”
Unsupervised Learning of Disentangled and Interpretable Representations from Sequential Data (2017, methodology): “(2) train an LSTM acoustic model [12, 35, 42] on the set that only covers partial domains using mean and log variance of the latent segment variable z1 extracted from the trained Seq2Seq-FHVAE.”
A Simple Way to Initialize Recurrent Networks of Rectified Linear Units (2015, methodology): “We performed Phoneme recognition experiments on TIMIT with IRNNs and Bidirectional IRNNs and compared them to RNNs, LSTMs and Bidirectional LSTMs and RNNs. Bidirectional LSTMs have been applied previously to TIMIT in [11].”
Document Modeling with Gated Recurrent Neural Network for Sentiment Classification (2015, methodology): “The approach is analogous to the recently emerged LSTM (Graves et al., 2013; Zaremba and Sutskever, 2014; Sutskever et al., 2014; Xu et al., 2015) and gated neural network (Cho et al.”
End-to-end Continuous Speech Recognition using Attention-based Recurrent NN: First Results (2014, methodology): “Graves et al. (2013b), on the other hand, used a separate multi-layer perceptron (MLP) to combine the two scores from the transcription and prediction networks and achieved the state-of-the-art performance on the TIMIT dataset.”