handler.py

from typing import  Dict
from transformers.pipelines.audio_utils import ffmpeg_read
import whisper
import torch

SAMPLE_RATE = 16000

MODEL_NAME = "openai/whisper-large" #this always needs to stay in line 8 :D sorry for the hackiness
lang = "dk"




class EndpointHandler():
    def __init__(self, path=""):
        pipe = pipeline(
            task="automatic-speech-recognition",
            model=MODEL_NAME,
            chunk_length_s=30,
            device=device,
        )
        
        # load the model
        #self.model = whisper.load_model("large")
        self.model.config.forced_decoder_ids = pipe.tokenizer.get_decoder_prompt_ids(language=lang, task="transcribe")


    def __call__(self, data: Dict[str, bytes]) -> Dict[str, str]:
        """
        Args:
            data (:obj:):
                includes the deserialized audio file as bytes
        Return:
            A :obj:`dict`:. base64 encoded image
        """
        # process input
        inputs = data.pop("inputs", data)
        audio_nparray = ffmpeg_read(inputs, SAMPLE_RATE)
        audio_tensor= torch.from_numpy(audio_nparray)
        
        # run inference pipeline        
        result = self.model.transcribe(audio_nparray)


        # postprocess the prediction
        return {"tekst": result["text"]}