app.py

from flask import Flask, request, jsonify
import requests
import torch
import librosa
import os
from transformers import WhisperProcessor, WhisperForConditionalGeneration

# Set Hugging Face cache directory explicitly
os.environ["HF_HOME"] = "/tmp/hf_cache"

app = Flask(__name__)

model_id = "openai/whisper-base"
processor = WhisperProcessor.from_pretrained(model_id)
model = WhisperForConditionalGeneration.from_pretrained(model_id)

device = "cuda" if torch.cuda.is_available() else "cpu"
model.to(device)

forced_decoder_ids = processor.get_decoder_prompt_ids(language="he", task="transcribe")

# rest of the code remains unchanged...