SSL speech models encode articulatory movements cross-lingually, reaching Pearson r of 0.68 with only ~5 minutes of EMA training data from bilingual speakers, and multilingual models outperform monolingual ones.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
eess.AS 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
How Bilingual Are SSL Speech Models? Cross-Lingual Probing of Articulatory Encoding with Finnish and Russian EMA
SSL speech models encode articulatory movements cross-lingually, reaching Pearson r of 0.68 with only ~5 minutes of EMA training data from bilingual speakers, and multilingual models outperform monolingual ones.