Whereas conventional spoken language understanding (SLU) systems map speech to text, and then text to intent, end-to-end SLU systems map speech directly to intent through a single trainable model. Achieving high accuracy with these end-to-end models without a large amount of training data is difficult. We propose a method to reduce the data requirements of end-to-end SLU in which the model is first pre-trained to predict words and phonemes, thus learning good features for SLU. We introduce a new SLU dataset, Fluent Speech Commands, and show that our method improves performance both when the full dataset is used for training and when only a small subset is used. We also describe preliminary experiments to gauge the model's ability to generalize to new phrases not heard during training.
@article{arxiv.1904.03670,
title = {Speech Model Pre-training for End-to-End Spoken Language Understanding},
author = {Loren Lugosch and Mirco Ravanelli and Patrick Ignoto and Vikrant Singh Tomar and Yoshua Bengio},
journal= {arXiv preprint arXiv:1904.03670},
year = {2019}
}