mHuBERT-147: A Compact Multilingual HuBERT Model
This provides a parameter-efficient solution for multilingual speech tasks, though it is incremental as it builds on the HuBERT approach.
They tackled the problem of creating a compact multilingual speech representation model by training mHuBERT-147 on 90K hours of data, achieving second and first place on ML-SUPERB leaderboards and outperforming larger models like XLS-R and MMS in ASR/LID tasks.
We present mHuBERT-147, the first general-purpose massively multilingual HuBERT speech representation model trained on 90K hours of clean, open-license data. To scale up the multi-iteration HuBERT approach, we use faiss-based clustering, achieving 5.2x faster label assignment than the original method. We also apply a new multilingual batching up-sampling strategy, leveraging both language and dataset diversity. After 3 training iterations, our compact 95M parameter mHuBERT-147 outperforms larger models trained on substantially more data. We rank second and first on the ML-SUPERB 10min and 1h leaderboards, with SOTA scores for 3 tasks. Across ASR/LID tasks, our model consistently surpasses XLS-R (300M params; 436K hours) and demonstrates strong competitiveness against the much larger MMS (1B params; 491K hours). Our findings indicate that mHuBERT-147 is a promising model for multilingual speech tasks, offering an unprecedented balance between high performance and parameter efficiency.