Simpaudio v1.0.0 - fully working build
This replaces the initial v1.0.0 binaries (same version, no new release). The original release had broken AI features inside the packaged app - they now all work.
What was fixed:
- Kokoro TTS no longer produces "No audio was generated" (model is now actually passed to the pipeline)
- Piper, Kokoro and Whisper voice data is now bundled correctly in the app
- Bundled espeak-ng phonemizer + en_core_web_sm model so Kokoro handles names/unknown words in all 8 supported languages
- Blending voices (e.g. "af_heart,af_bella") works and the Test Play button actually plays audio
- MP3 preview/export works (ffmpeg bundled); format choice is remembered between sessions
- SRT subtitles now use real segment timestamps
- Fixed a UI crash risk from status updates coming from background threads
- Added a built-in self-test (Simpaudio.exe --selftest) to verify engines
First-run notes:
- Piper voices, the Kokoro model and the Whisper model download automatically on first use (one-time).
- Antivirus may flag unsigned executables; this is normal for PyInstaller apps.
Verified: all engine self-tests pass in the packaged exe (Piper, Kokoro, voice blending, Whisper STT).