-
Update short demo demo_zh.wav with complete utterance
-
Update short demo demo_en.wav with complete utterance
-
Force update demo_zh_long.wav with complete-ending version
-
Force update demo_en_long.wav with complete-ending version
-
Regenerate 20s demo audio with complete ending (EN 20.1s / ZH 17.3s, full sentence ASR-verified)
-
Add 20s long demo audio (en/zh, sampling 0.8/0.95/rep1.1); post_config sampling default for long-text stability
-
Create config.json
-
v2 audio fix: RMS-norm fine_local input, zero-pad decoder bits, fix ISTFT log-mag stride; refreshed demo audio (ASR-verified)
-
Update model_meta.json for s8 ax-llm route
-
Replace with s8 build: LM s8 (Pulsar2 7.0 llm_build2), deterministic greedy generation, refreshed demo audio; drop stale lm_v1_60fix/binaries
-
Delete bin/tts_driver with huggingface_hub
-
Delete bin/hojo_tts_cpp with huggingface_hub
-
Delete bin/hojo_tts_text with huggingface_hub
-
Delete folder models/lm_v1_60fix with huggingface_hub
-
Add zh/en demo audio samples (24kHz wav) + README audio players
-
Track wav samples with LFS/Xet
-
Add YAML frontmatter to README
-
Release Hojo-TTS-Light-40M AX650/NPU3: C++ SDK + axmodels (host-KV LM, whisper-style U16 config)
-
Track axmodel and board binaries with LFS/Xet
-
initial commit
