utils/model_validity.py

from huggingface_hub import model_info
from huggingface_hub.utils import RepositoryNotFoundError
import re

def is_valid_asr_model(model_id: str) -> bool:
    try:
        model_id = re.sub(r"[^a-zA-Z0-9/_\-.]", "", model_id)  # Sanitize the model ID
        info = model_info(model_id)
        # Optionally check if it's an ASR model (i.e., "automatic-speech-recognition" in the tags)
        return "automatic-speech-recognition" in info.tags
    except RepositoryNotFoundError:
        return False

# Test examples
# print(is_valid_asr_model("facebook/hubert-large-ls960-ft"))     # True
# print(is_valid_asr_model("facebook/hubert-largeXX-ls960-ft"))   # False