Improving Speech Recognition by Revising Gated Recurrent Units
By Mirco Ravanelli, Philemon Brakel, Maurizio Omologo, Yoshua Bengio. 2017
56 citations recorded by the source corpus.
Speech recognition is largely taking advantage of deep learning, showing that substantial benefits can be obtained by modern Recurrent Neural Networks (RNNs). The most popular RNNs are Long Short-Term Memory (LSTMs), which typically reach state-of-the-art performance in many tasks thanks to their ability to learn long-term dependencies and robustness to vanishing gradients. Nevertheless, LSTMs have a rather complex design with three multiplicative gates, that might impair their efficient implementation. An attempt to simplify LSTMs has recently led to Gated Recurrent Units (GRUs), which are based on just two multiplicative gates. This paper builds on these efforts by further revising GRUs and proposing a simplified architecture potentially more suitable for speech recognition. The contribution of this work is two-fold. First, we suggest to remove the reset gate in the GRU design, resulting in a more efficient single-gate architecture. Second, we propose to replace tanh with ReLU activations in the state update equations. Results show that, in our implementation, the revised architecture reduces the per-epoch training time with more than 30% and consistently improves recognition performance across different tasks, input features, and noisy conditions when compared to a standard GRU.
Subfields: Artificial Intelligence, Speech & Audio
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
- Yoshua Bengio
- Mirco Ravanelli
- Philemon Brakel
- Titouan Parcollet
- Dmitriy Serdyuk
- Shinji Watanabe
- Maurizio Omologo
- Quoc V. Le
In Speech & Audio
- Yoshua Bengio
- Mirco Ravanelli
- Titouan Parcollet
- Philemon Brakel
- Maurizio Omologo
- Dmitriy Serdyuk
- Santiago Pascual
- Jianyuan Zhong
Citations in the corpus
- Multi-Task Self-Supervised Learning for Robust Speech Recognition (2020, positive): “The QRNN provides a performance similar to that of more traditional LSTM or GRU models with lower computational load [16].”
- Light Gated Recurrent Units for Speech Recognition (2018, background): “The potential benefits Li-GRUs have been preliminarily observed as part of a work on speech recognition described in [52].”
- Interpretable Convolutional Filters with SincNet (2018, methodology): “Multiple standard convolutional, fully-connected or recurrent layers [15, 16, 17, 45] can then be stacked together to finally perform a classification with a softmax classifier.”
- Speaker Recognition from Raw Waveform with SincNet (2018, methodology): “Multiple standard convolutional, fully-connected or recurrent layers [37–40] can then be stacked together to finally perform a speaker classification with a softmax classifier.”
- Quaternion Convolutional Neural Networks for End-to-End Automatic Speech Recognition (2018, background): “Other recent results with real-valued vectors of similar features are reported in [29, 4, 30, 31].”
- Twin Regularization for online speech recognition (2018, methodology): “Beyond standard LSTM [36] and GRU [37], we also considered recently proposed architectural variations [38]: MGRU [39] is the minimal GRU architecture based on replacing reset gate with updated gate activations, while light GRU (LiGRU) [40] directly avoids the reset gate and exploits ReLU activations”
- The Pytorch-kaldi Speech Recognition Toolkit (2018, methodology): “Batch normalization was adopted for feed-forward connections only, as proposed in [41].”