Pre-trained Checkpoints

Pre-trained weights for the HuBERT and k-means models used in Spoken-Term Discovery using Discrete Speech Units.

Includes checkpoints for:

• HuBERT English from "HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Units" originally available in fairseq.
• HuBERT Chinese originally available from Tencent.
• HuBERT French (a multi-lingual model trained on French, English, and Spanish) from "Textless Speech-to-Speech Translation on Real Data" also originally available in fairseq.