diodiogod
edbcb3422d
Major fixes for VibeVoice engine functionality: Audio Quality Improvements: - Replace linear interpolation resampling with librosa.resample (matches official) - Fix normalization to use -25 dB FS RMS instead of peak normalization (matches official) - Implement proper stereo-to-mono conversion with channel averaging (matches official) Multi-Character Voice Switching: - Add proper character mapping using get_character_mapping() like F5-TTS does - Build complete voice_mapping with fallbacks for all parsed characters - Fix character parser to auto-discover characters when available_characters is None - Remove unnecessary fallback logic from processor (handled at wrapper level) Pause Tag Support: - Add modular pause tag processing like F5-TTS using generate_vibevoice_with_pause_tags() - Integrate PauseTagProcessor for [pause:X] support in VibeVoice - Follow established modular pattern from F5-TTS and ChatterBox implementations Cache System Improvements: - Fix floating point precision issues in cache key generation (round to 3 decimals) - Add proper parameter propagation to cached engine instances - Ensure cache keys include all generation parameters properly All changes tested and working - multi-character switching now works correctly and pause tags are processed properly using the modular system.
VibeVoice Official Demo Voice Samples
These are the official voice samples from Microsoft's VibeVoice repository, used under MIT License.
Sample Information
English Voices
- en-Alice_woman.wav (9.27s) - Short sample, female voice
- en-Carter_man.wav (27.73s) - Male voice
- en-Frank_man.wav (24.13s) - Male voice
- en-Maya_woman.wav (27.20s) - Female voice
- en-Mary_woman_bgm.wav (26.93s) - Female voice with background music
Other Languages
- in-Samuel_man.wav (27s) - Indian accent, male
- zh-Bowen_man.wav (29.47s) - Chinese male
- zh-Xinran_woman.wav (27.87s) - Chinese female
- zh-Anchen_man_bgm.wav (24.53s) - Chinese male with background music
Usage Notes
- Recommended duration: 24-30 seconds (most samples follow this)
- Audio specs: 16kHz sample rate, mono, WAV format
- BGM samples: Show VibeVoice can handle background music in voice references
- Clean samples: Recommended for voice cloning (avoid BGM variants unless desired)
License
These samples are from the Microsoft VibeVoice repository and are available under MIT License. Copyright (c) 2025 Microsoft.