Unable to download voxpopuli in unit 6 [Errno22]

Hi folks,

When executing the following block cell (from Unit 6 of the Audio Course - subsection "Fine-tune SpeechT5") in my notebook environnement:
```
from datasets import load_dataset

dataset = load_dataset("facebook/voxpopuli", "nl", split="train")
len(dataset)
```

I get the following error message. The wav file mentionned at the end seems to change at each execution of the cell.
I didn't have any problem with all the other datasets that I had to download in all previous units of the course.

> ---------------------------------------------------------------------------
> OSError Traceback (most recent call last)
> Cell In[6], line 1
> ----> 1 dataset = load_dataset("facebook/voxpopuli", "nl")
> 2 len(dataset)
>
> File C:\ProgramData\anaconda3\envs\hf_audio_course\Lib\site-packages\datasets\load.py:2154, in load_dataset(path, name, data_dir, data_files, split, cache_dir, features, download_config, download_mode, verification_mode, keep_in_memory, save_infos, revision, token, streaming, num_proc, storage_options, trust_remote_code, **config_kwargs)
> 2151 return builder_instance.as_streaming_dataset(split=split)
> 2153 # Download and prepare data
> -> 2154 builder_instance.download_and_prepare(
> 2155 download_config=download_config,
> 2156 download_mode=download_mode,
> 2157 verification_mode=verification_mode,
> 2158 num_proc=num_proc,
> 2159 storage_options=storage_options,
> 2160 )
> 2162 # Build dataset for splits
> 2163 keep_in_memory = (
> 2164 keep_in_memory if keep_in_memory is not None else is_small_dataset(builder_instance.info.dataset_size)
> 2165 )
>
> File C:\ProgramData\anaconda3\envs\hf_audio_course\Lib\site-packages\datasets\builder.py:924, in DatasetBuilder.download_and_prepare(self, output_dir, download_config, download_mode, verification_mode, dl_manager, base_path, file_format, max_shard_size, num_proc, storage_options, **download_and_prepare_kwargs)
> 922 if num_proc is not None:
> 923 prepare_split_kwargs["num_proc"] = num_proc
> --> 924 self._download_and_prepare(
> 925 dl_manager=dl_manager,
> 926 verification_mode=verification_mode,
> 927 **prepare_split_kwargs,
> 928 **download_and_prepare_kwargs,
> 929 )
> 930 # Sync info
> 931 self.info.dataset_size = sum(split.num_bytes for split in self.info.splits.values())
>
> File C:\ProgramData\anaconda3\envs\hf_audio_course\Lib\site-packages\datasets\builder.py:1648, in GeneratorBasedBuilder._download_and_prepare(self, dl_manager, verification_mode, **prepare_splits_kwargs)
> 1647 def _download_and_prepare(self, dl_manager, verification_mode, **prepare_splits_kwargs):
> -> 1648 super()._download_and_prepare(
> 1649 dl_manager,
> 1650 verification_mode,
> 1651 check_duplicate_keys=verification_mode == VerificationMode.BASIC_CHECKS
> 1652 or verification_mode == VerificationMode.ALL_CHECKS,
> 1653 **prepare_splits_kwargs,
> 1654 )
>
> File C:\ProgramData\anaconda3\envs\hf_audio_course\Lib\site-packages\datasets\builder.py:978, in DatasetBuilder._download_and_prepare(self, dl_manager, verification_mode, **prepare_split_kwargs)
> 976 split_dict = SplitDict(dataset_name=self.dataset_name)
> 977 split_generators_kwargs = self._make_split_generators_kwargs(prepare_split_kwargs)
> --> 978 split_generators = self._split_generators(dl_manager, **split_generators_kwargs)
> 980 # Checksums verification
> 981 if verification_mode == VerificationMode.ALL_CHECKS and dl_manager.record_checksums:
>
> File ~\.cache\huggingface\modules\datasets_modules\datasets\facebook--voxpopuli\b5ff837284f0778eefe0f642734e142d8c3f574eba8c9c8a4b13602297f73604\voxpopuli.py:146, in Voxpopuli._split_generators(self, dl_manager)
> 142 meta_paths = dl_manager.download_and_extract(meta_urls)
> 143 audio_paths = dl_manager.download(audio_urls)
> 145 local_extracted_audio_paths = (
> --> 146 dl_manager.extract(audio_paths) if not dl_manager.is_streaming else
> 147 {
> 148 split: {lang: [None] * len(audio_paths[split][lang]) for lang in self.config.languages} for split in splits
> 149 }
> 150 )
> 151 if self.config.name == "en_accented":
> 152 return [
> 153 datasets.SplitGenerator(
> 154 name=datasets.Split.TEST,
> (...)
> 163 ),
> 164 ]
>
> File C:\ProgramData\anaconda3\envs\hf_audio_course\Lib\site-packages\datasets\download\download_manager.py:299, in DownloadManager.extract(self, path_or_paths)
> 297 download_config.extract_compressed_file = True
> 298 extract_func = partial(self._download_single, download_config=download_config)
> --> 299 extracted_paths = map_nested(
> 300 extract_func,
> 301 path_or_paths,
> 302 num_proc=download_config.num_proc,
> 303 desc="Extracting data files",
> 304 )
> 305 path_or_paths = NestedDataStructure(path_or_paths)
> 306 extracted_paths = NestedDataStructure(extracted_paths)
>
> File C:\ProgramData\anaconda3\envs\hf_audio_course\Lib\site-packages\datasets\utils\py_utils.py:512, in map_nested(function, data_struct, dict_only, map_list, map_tuple, map_numpy, num_proc, parallel_min_length, batched, batch_size, types, disable_tqdm, desc)
> 509 batch_size = max(len(iterable) // num_proc + int(len(iterable) % num_proc > 0), 1)
> 510 iterable = list(iter_batched(iterable, batch_size))
> 511 mapped = [
> --> 512 _single_map_nested((function, obj, batched, batch_size, types, None, True, None))
> 513 for obj in hf_tqdm(iterable, disable=disable_tqdm, desc=desc)
> 514 ]
> 515 if batched:
> 516 mapped = [mapped_item for mapped_batch in mapped for mapped_item in mapped_batch]
>
> File C:\ProgramData\anaconda3\envs\hf_audio_course\Lib\site-packages\datasets\utils\py_utils.py:396, in _single_map_nested(args)
> 393 with hf_tqdm(pbar_iterable, disable=disable_tqdm, position=rank, unit="obj", desc=pbar_desc) as pbar:
> 394 if isinstance(data_struct, dict):
> 395 return {
> --> 396 k: _single_map_nested((function, v, batched, batch_size, types, None, True, None)) for k, v in pbar
> 397 }
> 398 else:
> 399 mapped = [_single_map_nested((function, v, batched, batch_size, types, None, True, None)) for v in pbar]
>
> File C:\ProgramData\anaconda3\envs\hf_audio_course\Lib\site-packages\datasets\utils\py_utils.py:399, in _single_map_nested(args)
> 395 return {
> 396 k: _single_map_nested((function, v, batched, batch_size, types, None, True, None)) for k, v in pbar
> 397 }
> 398 else:
> --> 399 mapped = [_single_map_nested((function, v, batched, batch_size, types, None, True, None)) for v in pbar]
> 400 if isinstance(data_struct, list):
> 401 return mapped
>
> File C:\ProgramData\anaconda3\envs\hf_audio_course\Lib\site-packages\datasets\utils\py_utils.py:373, in _single_map_nested(args)
> 371 return function([data_struct])[0]
> 372 else:
> --> 373 return function(data_struct)
> 374 if (
> 375 batched
> 376 and not isinstance(data_struct, dict)
> 377 and isinstance(data_struct, types)
> 378 and all(not isinstance(v, (dict, types)) for v in data_struct)
> 379 ):
> 380 return [mapped_item for batch in iter_batched(data_struct, batch_size) for mapped_item in function(batch)]
>
> File C:\ProgramData\anaconda3\envs\hf_audio_course\Lib\site-packages\datasets\download\download_manager.py:229, in DownloadManager._download_single(self, url_or_filename, download_config)
> 226 if is_relative_path(url_or_filename):
> 227 # append the relative path to the base_path
> 228 url_or_filename = url_or_path_join(self._base_path, url_or_filename)
> --> 229 out = cached_path(url_or_filename, download_config=download_config)
> 230 out = tracked_str(out)
> 231 out.set_origin(url_or_filename)
>
> File C:\ProgramData\anaconda3\envs\hf_audio_course\Lib\site-packages\datasets\utils\file_utils.py:251, in cached_path(url_or_filename, download_config, **download_kwargs)
> 248 return output_path
> 250 # Eager extraction
> --> 251 output_path = ExtractManager(cache_dir=download_config.cache_dir).extract(
> 252 output_path, force_extract=download_config.force_extract
> 253 )
> 254 return relative_to_absolute_path(output_path)
>
> File C:\ProgramData\anaconda3\envs\hf_audio_course\Lib\site-packages\datasets\utils\extract.py:48, in ExtractManager.extract(self, input_path, force_extract)
> 46 output_path = self._get_output_path(input_path)
> 47 if self._do_extract(output_path, force_extract):
> ---> 48 self.extractor.extract(input_path, output_path, extractor_format)
> 49 return output_path
>
> File C:\ProgramData\anaconda3\envs\hf_audio_course\Lib\site-packages\datasets\utils\extract.py:332, in Extractor.extract(cls, input_path, output_path, extractor_format)
> 330 shutil.rmtree(output_path, ignore_errors=True)
> 331 extractor = cls.extractors[extractor_format]
> --> 332 return extractor.extract(input_path, output_path)
>
> File C:\ProgramData\anaconda3\envs\hf_audio_course\Lib\site-packages\datasets\utils\extract.py:126, in TarExtractor.extract(input_path, output_path)
> 124 os.makedirs(output_path, exist_ok=True)
> 125 tar_file = tarfile.open(input_path)
> --> 126 tar_file.extractall(output_path, members=TarExtractor.safemembers(tar_file, output_path))
> 127 tar_file.close()
>
> File C:\ProgramData\anaconda3\envs\hf_audio_course\Lib\tarfile.py:2302, in TarFile.extractall(self, path, members, numeric_owner, filter)
> 2297 if tarinfo.isdir():
> 2298 # For directories, delay setting attributes until later,
> 2299 # since permissions can interfere with extraction and
> 2300 # extracting contents can reset mtime.
> 2301 directories.append(tarinfo)
> -> 2302 self._extract_one(tarinfo, path, set_attrs=not tarinfo.isdir(),
> 2303 numeric_owner=numeric_owner)
> 2305 # Reverse sort directories.
> 2306 directories.sort(key=lambda a: a.name, reverse=True)
>
> File C:\ProgramData\anaconda3\envs\hf_audio_course\Lib\tarfile.py:2369, in TarFile._extract_one(self, tarinfo, path, set_attrs, numeric_owner)
> 2365 self._extract_member(tarinfo, os.path.join(path, tarinfo.name),
> 2366 set_attrs=set_attrs,
> 2367 numeric_owner=numeric_owner)
> 2368 except OSError as e:
> -> 2369 self._handle_fatal_error(e)
> 2370 except ExtractError as e:
> 2371 self._handle_nonfatal_error(e)
>
> File C:\ProgramData\anaconda3\envs\hf_audio_course\Lib\tarfile.py:2365, in TarFile._extract_one(self, tarinfo, path, set_attrs, numeric_owner)
> 2362 self._check("r")
> 2364 try:
> -> 2365 self._extract_member(tarinfo, os.path.join(path, tarinfo.name),
> 2366 set_attrs=set_attrs,
> 2367 numeric_owner=numeric_owner)
> 2368 except OSError as e:
> 2369 self._handle_fatal_error(e)
>
> File C:\ProgramData\anaconda3\envs\hf_audio_course\Lib\tarfile.py:2448, in TarFile._extract_member(self, tarinfo, targetpath, set_attrs, numeric_owner)
> 2445 self._dbg(1, tarinfo.name)
> 2447 if tarinfo.isreg():
> -> 2448 self.makefile(tarinfo, targetpath)
> 2449 elif tarinfo.isdir():
> 2450 self.makedir(tarinfo, targetpath)
>
> File C:\ProgramData\anaconda3\envs\hf_audio_course\Lib\tarfile.py:2494, in TarFile.makefile(self, tarinfo, targetpath)
> 2492 source.seek(tarinfo.offset_data)
> 2493 bufsize = self.copybufsize
> -> 2494 with bltn_open(targetpath, "wb") as target:
> 2495 if tarinfo.sparse is not None:
> 2496 for offset, size in tarinfo.sparse:
>
> OSError: [Errno 22] Invalid argument: 'C:\\Users\\APU\\.cache\\huggingface\\datasets\\downloads\\extracted\\49afd6ed6a1455c643d97d6bc9dcba821db2c51ee26fe1279355df74a0d10a31\\train_part_4\\20140116-0900-PLENARY-4-nl_20140116-10:46:51_1.wav'
>

Start by running the Unit 6 “Fine-tune SpeechT5” notebook cell on the reported Windows Anaconda environment with load_dataset("facebook/voxpopuli", "nl", split="train") and inspect the extraction failure in the traceback. Done means the VoxPopuli dataset downloads and extracts successfully so len(dataset) returns a value; the issue names no repository file or test.

Written by the indexing model from the issue text.