Modular Speech-to-Text Translation for Zero-Shot Cross-Modal Transfer

Abstract

Recent research has shown that independently trained encoders and decoders, combined through a shared fixed-size representation, can achieve competitive performance in speech-to-text translation. In this work, we show that this type of approach can be further improved with multilingual training. We observe significant improvements in zero-shot cross-modal speech translation, even outperforming a supervised approach based on XLSR for several languages.

References in corpus (4)

• Robust Speech Recognition via Large-Scale Weak Supervision
• The FLORES-101 Evaluation Benchmark for Low-Resource and Multilingual Machine Translation
• mSLAM: Massively multilingual joint pre-training for speech and text
• DeepNet: Scaling Transformers to 1,000 Layers