Investigations on byte-level convolutional neural networks for language modeling in low resource speech recognition

Kazuki Irie; Pavel Golik; Ralf Schlüter; Hermann Ney

ICASSP 2017accepted0 citations

Investigations on byte-level convolutional neural networks for language modeling in low resource speech recognition

Kazuki Irie, Pavel Golik, Ralf Schlüter, Hermann Ney

Abstract

In this paper, we present an investigation on technical details of the byte-level convolutional layer which replaces the conventional linear word projection layer in the neural language model. In particular, we discuss and compare the effective filter configurations, pooling types and the use of bytes instead of characters. We carry out experiments on language packs released by the IARPA Babel project and measure the performance in terms of perplexity and word error rate. Introducing a convolutional layer consistently improves the results on all languages. Also, there is no degradation from using raw bytes instead of proper Unicode characters, even on syllabic alphabets like Amharic. In addition, we report improvements in word error rate from rescoring lattices and evaluate keyword search performance on several languages.

BibTeX

@inproceedings{icassp2017_investigationson,
  title = {Investigations on byte-level convolutional neural networks for language modeling in low resource speech recognition},
  author = {Kazuki Irie and Pavel Golik and Ralf Schlüter and Hermann Ney},
  booktitle = {ICASSP 2017},
  year = {2017}
}