A Comparative Study on Transformer vs RNN in Speech Applications
This work provides empirical evidence for the superiority of Transformer over RNN in various speech processing tasks, aiding researchers and practitioners in model selection.
This paper compared Transformer and RNN models across 15 ASR, one multilingual ASR, one ST, and two TTS benchmarks, finding that Transformer outperformed RNN in 13 out of 15 ASR tasks and showed benefits in other speech applications.
Sequence-to-sequence models have been widely used in end-to-end speech processing, for example, automatic speech recognition (ASR), speech translation (ST), and text-to-speech (TTS). This paper focuses on an emergent sequence-to-sequence model called Transformer, which achieves state-of-the-art performance in neural machine translation and other natural language processing applications. We undertook intensive studies in which we experimentally compared and analyzed Transformer and conventional recurrent neural networks (RNN) in a total of 15 ASR, one multilingual ASR, one ST, and two TTS benchmarks. Our experiments revealed various training tips and significant performance benefits obtained with Transformer for each task including the surprising superiority of Transformer in 13/15 ASR benchmarks in comparison with RNN. We are preparing to release Kaldi-style reproducible recipes using open source and publicly available datasets for all the ASR, ST, and TTS tasks for the community to succeed our exciting outcomes.