OpenAI's Whisper large-v3 model takes approximately 10–15 minutes to transcribe speech from a 1-hour video on an Apple M2 chip, compared to approximately 2 minutes on a cloud-based A100 GPU.