# Pretraining by Backtranslation for End-to-end ASR in Low-Resource Settings

## TL;DR

This paper presents a pretraining method using backtranslation with text and multilingual speech data to improve low-resource end-to-end ASR performance, achieving significant error rate reductions.

## Contribution

It introduces a novel pretraining approach for attention-based ASR models using only text and multilingual speech data, enhancing low-resource speech recognition.

## Key findings

- 20% average relative improvement over non-pretrained augmentation
- Additional 20-30% relative reduction with transcribed speech from nearby languages
- Effective pretraining method for low-resource end-to-end ASR

## Abstract

We explore training attention-based encoder-decoder ASR in low-resource settings. These models perform poorly when trained on small amounts of transcribed speech, in part because they depend on having sufficient target-side text to train the attention and decoder networks. In this paper we address this shortcoming by pretraining our network parameters using only text-based data and transcribed speech from other languages. We analyze the relative contributions of both sources of data. Across 3 test languages, our text-based approach resulted in a 20% average relative improvement over a text-based augmentation technique without pretraining. Using transcribed speech from nearby languages gives a further 20-30% relative reduction in character error rate.