fairseq S2T: Fast Speech-to-Text Modeling with fairseq

TL;DR

fairseq S2T is a scalable, extensible extension of fairseq for speech-to-text tasks, supporting various model architectures and integrating translation and language models for comprehensive end-to-end workflows.

Contribution

It introduces a flexible fairseq extension for speech-to-text modeling with multiple architectures and seamless integration of translation and language models.

Findings

Supports RNN, Transformer, and Conformer models

Provides detailed training recipes and workflows

Enables multi-task and transfer learning

Abstract

We introduce fairseq S2T, a fairseq extension for speech-to-text (S2T) modeling tasks such as end-to-end speech recognition and speech-to-text translation. It follows fairseq's careful design for scalability and extensibility. We provide end-to-end workflows from data pre-processing, model training to offline (online) inference. We implement state-of-the-art RNN-based, Transformer-based as well as Conformer-based models and open-source detailed training recipes. Fairseq's machine translation models and language models can be seamlessly integrated into S2T workflows for multi-task learning or transfer learning. Fairseq S2T documentation and examples are available at https://github.com/pytorch/fairseq/tree/master/examples/speech_to_text.