transcriber.py

import torch
from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor, pipeline

def transcribe(audio_path, translate):
    device = "cuda:0" if torch.cuda.is_available() else "cpu"
    torch_dtype = torch.float16 if torch.cuda.is_available() else torch.float32

    model_id = "openai/whisper-large-v3"

    model = AutoModelForSpeechSeq2Seq.from_pretrained(
        model_id, torch_dtype=torch_dtype, low_cpu_mem_usage=True, use_safetensors=True
    )
    model.to(device)

    processor = AutoProcessor.from_pretrained(model_id)

    pipe = pipeline(
        "automatic-speech-recognition",
        model=model,
        tokenizer=processor.tokenizer,
        feature_extractor=processor.feature_extractor,
        max_new_tokens=128,
        chunk_length_s=30,
        batch_size=16,
        return_timestamps=True,
        torch_dtype=torch_dtype,
        device=device,
    )

    options = {"task": "translate"} if translate else {"language": "polish"}
    print(f"Rozpoczęto tranksrypcję pliku {audio_path} z opcjami {options}")
    result = pipe(audio_path, generate_kwargs=options)
    print(f"Transkrypacja zakończona: {result}")
    text = [chunk.get('text') for chunk in result["chunks"]]
    return ''.join(map(str,text))