Release Jaiden Voice4 0.6B direct fine-tune (#2)

Upload v6_seg5.tar.gz with huggingface_hub

Upload v_old_transfer.tar.gz with huggingface_hub

Upload v6_transfer.tar.gz with huggingface_hub

v5: Vivian voice (young American female, F0=304Hz), UTMOS pending, 200x EOS

v4: Ender voice (young female), UTMOS pending, 200x EOS, 20 Stage2 epochs

v3: American female voice (serena), UTMOS 4.44, EOS 20/20, 200x EOS weight

v2: Voice-cloned two-stage (UTMOS 4.32, SIM 0.79) — better speaker similarity

Add detailed model card with training methodology and results

Upload two-stage Fish-distilled Qwen3-TTS 0.6B checkpoint (UTMOS 4.35, 250 samples)

initial commit