Advances In Very Deep Convolutional Neural Networks For LVCSR

Abstract

Very deep CNNs with small 3x3 kernels have recently been shown to achieve very strong performance as acoustic models in hybrid NN-HMM speech recognition systems. In this paper we investigate how to efficiently scale these models to larger datasets. Specifically, we address the design choice of pooling and padding along the time dimension which renders convolutional evaluation of sequences highly inefficient. We propose a new CNN design without timepadding and without timepooling, which is slightly suboptimal for accuracy, but has two significant advantages: it enables sequence training and deployment by allowing efficient convolutional evaluation of full utterances, and, it allows for batch normalization to be straightforwardly adopted to CNNs on sequence data. Through batch normalization, we recover the lost peformance from removing the time-pooling, while keeping the benefit of efficient convolutional evaluation. We demonstrate the performance of our models both on larger scale data

Stats

• citations0
• S2 citations—
• github stars0
• HF likes0
• heat score0.00
• arxiv keysercu2016advances

Related papers

• Deep Cnns Along The Time Axis With Intermap Pooling For Robustness To Spectral Variations (2016)6.77
• Very Deep Convolutional Neural Networks For Robust Speech Recognition (2016)12.40
• Analyzing Large Receptive Field Convolutional Networks For Distant Speech Recognition (2019)5.84
• Analyzing Deep Cnn-based Utterance Embeddings For Acoustic Model Adaptation (2018)6.77
• Long Short-term Memory Based Convolutional Recurrent Neural Networks For Large Vocabulary Speech Recognition (2016)6.77
• Linguistic Search Optimization For Deep Learning Based LVCSR (2018)0.00
• An Improved Deep Neural Network For Modeling Speaker Characteristics At Different Temporal Scales (2020)6.34
• Residual Convolutional CTC Networks For Automatic Speech Recognition (2017)0.00