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

TL;DR

Jasper is a deep convolutional neural network for end-to-end speech recognition that achieves state-of-the-art results on LibriSpeech without external data, using only convolutional layers, residual connections, and a new optimizer.

Contribution

The paper introduces Jasper, a deep convolutional architecture with 54 layers and a novel optimizer NovoGrad, achieving top performance on speech recognition benchmarks.

Findings

Achieves 2.95% WER on LibriSpeech test-clean with external language model.

Uses only 1D convolutions, batch normalization, ReLU, dropout, residuals.

Deep architecture performs as well or better than more complex models.

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.