May 7 16:26 README: drop preview status, add 25 voices, fix GH URL typo (shreyask -> shreyaskarnik), link demo + blog 14:27 Update README.md May 6 06:26 voices: bulk-add 25 deployed voicepacks (4 trained + 19 stock-Kokoro crossovers + 2 synthetic) Mirrors the Space-deployed voicepack set as of 2026-05-06. Same .pt files as published to shreyask/bol-tts-marathi-onnx/voices/ — content-equivalent to the .bin tensors served at the Space (just torch.save format). Apr 29 18:53 Upload voices/mm_dnyanesh.pt with huggingface_hub 18:53 Upload voices/mf_mukta.pt with huggingface_hub 18:52 Upload kokoro-mr-v0_2.pth with huggingface_hub Apr 25 21:03 re-extract Asha+Vivek from trim_silence-cleaned Rasa source — fixes ~15ms tail clip 13:08 Stage 2 final (epoch_2nd_00009) — v0.1 release 13:08 voicepacks re-extracted from Stage 2 final (epoch_2nd_00009) 01:22 re-extract voicepacks from Stage 2 epoch 5 checkpoint 01:17 update Kokoro-format weights from Stage 2 epoch 5 Apr 24 22:13 config: add model_type/architectures for transformers.js 21:45 v0.1-preview refresh: Stage 2 epoch 4 kokoro-format .pth (cp-alias fix) 21:44 v0.1-preview refresh: Stage 2 epoch 4 ckpt + all 4 voicepacks (Asha, Vivek, Mukta, Dnyanesh) 18:53 v0.1-preview: Stage 2 epoch 2 checkpoint (bs=8), Asha + Vivek voicepacks 18:53 initial commit 
