3 days ago
-
Use clean collection URL https://huggingface.co/collections/wq2012/flaxspeaker
-
Round 2: honest model card (training budget, variant labels, Vox1-O 10-spk subset, TFLite window, deviations), cleaned config.json, 26 dB Vox1-O subset metrics; remove misleading preprocessor_config.json
5 days ago
-
Round 2: Add scoring_type/loss_type/vad_config to config.json and voxceleb1_o_official to eval_results.json
8 days ago
-
Upload FlaxSpeaker pretrained model ls_conformer_tiny_ge2e