fix: add retry mechanism to prevent EOS early termination sentence dropping

feat: add fade_duration API parameter for smooth fade in/out transitions

feat: add fragment_interval API parameter for adjustable sentence gap

Delete API_DOCUMENTATION.md

Delete openapi.json

fix: add sentence interval padding and increase ref audio padding for cross-lang TTS

fix: ToneSandhi IndexError and proper error handling for missing output files

fix: clear ReferenceAudio cache before deleting temp files

feat: add speed control support

fix: add mzm model download to Dockerfile

sync all fixes: prompt leakage, cross-lang, ref_cache update, and file wait logic

Upload 1.txt

Upload 68 files

Delete Shared.py

Delete TTSPlayer.py

Delete Internal.py

Delete Inference.py

Upload 5 files

Upload Dockerfile

Upload 2 files

Upload 68 files

Upload app.py

Upload 28 files

initial commit