from transformers import pipeline
import ffmpeg

transcriber = pipeline(model="openai/whisper-large-v2", torch_dtype="auto")
transcriber("https://huggingface.co/datasets/Narsil/asr_dummy/resolve/main/mlk.flac")