Commit History

  1. Add reusable speaker and emotion-reference dubbing API

    Verified signature
    f508498

  1. fix: pre-download aux models only (avoid build OOM)

    Verified signature
    755640e
  2. perf: bake aux models + wetext FST into image (faster cold start)

    Verified signature
    6a2f21a
  3. fix: drop numpy pin (indextts requires numpy==2.2.6)

    Verified signature
    94f2358
  4. IndexTTS-2 FastAPI (GPU, baked weights)

    Verified signature
    e15e1c1
  5. initial commit

    Verified signature
    fd94d3e