from transformers import pipeline class WhisperTranscriber: def __init__(self, model_name: str, device: str = "cuda"): print("Loading Whisper model...") self.pipe = pipeline( task="automatic-speech-recognition", model=model_name, tokenizer=model_name, device=0 if device == "cuda" else -1 ) print("Whisper model loaded.") def transcribe(self, audio_file: str) -> str: result = self.pipe(audio_file, batch_size=4, return_timestamps=True) return result.get("text", "").strip()