BriefGPT.xyz
Sep, 2021
简单而有效的零样本跨语言音素识别
Simple and Effective Zero-shot Cross-lingual Phoneme Recognition
HTML
PDF
Qiantong Xu, Alexei Baevski, Michael Auli
TL;DR
本文通过使用发音特征将多种训练语言的音素映射到目标语言中,对多语言预训练的 wav2vec 2.0 模型进行微调,以在没有标记数据的情况下提高其对未见过的语言的识别能力,并在实验中取得了较优效果。
Abstract
Recent progress in
self-training
,
self-supervised pretraining
and
unsupervised learning
enabled well performing speech recognition systems
→