XLSR-53

XLSR-53 is a multilingual benchmark used to evaluate self-supervised speech models on their performance in acoustic word embeddings across multiple languages.

Papers using XLSR-53 (4)

Analyzing Acoustic Word Embeddings From Pre-trained Self-supervised Speech Models2022 · 17 citesImproving Low-resource Speech Recognition With Pretrained Speech Models: Continued Pretraining Vs. Semi-supervised Training2022 · 5 citesThe THUEE System Description for the IARPA OpenASR21 Challenge2022 · 3 citesAnalyzing Acoustic Word Embeddings from Pre-trained Self-supervised Speech Models2022 · 1 cites