male acoustic v2 data

male acoustic v2: mel-GAN

male long v4 (mel-GAN)

male v4 (mel-GAN)

add tianqi demo + v2 config note

female acoustic v2 data

female acoustic v2: mel-GAN + length_scale=1.0

default noise_scale=0.0 (stable long-sequence decoding)

v4 (ls=1.0, noise=0): demo_female_tianqi.wav

v4 (ls=1.0, noise=0): demo_female_long.wav

v4 (ls=1.0, noise=0): demo_female_news.wav

v4 (ls=1.0, noise=0): demo_female_casual.wav

add mel sharpen (before soft-gate): restore aspiration detail

v3 (mel-GAN + sharpen): demo_female_long.wav

v3 (mel-GAN + sharpen): demo_female_news.wav

v3 (mel-GAN + sharpen): demo_female_casual.wav

strip trailing silence padding in vocoder_chunked (single-chunk path)

strip tail silence: demo_male.wav

strip tail silence: demo_female_news.wav

strip tail silence: demo_female_casual.wav

add long-text demo audio cards

add long demo audio demo_male_long.wav

add long demo audio demo_female_long.wav

add demo audio cards to README

add demo audio demo_male.wav

add demo audio demo_female_news.wav

add demo audio demo_female_casual.wav

remove __pycache__

Use inline ONNX (single-file), fix tail_stretch dtype

Delete models/bigvgan_base.onnx.data with huggingface_hub

Delete models/acoustic_male.onnx.data with huggingface_hub

Delete models/acoustic_female.onnx.data with huggingface_hub

Release 1.0.0: Chinese TTS (Inflect-Micro-v2 arch + BigVGAN), ONNX + AX650 axmodel + checkpoints

initial commit