Diane v0.2.0

Whisper transcription + audio event detection.

New in v0.2.0:

• Replaced Apple SFSpeechRecognizer with whisper.cpp for better accuracy and punctuation
• Audio event detection via HuggingFace AST — 527 sound categories
• Environmental sounds appear as [birds chirping], [music], [clapping] etc.
• First-run download overlay while Whisper model is fetched (~142MB)
• Forward button (hold to fast-forward through tape)
• Tape spool animation — reels visually grow/shrink during playback
• Click any word in the transcript to jump to that moment
• Dictation types words during playback
• Stop button stops both recording and playback
• Rewind releases into forward playback from the rewound position

macOS (Apple Silicon)