The CPC-big model and k-means checkpoints used in Analyzing Speaker Information in Self-Supervised Models to Improve Zero-Resource Speech Processing.