It seems like WhisperJAV has a hard time running on M1 MAX on transformer mode

[dadlaugh]

--mode transformers hangs silently on Apple Silicon (MPS) after model
loads

After the model loads successfully, the program hangs indefinitely with no
output, progress, or error:

2026-03-15 - whisperjav - INFO - Loaded in 4.7s
← hangs here forever, no further output

The process does not crash, does not print an error, and does not complete. It
must be force-killed with Ctrl+C.

Environment

• macOS Apple Silicon (MPS backend, auto-detected)
• Model: kotoba-tech/kotoba-whisper-bilingual-v1.0
• Command: --mode transformers --hf-device auto --hf-batch-size 16 (default)