"""Download the Nigerian/West African subset of the Speech Accent Archive used by eval/saa_gold.py. Source: Weinberger, S. (2015). Speech Accent Archive, George Mason University; audio and IPA transcriptions on OSF (https://osf.io/yh23d, https://osf.io/e2j5y), licence CC BY-NC-SA 4.0. The file list is eval/saa_west_african_subset.tsv. python -m eval.fetch_saa """ from __future__ import annotations import csv import urllib.request from pathlib import Path HERE = Path(__file__).resolve().parent DEST = HERE.parent / "data" / "resources" / "speech_accent_archive" def main(): with open(HERE / "saa_west_african_subset.tsv", encoding="utf-8") as f: rows = list(csv.DictReader(f, delimiter="\t")) for sub in ("flac", "transcription_texts"): (DEST / sub).mkdir(parents=True, exist_ok=True) got = 0 for r in rows: for url, path in ((r.get("flac_url"), DEST / "flac" / f"{r['sample']}.flac"), (r.get("txt_url") if r.get("has_text_ipa") == "True" else None, DEST / "transcription_texts" / f"{r['sample']}.txt")): if url and not path.exists(): urllib.request.urlretrieve(url, path) got += 1 print(f"downloaded {got} files into {DEST}") if __name__ == "__main__": main()