test-audio-len.py

import datasets
from datasets import load_dataset
import pandas as pd

# test reading all subsets for "test" split
_BIGOS_SUBSETS = ["pjatk-clarin_mobile-15", "pjatk-clarin_studio-15", "fair-mls-20", "mailabs-corpus_librivox-19", "mozilla-common_voice_15-23", "pwr-azon_read-20", "pwr-azon_spont-20", "pwr-maleset-unk", "pwr-shortwords-unk", "pwr-viu-unk", "google-fleurs-22", "polyai-minds14-21"]

splits=["test", "validation", "train"]
# Refer to documentation for the descriptions of splits and subsets

print("Testing dataset all subsets")

dataset_name = "pl-asr-bigos-v2"
hf_account = "amu-cai"

hf_db_name="/".join([hf_account,dataset_name])
print(hf_db_name, hf_db_name)

print(f"Testing dataset: {dataset_name} from account {hf_account} for all subsets")

print("Checking build script locally - test split")
dataset_local = load_dataset(f"{dataset_name}.py", "pwr-viu-unk")

#print(dataset_local)
print(dataset_local["test"][1])
print(dataset_local["test"][1]['audio']['array'])
print(type(dataset_local["test"][1]['audio']['array']))
print(len(dataset_local["test"][1]['audio']['array']))