Jasper: An End-to-End Convolutional Neural Acoustic Model

Abstract

In this paper, we report state-of-the-art results on LibriSpeech among end-to-end speech recognition models without any external training data. Our model, Jasper, uses only 1D convolutions, batch normalization, ReLU, dropout, and residual connections. To improve training, we further introduce a new layer-wise optimizer called NovoGrad. Through experiments, we demonstrate that the proposed deep architecture performs as well or better than more complex choices. Our deepest Jasper variant uses 54 convolutional layers. With this architecture, we achieve 2.95% WER using a beam-search decoder with an external neural language model and 3.86% WER with a greedy decoder on LibriSpeech test-clean. We also report competitive results on the Wall Street Journal and the Hub5'00 conversational evaluation datasets.

arXiv commentAccepted to INTERSPEECH 2019

References in corpus (15)

• Decoupled Weight Decay Regularization
• SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognition
• Deep Speech 2: End-to-End Speech Recognition in English and Mandarin
• Conditional Image Generation with PixelCNN Decoders
• Densely Connected Convolutional Networks
• Mixed Precision Training
• Improved training of end-to-end attention models for speech recognition
• Wav2Letter: an End-to-End ConvNet-based Speech Recognition System
• Stochastic Gradient Methods with Layer-wise Adaptive Moments for Training of Deep Networks
• Fully Convolutional Speech Recognition
• The CAPIO 2017 Conversational Speech Recognition System
• Exploring Neural Transducers for End-to-End Speech Recognition
• Letter-Based Speech Recognition with Gated ConvNets
• Normalized Direction-preserving Adam
• A novel pyramidal-FSMN architecture with lattice-free MMI for speech recognition

Cited by in corpus (14)

• Conformer: Convolution-augmented Transformer for Speech Recognition
• Pushing the Limits of Semi-Supervised Learning for Automatic Speech Recognition
• NeMo: a toolkit for building AI applications using Neural Modules
• ContextNet: Improving Convolutional Neural Networks for Automatic Speech Recognition with Global Context
• Text-Conditioned Transformer for Automatic Pronunciation Error Detection
• QuartzNet: Deep Automatic Speech Recognition with 1D Time-Channel Separable Convolutions
• Efficient and Generic 1D Dilated Convolution Layer for Deep Learning
• Improved Conformer-based End-to-End Speech Recognition Using Neural Architecture Search
• Multimodal Target Speech Separation with Voice and Face References
• Improved Multi-Stage Training of Online Attention-based Encoder-Decoder Models
• Speech recognition for air traffic control via feature learning and end-to-end training
• Real World Robustness from Systematic Noise
• CarneliNet: Neural Mixture Model for Automatic Speech Recognition
• Frame-level SpecAugment for Deep Convolutional Neural Networks in Hybrid ASR Systems