Problem

Both small and large Whisper models fail to recognize Mandarin Chinese. Clear, noise-free Mandarin audio during live meeting recordings is being transcribed as English text instead.

Steps to Reproduce

1. Start a live meeting recording with Mandarin Chinese audio
2. Use either small or large Whisper model for transcription
3. Observe the transcribed output

Expected Behavior

Mandarin Chinese speech should be transcribed as Mandarin Chinese text

Actual Behavior

Clear Mandarin Chinese audio is being transcribed as English text

Environment

• Platform: macOS (aarch64)
• OS Version: macOS 26.3.1
• App Version: 1.0.19
• Build: f4abc33
• Locale: en-US
• Audio Quality: Clear, no background noise

Additional Context

• This has never worked (not a regression)
• Tested with both small and large Whisper models
• Issue occurs during live meeting recordings