v0.5.4

andimarafioti/faster-qwen3-ttsv0.5.4Oct 5, 2026by andimarafioti

Upgrading?

AI Summary

This release upgrades the GGML runtime dependency to qwentts-cpp-python 0.5.0, utilizing the new qwentts.cpp ABI v5. This update applies to the Apple Silicon default install and the GGML extra, requiring native libraries to be rebuilt for the new ABI.

Key Highlights

  • Upgraded to qwentts-cpp-python 0.5.0
  • Switch to qwentts.cpp ABI v5
  • Requires native library rebuilds for ABI v5 compatibility
  • Validated with real buffered and streaming synthesis

Breaking Changes

  • Native libraries must be rebuilt for the wrapper's ABI v5 revision

Full Release Notes

Upgrade the GGML runtime dependency to qwentts-cpp-python 0.5.0 with qwentts.cpp ABI v5.

- Require 0.5.0 for the Apple Silicon default install and the GGML extra.
- Preserve quiet native logging by default, with opt-in verbose diagnostics.
- Update PyPI, Hugging Face wheel, and local-build installation instructions.
- Explicitly supplied native libraries must be rebuilt for the wrapper's ABI v5 revision.

Validated with 75 passing tests (15 CUDA-dependent tests skipped) and real buffered and streaming synthesis using the released 0.5.0 macOS wheel on CPU.

Upgrade with `pip install --upgrade faster-qwen3-tts` or `pip install --upgrade "faster-qwen3-tts[ggml]"`.

Related Repositories