v0.8.8 v0.8.9
low risk
azophy/sshifuSep 26, 2026 → Sep 26, 20260 days between releases
What's New
Fixes synthesis failures, adds new preset, and bumps CPU native version.
New Features
- New `--preset windows-3900x` profile for optimized performance.
- Automatic demo voice download and verification.
- New `--cpu-fused-attention` flag for CPU native 0.4.1.
Improvements
- RTF reduced by 6.4% on Ryzen 3900X.
- CPU native version bumped from 0.4.0 to 0.4.1.
- Streaming playback corrected in Voice Studio.
Bug Fixes
- Fixed standalone synthesis failures for language codes like 'de' and 'en'.
- Corrected streaming playback issues in Voice Studio.
Upgrade Notes
Upgrade CPU native from 0.4.0 to 0.4.1 and use the new `--cpu-fused-attention` flag. Be aware that floating-point evaluation order changes may affect seeded audio output.
The Two Releases
v0.8.8
RealtimeTTS 0.8.8
Sep 26, 2026· 25 days ago
Adds optional Qwen CPU decoder overlap for performance gains and restores the full emotional demo with updated Numba support for Intel Macs.
Full release notesv0.8.9
RealtimeTTS 0.8.9
Sep 26, 2026· 25 days ago
Fixes standalone synthesis failures for specific language codes and corrects streaming playback in Voice Studio with a new preset and demo voice support.
Full release notes