from transformers import AutoModelForCTC, AutoProcessor
from datasets import load_dataset
train_ds = load_dataset('mozilla-foundation/common_voice_7_0', 'en', split='train[10:20]')

print(len(train_ds))