WhisperCore is in pre-beta hardening with ongoing reliability and readiness work.
Features
- On-device transcription using OpenAI Whisper models (tiny → large-v3)
- Speaker diarization — automatically labels who said what
- CUDA GPU acceleration for NVIDIA GPUs
- Support for 99+ languages with auto-detection
- Export to SRT, VTT, TXT, and JSON formats
- Batch processing — queue multiple files at once
- Real-time progress tracking with ETA
- Audio format support: MP3, WAV, M4A, FLAC, OGG, WMA, AAC, WEBM, MP4, MKV
Privacy
- 100% offline processing — no audio data ever leaves your device
- One-time license activation, then fully offline
- Zero telemetry by default
Technical
- Windows 10/11 (64-bit) support
- CPU and GPU inference modes
- Automatic model downloading and caching
- Session-based error recovery