Direct Acoustics-to-Word Models for English Conversational Speech Recognition

TL;DR

This paper demonstrates direct acoustics-to-word CTC models for English conversational speech recognition, achieving competitive accuracy without the need for an external language model or decoder, and discusses techniques to address data requirements.

Contribution

First implementation of direct acoustics-to-word CTC models on benchmark tasks, eliminating the need for separate language models and decoders in speech recognition.

Findings

Achieved 13.0% WER on Switchboard without LM or decoder.

Presented techniques to mitigate large data requirements for word models.

Compared performance of word and phone CTC models.

Abstract

Recent work on end-to-end automatic speech recognition (ASR) has shown that the connectionist temporal classification (CTC) loss can be used to convert acoustics to phone or character sequences. Such systems are used with a dictionary and separately-trained Language Model (LM) to produce word sequences. However, they are not truly end-to-end in the sense of mapping acoustics directly to words without an intermediate phone representation. In this paper, we present the first results employing direct acoustics-to-word CTC models on two well-known public benchmark tasks: Switchboard and CallHome. These models do not require an LM or even a decoder at run-time and hence recognize speech with minimal complexity. However, due to the large number of word output units, CTC word models require orders of magnitude more data to train reliably compared to traditional systems. We present some techniques to mitigate this issue. Our CTC word model achieves a word error rate of 13.0%/18.8% on the Hub5-2000 Switchboard/CallHome test sets without any LM or decoder compared with 9.6%/16.0% for phone-based CTC with a 4-gram LM. We also present rescoring results on CTC word model lattices to quantify the performance benefits of a LM, and contrast the performance of word and phone CTC models.