BriefGPT.xyz
Oct, 2023
零對應跨模態轉換的模塊化語音轉文本翻譯
Modular Speech-to-Text Translation for Zero-Shot Cross-Modal Transfer
HTML
PDF
Paul-Ambroise Duquenne, Holger Schwenk, Benoît Sagot
TL;DR
通过独立训练的编码器和解码器，通过共享的固定大小表示组合，可以在语音到文本翻译中取得竞争力的性能，本研究表明这种方法可以通过多语种训练进一步改进，我们观察到在零-shot跨模态语音翻译中显著提高，甚至在几种语言上胜过基于XLSR的有监督方法。
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
→