# pip install numpy==1.26.4 transformers==4.45.2 datasets==3.0.1 librosa==0.10.2.post1 jiwer==3.0.4 bitsandbytes==0.38.0.post1 accelerate==0.26.0

dataset_name = "dlxjj/common_voice_17_0"
language = "Japanese"
language_abbr = "ja"

from datasets import load_dataset, DatasetDict

common_voice = DatasetDict()

common_voice["train"] = load_dataset(dataset_name, language_abbr, split="train", trust_remote_code=True)

print(common_voice["train"][0])