Wav2Seq: Pre-training Speech-to-Text Encoder-Decoder Models Using Pseudo Languages

TL;DR

Wav2Seq introduces a self-supervised pre-training method for speech-to-text models using pseudo languages, improving performance across ASR, translation, and named entity recognition without relying on extensive labeled data.

Contribution

It is the first to pre-train both encoder and decoder parts of speech models using pseudo language representations, enhancing various speech tasks.

Findings

State-of-the-art results in spoken named entity recognition.

Consistent improvements in speech-to-text translation across 20 language pairs.

Comparable performance to recent optimized methods in ASR.

Abstract

We introduce Wav2Seq, the first self-supervised approach to pre-train both parts of encoder-decoder models for speech data. We induce a pseudo language as a compact discrete representation, and formulate a self-supervised pseudo speech recognition task -- transcribing audio inputs into pseudo subword sequences. This process stands on its own, or can be applied as low-cost second-stage pre-training. We experiment with automatic speech recognition (ASR), spoken named entity recognition, and speech-to-text translation. We set new state-of-the-art results for end-to-end spoken named entity recognition, and show consistent improvements on 20 language pairs for speech-to-text translation, even when competing methods use additional text data for training. Finally, on ASR, our approach enables encoder-decoder methods to benefit from pre-training for all parts of the network, and shows comparable performance to highly optimized recent methods.