Whisper-test

Run

uv run python ./transcribe_whisper_v3.py --input "path/to/file.m4a" --device "cuda" --fp16 --model "large-v3"

test with RTX 5070 Ti