Fix download in LibriCSS recipe (#1148)

* add transform attribute for MixedCut

* add mix_first option in normalize_loudness

* handle the case when mix is called on MixedCut with existing transforms

* add test for mixing with transformed MixedCut

* enhancements and bug fixes

* small changes in some cutset methods

* small fix in error message

* return word alignments from ami recipe

* add word alignments for ICSI

* remove unwanted whitespace

* fix IHM preparation

* remove words with zero or negative duration

* ensure word alignments respect segment boundary

* add save-to-wav option for icsi

* add test for mixing cut with recording

* style fix

* add data prep for voxpopuli

* small changes in recipes

* changes for max segment duration

* remove extra code

* add alignment scores from CTM

* minor change

* made suggested changes

* apply change to multi custom merge func

* remove old code

* fix failing tests

* add tests for trim to alignments with max segment duration

* add tests for merge supervisions

* fix bug in eval2000

* remove storing unnecessary things

* add fix_manifests for all recipes

* fix unzipping

lhotse/recipes/libricss.py

@@ -10,10 +10,13 @@
 import json
 import logging
 import subprocess
+import zipfile
 from collections import defaultdict
 from pathlib import Path
 from typing import Dict, Union
 
+from tqdm import tqdm
+
 from lhotse import (
     CutSet,
     RecordingSet,
@@ -120,7 +123,9 @@ def download_libricss(target_dir: Pathlike, force_download: bool = False) -> Pat
     # Extract the zipped file
     if not corpus_dir.exists() or force_download:
         logging.info(f"Extracting {corpus_zip} to {target_dir}")
-        corpus_zip.unzip(target_dir)
+        with zipfile.ZipFile(corpus_zip, "r") as corpus_zip:
+            for member in tqdm(corpus_zip.infolist(), desc="Extracting"):
+                corpus_zip.extract(member, target_dir)
 
     return target_dir