From 650057f96946d4766e7a5dfbb958058e38bd73fb Mon Sep 17 00:00:00 2001
From: Labbeti <etienne.labbe31@gmail.com>
Date: Mon, 4 Mar 2024 18:43:32 +0100
Subject: [PATCH] Version 0.5.1

---
 .pre-commit-config.yaml                       |  36 +++++
 CHANGELOG.md                                  |   7 +-
 CITATION.cff                                  |   6 +-
 README.md                                     |  14 +-
 docs/conf.py                                  |   3 +-
 docs/data_subsets.rst                         |  14 +-
 docs/usage.rst                                |   9 --
 examples/dataloader.ipynb                     |   8 +-
 requirements-dev.txt                          |   9 +-
 src/aac_datasets/__init__.py                  |   5 +-
 src/aac_datasets/check.py                     |   9 +-
 src/aac_datasets/datasets/audiocaps.py        |  33 ++--
 src/aac_datasets/datasets/base.py             | 142 +++++++++---------
 src/aac_datasets/datasets/clotho.py           |  18 +--
 .../datasets/functional/audiocaps.py          |  47 +++---
 .../datasets/functional/clotho.py             |  20 +--
 src/aac_datasets/datasets/functional/macs.py  |  15 +-
 .../datasets/functional/wavcaps.py            | 131 +++++++++++-----
 src/aac_datasets/datasets/macs.py             |  21 +--
 src/aac_datasets/datasets/wavcaps.py          |  25 ++-
 src/aac_datasets/utils/collections.py         |   7 +-
 src/aac_datasets/utils/download.py            |  31 ++--
 tests/test_datasets_base.py                   |   5 +-
 23 files changed, 332 insertions(+), 283 deletions(-)
 create mode 100644 .pre-commit-config.yaml

diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yaml
new file mode 100644
index 0000000..15b5d0f
--- /dev/null
+++ b/.pre-commit-config.yaml
@@ -0,0 +1,36 @@
+# exclude: ""
+
+repos:
+    # Format Code
+    - repo: https://github.com/ambv/black
+      rev: 22.10.0
+      hooks:
+        - id: black
+
+    # Sort imports
+    - repo: https://github.com/PyCQA/isort
+      rev: 5.12.0
+      hooks:
+      - id: isort
+        args: ["--profile", "black"]
+
+    # Formatting, Whitespace, etc
+    - repo: https://github.com/pre-commit/pre-commit-hooks
+      rev: v2.2.3
+      hooks:
+      - id: trailing-whitespace
+      - id: check-added-large-files
+        args: ['--maxkb=1000']
+      - id: check-ast
+      - id: check-json
+      - id: check-merge-conflict
+      - id: check-xml
+      - id: check-yaml
+      - id: debug-statements
+      - id: end-of-file-fixer
+      - id: requirements-txt-fixer
+      - id: mixed-line-ending
+        args: ['--fix=no']
+      - id: flake8
+        # args: ['--ignore=E203,E501,F811,E712,W503']
+        args: ['--config=.flake8']
diff --git a/CHANGELOG.md b/CHANGELOG.md
index 0194f92..c5a0040 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -2,6 +2,11 @@
 
 All notable changes to this project will be documented in this file.
 
+## [0.5.1] 2024-03-04
+### Fixed
+- WavCaps download preparation (#3).
+- `safe_rmdir` function when sub-directories are deleted.
+
 ## [0.5.0] 2024-01-05
 ### Changed
 - Update typing for paths with python class `Path`.
@@ -83,7 +88,7 @@ All notable changes to this project will be documented in this file.
 - Weak private methods are now strongly private in datasets.
 - Rename `item_transform` to `transform` in datasets.
 - Rename `load_tags` to `with_tags` in `AudioCaps`.
- 
+
 ### Fixed
 - AudioCaps loading when `with_tags` is False.
 - Clotho files download.
diff --git a/CITATION.cff b/CITATION.cff
index db5dd24..edde21f 100644
--- a/CITATION.cff
+++ b/CITATION.cff
@@ -11,7 +11,7 @@ authors:
     affiliation: IRIT
     orcid: 'https://orcid.org/0000-0002-7219-5463'
 repository-code: 'https://github.com/Labbeti/aac-datasets/'
-abstract: Audio Captioning datasets for Pytorch.
+abstract: Audio Captioning datasets for PyTorch.
 keywords:
   - audio
   - deep-learning
@@ -22,5 +22,5 @@ keywords:
   - captioning
   - audio-captioning
 license: MIT
-version: 0.5.0
-date-released: '2024-01-05'
+version: 0.5.1
+date-released: '2024-03-04'
diff --git a/README.md b/README.md
index 1143565..5bc8739 100644
--- a/README.md
+++ b/README.md
@@ -67,7 +67,7 @@ aac-datasets-download --root "." clotho --subsets "dev"
 ```
 
 ## Datasets information
-Here is the statistics for each dataset :
+`aac-datasets` package contains 4 different datasets :
 
 <!-- | | AudioCaps | Clotho | MACS | WavCaps |
 |:---:|:---:|:---:|:---:|:---:|
@@ -85,7 +85,7 @@ Here is the statistics for each dataset :
 
 For Clotho, the **dev** subset should be used for training, val for validation and eval for testing.
 
-Here is additional statistics on the train subset for AudioCaps, Clotho and MACS:
+Here is additional statistics on the train subset for AudioCaps, Clotho, MACS and WavCaps:
 
 | | AudioCaps/train | Clotho/dev | MACS/full | WavCaps/full |
 |:---:|:---:|:---:|:---:|:---:|
@@ -96,7 +96,9 @@ Here is additional statistics on the train subset for AudioCaps, Clotho and MACS
 | Nb captions | 49,838 | 19,195 | 17,275 | 403,050 |
 | Total nb words<sup>2</sup> | 402,482 | 217,362 | 160,006 | 3,161,823 |
 | Sentence size<sup>2</sup> | 2-52 | 8-20 | 5-40 | 2-38 |
-| Vocabulary<sup>2</sup> | 4724 | 4369 | 2721 | 24600 |
+| Vocabulary<sup>2</sup> | 4724 | 4369 | 2721 | 24,600 |
+| Annotated by | Human | Human | Human | Machine |
+| Corrected by | Human | Human | None | None |
 
 <sup>1</sup> This duration is estimated on the total duration of 46230/49838 files of 126.7h.
 
@@ -167,7 +169,7 @@ dataset = AudioCaps(
 [3] F. Font, A. Mesaros, D. P. W. Ellis, E. Fonseca, M. Fuentes, and B. Elizalde, Proceedings of the 6th Workshop on Detection and Classication of Acoustic Scenes and Events (DCASE 2021). Barcelona, Spain: Music Technology Group - Universitat Pompeu Fabra, Nov. 2021. Available: https://doi.org/10.5281/zenodo.5770113
 
 #### WavCaps
-[4] X. Mei et al., “WavCaps: A ChatGPT-Assisted Weakly-Labelled Audio Captioning Dataset for Audio-Language Multimodal Research,” arXiv preprint arXiv:2303.17395, 2023, [Online]. Available: https://arxiv.org/pdf/2303.17395.pdf 
+[4] X. Mei et al., “WavCaps: A ChatGPT-Assisted Weakly-Labelled Audio Captioning Dataset for Audio-Language Multimodal Research,” arXiv preprint arXiv:2303.17395, 2023, [Online]. Available: https://arxiv.org/pdf/2303.17395.pdf
 
 ## Cite the aac-datasets package
 If you use this software, please consider cite it as "Labbe, E. (2013). aac-datasets: Audio Captioning datasets for PyTorch.", or use the following BibTeX citation:
@@ -177,10 +179,10 @@ If you use this software, please consider cite it as "Labbe, E. (2013). aac-data
     Labbe_aac_datasets_2024,
     author = {Labbé, Etienne},
     license = {MIT},
-    month = {01},
+    month = {03},
     title = {{aac-datasets}},
     url = {https://github.com/Labbeti/aac-datasets/},
-    version = {0.5.0},
+    version = {0.5.1},
     year = {2024}
 }
 ```
diff --git a/docs/conf.py b/docs/conf.py
index c191960..a7f7237 100644
--- a/docs/conf.py
+++ b/docs/conf.py
@@ -17,7 +17,6 @@
 
 import aac_datasets
 
-
 # -- Project information -----------------------------------------------------
 
 project = aac_datasets.__name__
@@ -96,5 +95,5 @@
     "torchaudio": ("https://pytorch.org/audio/stable/", None),
 }
 
-# TODO: to be used with sphinx>=7.1
+# Only works with sphinx>=7.1
 maximum_signature_line_length = 10
diff --git a/docs/data_subsets.rst b/docs/data_subsets.rst
index 52fb158..bc4307f 100644
--- a/docs/data_subsets.rst
+++ b/docs/data_subsets.rst
@@ -7,7 +7,7 @@ The original AudioCaps dataset contains only 3 subsets : `train`, `val` and `tes
 
 A fourth subset named `train_v2` is another version of the train subset where captions has been manually corrected or deleted. For more details, see paper "CoNeTTE: An efficient Audio Captioning system leveraging multiple datasets with Task Embedding".
 
-Clotho 
+Clotho
 ########################
 Clotho contains 7 subsets:
 
@@ -28,12 +28,12 @@ WavCaps
 ########################
 WavCaps contains 6 subsets:
 
-- `as` : contains 108K files from AudioSet strongly labeled dataset,
+- `audioset` : contains 108K files from AudioSet strongly labeled dataset,
 - `bbc` : contains 31K files from BBC Sound Effects website,
-- `fsd` : contains 262K files from FreeSound website,
-- `sb` : contains 1.2K files from SoundBible website,
-- `as_noac` : contains 99K files from as subset without overlapping data with AudioCaps,
-- `fsd_nocl` : contains 258K files from fsd subset without overlapping data with Clotho (except for subsets of task 6a).
+- `freesound` : contains 262K files from FreeSound website,
+- `soundbible` : contains 1.2K files from SoundBible website,
+- `audioset_no_audiocaps` : contains 99K files from as subset without overlapping data with AudioCaps,
+- `freesound_no_clotho` : contains 258K files from fsd subset without overlapping data with Clotho (except for subsets of task 6a).
 
 Since WavCaps does not contains validation or testing subsets, all of their data is used as additional training data.
-The subsets as_noac and `fsd_nocl` are provided to avoid biases when evaluating on AudioCaps or Clotho datasets.
+The subsets as_noac and `freesound_no_clotho` are provided to avoid biases when evaluating on AudioCaps or Clotho datasets.
diff --git a/docs/usage.rst b/docs/usage.rst
index cbf70fc..ef3b2fd 100644
--- a/docs/usage.rst
+++ b/docs/usage.rst
@@ -13,15 +13,6 @@ You can download each dataset subset by using the download=True option in datase
 
     _ = Clotho("/my/path/to/data", subset="dev", download=True)
 
-You can also do the same by using functions :
-.. :caption: Download Clotho development dataset (python).
-
-.. code-block:: python
-
-    from aac_datasets.download import download_clotho_dataset
-
-    download_clotho_dataset("/my/path/to/data", subset="dev", download=True)
-
 Or by the command line :
 .. :caption: Download Clotho development dataset (command line).
 
diff --git a/examples/dataloader.ipynb b/examples/dataloader.ipynb
index ab5f405..4f0f1de 100644
--- a/examples/dataloader.ipynb
+++ b/examples/dataloader.ipynb
@@ -60,7 +60,7 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "aac-datasets version: 0.5.0\n"
+      "aac-datasets version: 0.5.1\n"
      ]
     }
    ],
@@ -108,7 +108,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 7,
+   "execution_count": 6,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -123,7 +123,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 8,
+   "execution_count": 7,
    "metadata": {},
    "outputs": [
     {
@@ -188,7 +188,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.10.4"
+   "version": "3.10.11"
   },
   "orig_nbformat": 4,
   "vscode": {
diff --git a/requirements-dev.txt b/requirements-dev.txt
index 8d9e47e..9f7070e 100644
--- a/requirements-dev.txt
+++ b/requirements-dev.txt
@@ -1,10 +1,11 @@
 # -*- coding: utf-8 -*-
 
-pytest==6.2.5
-flake8==4.0.1
-click==8.0.4
 black==22.10.0
+click==8.0.4
+flake8==4.0.1
 ipykernel==6.9.1
-twine==4.0.1
+pre-commit
+pytest==6.2.5
 sphinx
 sphinx-press-theme==0.8.0
+twine==4.0.1
diff --git a/src/aac_datasets/__init__.py b/src/aac_datasets/__init__.py
index eae6098..948962c 100644
--- a/src/aac_datasets/__init__.py
+++ b/src/aac_datasets/__init__.py
@@ -6,11 +6,13 @@
 
 __author__ = "Etienne Labbé (Labbeti)"
 __author_email__ = "labbeti.pub@gmail.com"
+__docs__ = "Audio Captioning Datasets"
+__docs_url__ = "https://aac-datasets.readthedocs.io/en/stable/"
 __license__ = "MIT"
 __maintainer__ = "Etienne Labbé (Labbeti)"
 __name__ = "aac-datasets"
 __status__ = "Development"
-__version__ = "0.5.0"
+__version__ = "0.5.1"
 
 
 from .datasets.audiocaps import AudioCaps
@@ -26,7 +28,6 @@
     set_default_ytdlp_path,
 )
 
-
 __all__ = [
     "AudioCaps",
     "Clotho",
diff --git a/src/aac_datasets/check.py b/src/aac_datasets/check.py
index 49e8ed5..7ddaab1 100644
--- a/src/aac_datasets/check.py
+++ b/src/aac_datasets/check.py
@@ -4,21 +4,18 @@
 import logging
 import os.path as osp
 import random
-
 from argparse import ArgumentParser, Namespace
 from typing import Dict, Iterable, Union
 
 import yaml
 
 import aac_datasets
-
 from aac_datasets.datasets.audiocaps import AudioCaps, AudioCapsCard
 from aac_datasets.datasets.clotho import Clotho, ClothoCard
 from aac_datasets.datasets.macs import MACS, MACSCard
 from aac_datasets.datasets.wavcaps import WavCaps, WavCapsCard
-from aac_datasets.utils.globals import get_default_root
 from aac_datasets.download import _setup_logging
-
+from aac_datasets.utils.globals import get_default_root
 
 DATASETS_NAMES = (AudioCapsCard.NAME, ClothoCard.NAME, MACSCard.NAME, WavCapsCard.NAME)
 
@@ -66,8 +63,8 @@ def check_directory(
                 ds = ds_class(root, subset, verbose=0)
                 if len(ds) > 0:
                     # Try to load a random item
-                    idx = random.randint(0, len(ds) - 1)
-                    _item = ds[idx]
+                    index = random.randint(0, len(ds) - 1)
+                    ds[index]
                 found_dsets[subset] = ds
 
             except RuntimeError:
diff --git a/src/aac_datasets/datasets/audiocaps.py b/src/aac_datasets/datasets/audiocaps.py
index d2a6f36..2b9fde6 100644
--- a/src/aac_datasets/datasets/audiocaps.py
+++ b/src/aac_datasets/datasets/audiocaps.py
@@ -3,23 +3,13 @@
 
 import logging
 import os.path as osp
-
 from pathlib import Path
-from typing import (
-    Any,
-    Callable,
-    ClassVar,
-    Dict,
-    List,
-    Optional,
-    Union,
-)
+from typing import Any, Callable, ClassVar, Dict, List, Optional, Union
 
 import torch
 import torchaudio
-
 from torch import Tensor
-from typing_extensions import TypedDict, NotRequired
+from typing_extensions import NotRequired, TypedDict
 
 try:
     # To support torchaudio >= 2.1.0
@@ -30,12 +20,11 @@
 from aac_datasets.datasets.base import AACDataset
 from aac_datasets.datasets.functional.audiocaps import (
     AudioCapsCard,
+    _get_audio_subset_dpath,
     download_audiocaps_dataset,
     load_audiocaps_dataset,
-    _get_audio_subset_dpath,
 )
-from aac_datasets.utils.globals import _get_root, _get_ffmpeg_path, _get_ytdlp_path
-
+from aac_datasets.utils.globals import _get_ffmpeg_path, _get_root, _get_ytdlp_path
 
 pylog = logging.getLogger(__name__)
 
@@ -98,7 +87,7 @@ def __init__(
         root: Union[str, Path, None] = None,
         subset: str = AudioCapsCard.DEFAULT_SUBSET,
         download: bool = False,
-        transform: Optional[Callable[[Dict[str, Any]], Any]] = None,
+        transform: Optional[Callable[[AudioCapsItem], Any]] = None,
         verbose: int = 0,
         force_download: bool = False,
         verify_files: bool = False,
@@ -284,10 +273,10 @@ def __repr__(self) -> str:
         return f"{AudioCapsCard.PRETTY_NAME}({repr_str})"
 
     # Private methods
-    def _load_audio(self, idx: int) -> Tensor:
-        if not self._raw_data["is_on_disk"][idx]:
+    def _load_audio(self, index: int) -> Tensor:
+        if not self._raw_data["is_on_disk"][index]:
             return torch.empty((0,))
-        fpath = self.at(idx, "fpath")
+        fpath = self.at(index, "fpath")
         audio, sr = torchaudio.load(fpath)  # type: ignore
 
         # Sanity check
@@ -302,9 +291,9 @@ def _load_audio(self, idx: int) -> Tensor:
             )
         return audio
 
-    def _load_audio_metadata(self, idx: int) -> AudioMetaData:
-        if not self._raw_data["is_on_disk"][idx]:
+    def _load_audio_metadata(self, index: int) -> AudioMetaData:
+        if not self._raw_data["is_on_disk"][index]:
             return AudioMetaData(-1, -1, -1, -1, "unknown_encoding")
-        fpath = self.at(idx, "fpath")
+        fpath = self.at(index, "fpath")
         audio_metadata = torchaudio.info(fpath)  # type: ignore
         return audio_metadata
diff --git a/src/aac_datasets/datasets/base.py b/src/aac_datasets/datasets/base.py
index d3a93c8..d89550a 100644
--- a/src/aac_datasets/datasets/base.py
+++ b/src/aac_datasets/datasets/base.py
@@ -3,7 +3,6 @@
 
 import logging
 import os.path as osp
-
 from typing import (
     Any,
     Callable,
@@ -21,8 +20,6 @@
 import torchaudio
 import tqdm
 
-from typing_extensions import TypedDict
-
 try:
     # To support torchaudio >= 2.1.0
     from torchaudio import AudioMetaData  # type: ignore
@@ -32,11 +29,10 @@
 from torch import Tensor
 from torch.utils.data.dataset import Dataset
 
-
 pylog = logging.getLogger(__name__)
 
 
-ItemType = TypeVar("ItemType", bound=TypedDict, covariant=True)
+ItemType = TypeVar("ItemType", covariant=True)
 
 
 class AACDataset(Generic[ItemType], Dataset[ItemType]):
@@ -46,7 +42,7 @@ class AACDataset(Generic[ItemType], Dataset[ItemType]):
     def __init__(
         self,
         raw_data: Optional[Dict[str, List[Any]]] = None,
-        transform: Optional[Callable] = None,
+        transform: Optional[Callable[[ItemType], Any]] = None,
         column_names: Optional[Iterable[str]] = None,
         flat_captions: bool = False,
         sr: Optional[int] = None,
@@ -149,31 +145,37 @@ def column_names(
         self._columns = columns
 
     @transform.setter
-    def transform(self, transform: Optional[Callable]) -> None:
+    def transform(self, transform: Optional[Callable[[ItemType], Any]]) -> None:
         self._transform = transform
 
     # Public methods
     @overload
-    def at(self, idx: int) -> ItemType:
+    def at(self, index: int) -> ItemType:
+        ...
+
+    @overload
+    def at(self, index: Union[Iterable[int], slice, None], column: str) -> List:
         ...
 
     @overload
-    def at(self, idx: Union[Iterable[int], slice, None], column: str) -> List:
+    def at(self, index: Union[Iterable[int], slice, None]) -> Dict[str, List]:
         ...
 
     @overload
     def at(
-        self, idx: Union[Iterable[int], slice, None], column: Union[Iterable[str], None]
+        self,
+        index: Union[Iterable[int], slice, None],
+        column: Union[Iterable[str], None],
     ) -> Dict[str, List]:
         ...
 
     @overload
-    def at(self, idx: Any, column: Any) -> Any:
+    def at(self, index: Any, column: Any) -> Any:
         ...
 
     def at(
         self,
-        idx: Union[int, Iterable[int], None, slice] = None,
+        index: Union[int, Iterable[int], None, slice] = None,
         column: Union[str, Iterable[str], None] = None,
     ) -> Any:
         """Get a specific data field.
@@ -182,63 +184,63 @@ def at(
         :param column: The name(s) of the column. Can be any value of :meth:`~Clotho.columns`.
         :returns: The field value. The type depends of the column.
         """
-        if idx is None:
-            idx = slice(None)
-        elif isinstance(idx, Tensor):
-            if idx.ndim not in (0, 1):
+        if index is None:
+            index = slice(None)
+        elif isinstance(index, Tensor):
+            if index.ndim not in (0, 1):
                 raise ValueError(
-                    f"Invalid number of dimensions for idx argument. (found idx.ndim={idx.ndim} but expected 0 or 1)"
+                    f"Invalid number of dimensions for index argument. (found index.ndim={index.ndim} but expected 0 or 1)"
                 )
-            elif idx.is_floating_point():
+            elif index.is_floating_point():
                 raise TypeError(
                     "Invalid tensor dtype. (found floating-point tensor but expected integer tensor)"
                 )
-            idx = idx.tolist()
+            index = index.tolist()
 
         if column is None:
             column = self.column_names
 
         if not isinstance(column, str) and isinstance(column, Iterable):
-            return {column_i: self.at(idx, column_i) for column_i in column}
+            return {column_i: self.at(index, column_i) for column_i in column}
 
-        if isinstance(idx, (int, slice)) and (
+        if isinstance(index, (int, slice)) and (
             column in self._raw_data.keys() and column not in self._online_fns
         ):
-            return self._raw_data[column][idx]  # type: ignore
+            return self._raw_data[column][index]  # type: ignore
 
-        if isinstance(idx, slice):
-            idx = range(len(self))[idx]
+        if isinstance(index, slice):
+            index = range(len(self))[index]
 
-        if isinstance(idx, Iterable):
-            idx = list(idx)
-            if all(isinstance(idx_i, bool) for idx_i in idx):
-                if len(idx) != len(self):
+        if isinstance(index, Iterable):
+            index = list(index)
+            if all(isinstance(idx_i, bool) for idx_i in index):
+                if len(index) != len(self):
                     raise IndexError(
-                        f"The length of the mask ({len(idx)}) does not match the length of the dataset ({len(self)})."
+                        f"The length of the mask ({len(index)}) does not match the length of the dataset ({len(self)})."
                     )
-                idx = [i for i, idx_i in enumerate(idx) if idx_i]
+                index = [i for i, idx_i in enumerate(index) if idx_i]
 
-            elif not all(isinstance(idx_i, int) for idx_i in idx):
+            elif not all(isinstance(idx_i, int) for idx_i in index):
                 raise TypeError(
-                    f"Invalid input type for idx={idx}. (expected Iterable[int], not Iterable[{idx[0].__class__.__name__}])"
+                    f"Invalid input type for index={index}. (expected Iterable[int], not Iterable[{index[0].__class__.__name__}])"
                 )
 
             values = [
                 self.at(idx_i, column)
                 for idx_i in tqdm.tqdm(
-                    idx,
+                    index,
                     desc=f"Loading column '{column}'...",
                     disable=self._verbose < 2,
                 )
             ]
             return values
 
-        if isinstance(idx, int):
-            return self._load_online_value(column, idx)
+        if isinstance(index, int):
+            return self._load_online_value(column, index)
         else:
             IDX_TYPES = ("int", "Iterable[int]", "None", "slice", "Tensor")
             raise TypeError(
-                f"Invalid argument type {type(idx)}. (expected one of {IDX_TYPES})"
+                f"Invalid argument type {type(index)}. (expected one of {IDX_TYPES})"
             )
 
     def has_raw_column(self, column: str) -> bool:
@@ -342,44 +344,45 @@ def preload_online_column(
 
     # Magic methods
     @overload
-    def __getitem__(self, idx: int) -> ItemType:
+    def __getitem__(self, index: int) -> ItemType:
         ...
 
     @overload
-    def __getitem__(self, idx: Tuple[Union[Iterable[int], slice, None], str]) -> List:
+    def __getitem__(self, index: Tuple[Union[Iterable[int], slice, None], str]) -> List:
         ...
 
     @overload
-    def __getitem__(self, idx: Union[Iterable[int], slice, None]) -> Dict[str, List]:
+    def __getitem__(self, index: Union[Iterable[int], slice, None]) -> Dict[str, List]:
         ...
 
     @overload
     def __getitem__(
-        self, idx: Tuple[Union[Iterable[int], slice, None], Union[Iterable[str], None]]
+        self,
+        index: Tuple[Union[Iterable[int], slice, None], Union[Iterable[str], None]],
     ) -> Dict[str, List]:
         ...
 
     @overload
-    def __getitem__(self, idx: Any) -> Any:
+    def __getitem__(self, index: Any) -> Any:
         ...
 
-    def __getitem__(self, idx: Any) -> Any:
+    def __getitem__(self, index: Any) -> Any:
         if (
-            isinstance(idx, tuple)
-            and len(idx) == 2
-            and (isinstance(idx[1], (str, Iterable)) or idx[1] is None)
+            isinstance(index, tuple)
+            and len(index) == 2
+            and (isinstance(index[1], (str, Iterable)) or index[1] is None)
         ):
-            idx, column = idx
+            index, column = index
         else:
             column = None
 
-        item = self.at(idx, column)
+        item = self.at(index, column)
         if (
-            isinstance(idx, int)
-            and (column is None or column == self._columns)
+            isinstance(index, int)
             and self._transform is not None
+            and (column is None or set(column) == set(self._columns))
         ):
-            item = self._transform(item)
+            item = self._transform(item)  # type: ignore
         return item
 
     def __len__(self) -> int:
@@ -394,7 +397,7 @@ def __len__(self) -> int:
     def __repr__(self) -> str:
         info = {
             "size": len(self),
-            "num_columns": len(self.column_names),
+            "num_columns": self.num_columns,
         }
         repr_str = ", ".join(f"{k}={v}" for k, v in info.items())
         return f"{self.__class__.__name__}({repr_str})"
@@ -421,17 +424,17 @@ def _unflat_raw_data(self) -> None:
         raw_data = _unflat_raw_data(self._raw_data, self._sizes)
         self._raw_data = raw_data
 
-    def _load_online_value(self, column: str, idx: int) -> Any:
+    def _load_online_value(self, column: str, index: int) -> Any:
         if column in self._online_fns:
             fn = self._online_fns[column]
-            return fn(self, idx)
+            return fn(self, index)
         else:
             raise ValueError(
-                f"Invalid argument column={column} at idx={idx}. (expected one of {self.all_columns})"
+                f"Invalid argument column={column} at index={index}. (expected one of {self.all_columns})"
             )
 
-    def _load_audio(self, idx: int) -> Tensor:
-        fpath = self.at(idx, "fpath")
+    def _load_audio(self, index: int) -> Tensor:
+        fpath = self.at(index, "fpath")
         audio_and_sr: Tuple[Tensor, int] = torchaudio.load(fpath)  # type: ignore
         audio, sr = audio_and_sr
 
@@ -447,33 +450,33 @@ def _load_audio(self, idx: int) -> Tensor:
             )
         return audio
 
-    def _load_audio_metadata(self, idx: int) -> AudioMetaData:
-        fpath = self.at(idx, "fpath")
+    def _load_audio_metadata(self, index: int) -> AudioMetaData:
+        fpath = self.at(index, "fpath")
         audio_metadata = torchaudio.info(fpath)  # type: ignore
         return audio_metadata
 
-    def _load_duration(self, idx: int) -> float:
-        audio_metadata: AudioMetaData = self.at(idx, "audio_metadata")
+    def _load_duration(self, index: int) -> float:
+        audio_metadata: AudioMetaData = self.at(index, "audio_metadata")
         duration = audio_metadata.num_frames / audio_metadata.sample_rate
         return duration
 
-    def _load_fname(self, idx: int) -> str:
-        fpath = self.at(idx, "fpath")
+    def _load_fname(self, index: int) -> str:
+        fpath = self.at(index, "fpath")
         fname = osp.basename(fpath)
         return fname
 
-    def _load_num_channels(self, idx: int) -> int:
-        audio_metadata = self.at(idx, "audio_metadata")
+    def _load_num_channels(self, index: int) -> int:
+        audio_metadata = self.at(index, "audio_metadata")
         num_channels = audio_metadata.num_channels
         return num_channels
 
-    def _load_num_frames(self, idx: int) -> int:
-        audio_metadata = self.at(idx, "audio_metadata")
+    def _load_num_frames(self, index: int) -> int:
+        audio_metadata = self.at(index, "audio_metadata")
         num_frames = audio_metadata.num_frames
         return num_frames
 
-    def _load_sr(self, idx: int) -> int:
-        audio_metadata = self.at(idx, "audio_metadata")
+    def _load_sr(self, index: int) -> int:
+        audio_metadata = self.at(index, "audio_metadata")
         sr = audio_metadata.sample_rate
         return sr
 
@@ -525,7 +528,8 @@ def _unflat_raw_data(
         for key in raw_data.keys():
             if key == caps_column:
                 caps = [
-                    raw_data_flat[key][idx][0] for idx in range(cumsize, cumsize + size)
+                    raw_data_flat[key][index][0]
+                    for index in range(cumsize, cumsize + size)
                 ]
                 raw_data[key].append(caps)
             else:
diff --git a/src/aac_datasets/datasets/clotho.py b/src/aac_datasets/datasets/clotho.py
index 96f048d..a9c0e62 100644
--- a/src/aac_datasets/datasets/clotho.py
+++ b/src/aac_datasets/datasets/clotho.py
@@ -3,29 +3,21 @@
 
 import logging
 import os.path as osp
-
 from pathlib import Path
-from typing import (
-    Callable,
-    ClassVar,
-    List,
-    Optional,
-    Union,
-)
+from typing import Any, Callable, ClassVar, List, Optional, Union
 
 from torch import Tensor
-from typing_extensions import TypedDict, NotRequired
+from typing_extensions import NotRequired, TypedDict
 
 from aac_datasets.datasets.base import AACDataset
 from aac_datasets.datasets.functional.clotho import (
     ClothoCard,
-    load_clotho_dataset,
-    download_clotho_dataset,
     _get_audio_subset_dpath,
+    download_clotho_dataset,
+    load_clotho_dataset,
 )
 from aac_datasets.utils.globals import _get_root
 
-
 pylog = logging.getLogger(__name__)
 
 
@@ -108,7 +100,7 @@ def __init__(
         root: Union[str, Path, None] = None,
         subset: str = ClothoCard.DEFAULT_SUBSET,
         download: bool = False,
-        transform: Optional[Callable] = None,
+        transform: Optional[Callable[[ClothoItem], Any]] = None,
         verbose: int = 0,
         force_download: bool = False,
         verify_files: bool = False,
diff --git a/src/aac_datasets/datasets/functional/audiocaps.py b/src/aac_datasets/datasets/functional/audiocaps.py
index 7e01126..1a7f466 100644
--- a/src/aac_datasets/datasets/functional/audiocaps.py
+++ b/src/aac_datasets/datasets/functional/audiocaps.py
@@ -7,19 +7,10 @@
 import os.path as osp
 import subprocess
 import time
-
 from concurrent.futures import ThreadPoolExecutor
 from pathlib import Path
 from subprocess import CalledProcessError
-from typing import (
-    Any,
-    Dict,
-    Iterable,
-    List,
-    Optional,
-    Tuple,
-    Union,
-)
+from typing import Any, Dict, Iterable, List, Optional, Tuple, Union
 
 import torchaudio
 import tqdm
@@ -30,8 +21,7 @@
     load_audioset_mapping,
 )
 from aac_datasets.utils.download import download_file
-from aac_datasets.utils.globals import _get_root, _get_ffmpeg_path, _get_ytdlp_path
-
+from aac_datasets.utils.globals import _get_ffmpeg_path, _get_root, _get_ytdlp_path
 
 pylog = logging.getLogger(__name__)
 
@@ -194,19 +184,19 @@ def load_audiocaps_dataset(
 
         fname = _AUDIO_FNAME_FORMAT.format(**line, audio_format=audio_format)
         if fname in fname_to_idx:
-            idx = fname_to_idx[fname]
+            index = fname_to_idx[fname]
 
-            if all_caps_dic["start_time"][idx] is None:
-                all_caps_dic["start_time"][idx] = start_time
-                all_caps_dic["youtube_id"][idx] = youtube_id
-                all_caps_dic["audiocaps_ids"][idx] = [audiocap_id]
-                all_caps_dic["captions"][idx] = [caption]
+            if all_caps_dic["start_time"][index] is None:
+                all_caps_dic["start_time"][index] = start_time
+                all_caps_dic["youtube_id"][index] = youtube_id
+                all_caps_dic["audiocaps_ids"][index] = [audiocap_id]
+                all_caps_dic["captions"][index] = [caption]
             else:
-                assert all_caps_dic["start_time"][idx] == start_time
-                assert all_caps_dic["youtube_id"][idx] == youtube_id
+                assert all_caps_dic["start_time"][index] == start_time
+                assert all_caps_dic["youtube_id"][index] == youtube_id
 
-                all_caps_dic["audiocaps_ids"][idx].append(audiocap_id)
-                all_caps_dic["captions"][idx].append(caption)
+                all_caps_dic["audiocaps_ids"][index].append(audiocap_id)
+                all_caps_dic["captions"][index].append(caption)
 
     # Load tags from audioset data
     all_tags_lst = [[] for _ in range(dataset_size)]
@@ -228,8 +218,8 @@ def load_audiocaps_dataset(
             tags_mid = tags_mid.split(",")
             tags_indexes = [mid_to_index[tag_mid] for tag_mid in tags_mid]
 
-            idx = fname_to_idx[fname]
-            all_tags_lst[idx] = tags_indexes
+            index = fname_to_idx[fname]
+            all_tags_lst[index] = tags_indexes
 
     raw_data = {
         "fname": fnames_lst,
@@ -642,6 +632,15 @@ def _download_from_youtube(
         return False
     _video_link, audio_link = lines[:2]
 
+    # if yt-dlp only returns one link, it is a combined video audio
+    if len(audio_link) == 0:
+        if verbose >= 2:
+            pylog.debug(
+                f"youtube_id={youtube_id} is combined video audio only (cant download)"
+            )
+        # audio_link = video_link # this does not work, not sure why. probably requires changes to ffmpeg command
+        return False
+
     # Download and extract audio from audio_link to fpath_out with ffmpeg
     extract_command = [
         ffmpeg_path,
diff --git a/src/aac_datasets/datasets/functional/clotho.py b/src/aac_datasets/datasets/functional/clotho.py
index be1d80e..77626e0 100644
--- a/src/aac_datasets/datasets/functional/clotho.py
+++ b/src/aac_datasets/datasets/functional/clotho.py
@@ -6,17 +6,8 @@
 import logging
 import os
 import os.path as osp
-
 from pathlib import Path
-from typing import (
-    Any,
-    Dict,
-    Iterable,
-    List,
-    Optional,
-    Tuple,
-    Union,
-)
+from typing import Any, Dict, Iterable, List, Optional, Tuple, Union
 from zipfile import ZipFile
 
 from py7zr import SevenZipFile
@@ -25,7 +16,6 @@
 from aac_datasets.utils.download import download_file, hash_file
 from aac_datasets.utils.globals import _get_root
 
-
 pylog = logging.getLogger(__name__)
 
 
@@ -167,8 +157,8 @@ def load_clotho_dataset(
 
     for line in captions_data:
         fname = line["file_name"]
-        idx = fname_to_idx[fname]
-        all_captions_lst[idx] = [line[caption_key] for caption_key in captions_keys]
+        index = fname_to_idx[fname]
+        all_captions_lst[index] = [line[caption_key] for caption_key in captions_keys]
 
     all_metadata_dic: Dict[str, List[Any]] = {
         key: [None for _ in range(dataset_size)] for key in subset_metadata_keys
@@ -179,11 +169,11 @@ def load_clotho_dataset(
             raise KeyError(
                 f"Cannot find metadata fname={fname} in captions file. (subset={subset})"
             )
-        idx = fname_to_idx[fname]
+        index = fname_to_idx[fname]
         for key in subset_metadata_keys:
             # The test subset does not have keywords in metadata, but has sound_id, sound_link, etc.
             if key in line:
-                all_metadata_dic[key][idx] = line[key]
+                all_metadata_dic[key][index] = line[key]
 
     raw_data = {
         "fname": fnames_lst,
diff --git a/src/aac_datasets/datasets/functional/macs.py b/src/aac_datasets/datasets/functional/macs.py
index 2bb0ac3..7b21768 100644
--- a/src/aac_datasets/datasets/functional/macs.py
+++ b/src/aac_datasets/datasets/functional/macs.py
@@ -7,16 +7,8 @@
 import os.path as osp
 import shutil
 import zipfile
-
 from pathlib import Path
-from typing import (
-    Any,
-    Dict,
-    Iterable,
-    List,
-    Tuple,
-    Union,
-)
+from typing import Any, Dict, Iterable, List, Tuple, Union
 
 import yaml
 
@@ -24,7 +16,6 @@
 from aac_datasets.utils.download import download_file, hash_file
 from aac_datasets.utils.globals import _get_root
 
-
 pylog = logging.getLogger(__name__)
 
 
@@ -149,9 +140,9 @@ def load_macs_dataset(
     for tau_tags in tau_tags_data:
         fname = osp.basename(tau_tags["filename"])
         if fname in fname_to_idx:
-            idx = fname_to_idx[fname]
+            index = fname_to_idx[fname]
             for key in tau_additional_keys:
-                raw_data[key][idx] = tau_tags[key]
+                raw_data[key][index] = tau_tags[key]
 
     # Sanity checks
     assert all(
diff --git a/src/aac_datasets/datasets/functional/wavcaps.py b/src/aac_datasets/datasets/functional/wavcaps.py
index 69ccb8f..7c9f0e9 100644
--- a/src/aac_datasets/datasets/functional/wavcaps.py
+++ b/src/aac_datasets/datasets/functional/wavcaps.py
@@ -8,12 +8,10 @@
 import os.path as osp
 import subprocess
 import zipfile
-
 from pathlib import Path
 from typing import Any, Dict, Iterable, List, Optional, Tuple, Union
 
 import tqdm
-
 from huggingface_hub import snapshot_download
 from huggingface_hub.constants import HUGGINGFACE_HUB_CACHE
 from huggingface_hub.utils.tqdm import (
@@ -25,22 +23,21 @@
 
 from aac_datasets.datasets.functional.common import DatasetCard
 from aac_datasets.utils.collections import list_dict_to_dict_list
-from aac_datasets.utils.download import safe_rmdir
+from aac_datasets.utils.download import download_file, safe_rmdir
 from aac_datasets.utils.globals import _get_root, _get_zip_path
 
-
 pylog = logging.getLogger(__name__)
 
 
 class WavCapsCard(DatasetCard):
     ANNOTATIONS_CREATORS: Tuple[str, ...] = ("machine-generated",)
     CAPTIONS_PER_AUDIO: Dict[str, int] = {
-        "as": 1,
+        "audioset": 1,
         "bbc": 1,
-        "fsd": 1,
-        "sb": 1,
-        "as_noac": 1,
-        "fsd_nocl": 1,
+        "freesound": 1,
+        "soundbible": 1,
+        "audioset_no_audiocaps": 1,
+        "freesound_no_clotho": 1,
     }
     CITATION: str = r"""
     @article{mei2023WavCaps,
@@ -52,7 +49,7 @@ class WavCapsCard(DatasetCard):
     }
     """
     DEFAULT_REVISION: str = "85a0c21e26fa7696a5a74ce54fada99a9b43c6de"
-    DEFAULT_SUBSET: str = "as_noac"
+    DEFAULT_SUBSET: str = "audioset_no_audiocaps"
     DESCRIPTION: str = "WavCaps: A ChatGPT-Assisted Weakly-Labelled Audio Captioning Dataset for Audio-Language Multimodal Research."
     EXPECTED_SIZES: Dict[str, int] = {
         "AudioSet_SL": 108317,
@@ -87,7 +84,7 @@ def load_wavcaps_dataset(
     :param root: Dataset root directory.
         defaults to ".".
     :param subset: The subset of MACS to use. Can be one of :attr:`~MACSCard.SUBSETS`.
-        defaults to "as_noac".
+        defaults to "audioset_no_audiocaps".
     :param verbose: Verbose level.
         defaults to 0.
 
@@ -97,13 +94,22 @@ def load_wavcaps_dataset(
         defaults to None.
     :returns: A dictionnary of lists containing each metadata.
     """
+
+    if subset in _WAVCAPS_OLD_SUBSETS_NAMES:
+        new_subset = _WAVCAPS_OLD_SUBSETS_NAMES[subset]
+        if verbose >= 0:
+            pylog.warning(
+                f"Deprecated subset name '{subset}', use '{new_subset}' instead."
+            )
+        subset = new_subset
+
     root = _get_root(root)
     if subset not in WavCapsCard.SUBSETS:
         raise ValueError(
             f"Invalid argument subset={subset}. (expected one of {WavCapsCard.SUBSETS})"
         )
 
-    if subset == "as":
+    if subset == "audioset":
         overlapped_ds = "AudioCaps"
         overlapped_subsets = ("val", "test")
         pylog.warning(
@@ -111,7 +117,7 @@ def load_wavcaps_dataset(
             "You can use as_noac subset for to avoid this bias with AudioCaps."
         )
 
-    elif subset == "fsd":
+    elif subset == "freesound":
         overlapped_ds = "Clotho"
         overlapped_subsets = (
             "val",
@@ -126,14 +132,14 @@ def load_wavcaps_dataset(
             f"You can use fsd_nocl subset for to avoid this bias for Clotho val, eval, dcase_t2a_audio and dcase_t2a_captions subsets. Data could still overlap with Clotho dcase_aac_test and dcase_aac_analysis subsets."
         )
 
-    if subset in ("as_noac", "fsd_nocl"):
-        if subset == "as_noac":
-            target_subset = "as"
-            csv_fname = "blacklist_audiocaps.full.csv"
+    if subset in ("audioset_no_audiocaps", "freesound_no_clotho"):
+        if subset == "audioset_no_audiocaps":
+            target_subset = "audioset"
+            csv_fname = _WAVCAPS_LINKS["blacklist_audiocaps"]["fname"]
 
-        elif subset == "fsd_nocl":
-            target_subset = "fsd"
-            csv_fname = "blacklist_clotho.full.csv"
+        elif subset == "freesound_no_clotho":
+            target_subset = "freesound"
+            csv_fname = _WAVCAPS_LINKS["blacklist_clotho"]["fname"]
 
         else:
             raise ValueError(f"INTERNAL ERROR: Invalid argument subset={subset}.")
@@ -147,14 +153,13 @@ def load_wavcaps_dataset(
         )
         wavcaps_ids = raw_data["id"]
 
-        csv_fpath = (
-            Path(__file__)
-            .parent.parent.parent.parent.joinpath("data")
-            .joinpath(csv_fname)
-        )
+        wavcaps_root = _get_wavcaps_root(root, hf_cache_dir, revision)
+        csv_fpath = Path(wavcaps_root).joinpath(csv_fname)
+
         with open(csv_fpath, "r") as file:
             reader = csv.DictReader(file)
             data = list(reader)
+
         other_ids = [data_i["id"] for data_i in data]
         other_ids = dict.fromkeys(other_ids)
 
@@ -166,7 +171,7 @@ def load_wavcaps_dataset(
             )
 
         raw_data = {
-            column: [column_data[idx] for idx in indexes]
+            column: [column_data[index] for index in indexes]
             for column, column_data in raw_data.items()
         }
         return raw_data
@@ -270,7 +275,7 @@ def download_wavcaps_dataset(
     :param root: Dataset root directory.
         defaults to ".".
     :param subset: The subset of MACS to use. Can be one of :attr:`~WavCapsCard.SUBSETS`.
-        defaults to "as_noac".
+        defaults to "audioset_no_audiocaps".
     :param force: If True, force to download again all files.
         defaults to False.
     :param verbose: Verbose level.
@@ -289,10 +294,24 @@ def download_wavcaps_dataset(
     :param zip_path: Path to zip executable path in shell.
         defaults to "zip".
     """
-    if subset == "as_noac":
+
+    if subset in _WAVCAPS_OLD_SUBSETS_NAMES:
+        new_subset = _WAVCAPS_OLD_SUBSETS_NAMES[subset]
+        if verbose >= 0:
+            pylog.warning(
+                f"Deprecated subset name '{subset}', use '{new_subset}' instead."
+            )
+        subset = new_subset
+
+    root = _get_root(root)
+    zip_path = _get_zip_path(zip_path)
+
+    if subset == "audioset_no_audiocaps":
+        _download_blacklist(root, hf_cache_dir, revision, "blacklist_audiocaps")
+
         return download_wavcaps_dataset(
             root=root,
-            subset="as",
+            subset="audioset",
             revision=revision,
             hf_cache_dir=hf_cache_dir,
             force=force,
@@ -301,10 +320,13 @@ def download_wavcaps_dataset(
             zip_path=zip_path,
             verbose=verbose,
         )
-    elif subset == "fsd_nocl":
+
+    elif subset == "freesound_no_clotho":
+        _download_blacklist(root, hf_cache_dir, revision, "blacklist_clotho")
+
         return download_wavcaps_dataset(
             root=root,
-            subset="fsd",
+            subset="freesound",
             revision=revision,
             hf_cache_dir=hf_cache_dir,
             force=force,
@@ -314,9 +336,6 @@ def download_wavcaps_dataset(
             verbose=verbose,
         )
 
-    root = _get_root(root)
-    zip_path = _get_zip_path(zip_path)
-
     if subset not in WavCapsCard.SUBSETS:
         raise ValueError(
             f"Invalid argument subset={subset}. (expected one of {WavCapsCard.SUBSETS})"
@@ -340,7 +359,7 @@ def download_wavcaps_dataset(
     ign_patterns = [
         pattern
         for source in ign_sources
-        for pattern in (f"json_files/{source}/*.json", "Zip_files/*")  # {source}/
+        for pattern in (f"json_files/{source}/*.json", f"Zip_files/{source}/*")
     ]
     if verbose >= 2:
         pylog.debug(f"ign_sources={ign_sources}")
@@ -373,7 +392,7 @@ def download_wavcaps_dataset(
     del snapshot_dpath
 
     # Build symlink to hf cache
-    if osp.exists(wavcaps_root):
+    if osp.lexists(wavcaps_root):
         if not osp.islink(wavcaps_root):
             raise RuntimeError("WavCaps root exists but it is not a symlink.")
         link_target_abspath = osp.abspath(osp.realpath(wavcaps_root))
@@ -626,10 +645,10 @@ def _is_prepared_wavcaps(
 def _use_source(source: str, subset: str) -> bool:
     return any(
         (
-            source == "AudioSet_SL" and subset in ("as", "as_noac"),
+            source == "AudioSet_SL" and subset in ("audioset", "audioset_no_audiocaps"),
             source == "BBC_Sound_Effects" and subset in ("bbc",),
-            source == "FreeSound" and subset in ("fsd", "fsd_nocl"),
-            source == "SoundBible" and subset in ("sb",),
+            source == "FreeSound" and subset in ("freesound", "freesound_no_clotho"),
+            source == "SoundBible" and subset in ("soundbible",),
         )
     )
 
@@ -643,6 +662,21 @@ def _load_json(fpath: str) -> Tuple[Dict[str, Any], int]:
     return data, size
 
 
+def _download_blacklist(
+    root: str,
+    hf_cache_dir: Optional[str],
+    revision: Optional[str],
+    name: str,
+    verbose: int = 0,
+) -> None:
+    info = _WAVCAPS_LINKS[name]
+    fname = info["fname"]
+    url = info["url"]
+    wavcaps_root = _get_wavcaps_root(root, hf_cache_dir, revision)
+    fpath = Path(wavcaps_root).joinpath(fname)
+    download_file(url, fpath, verbose=verbose)
+
+
 class _WavCapsRawItem(TypedDict):
     # Common values
     caption: str
@@ -685,3 +719,22 @@ class _WavCapsRawItem(TypedDict):
     "FreeSound": "FreeSound",
     "SoundBible": "SoundBible",
 }
+
+_WAVCAPS_LINKS = {
+    "blacklist_audiocaps": {
+        "url": "https://raw.githubusercontent.com/Labbeti/aac-datasets/main/data/wavcaps/blacklist_audiocaps.full.csv",
+        "fname": "blacklist_audiocaps.full.csv",
+    },
+    "blacklist_clotho": {
+        "url": "https://raw.githubusercontent.com/Labbeti/aac-datasets/main/data/wavcaps/blacklist_clotho.full.csv",
+        "fname": "blacklist_clotho.full.csv",
+    },
+}
+
+_WAVCAPS_OLD_SUBSETS_NAMES = {
+    "fsd": "freesound",
+    "as": "audioset",
+    "fsd_nocl": "freesound_no_clotho",
+    "as_noac": "audioset_no_audiocaps",
+    "sb": "soundbible",
+}
diff --git a/src/aac_datasets/datasets/macs.py b/src/aac_datasets/datasets/macs.py
index 44eaaf0..ce97e15 100644
--- a/src/aac_datasets/datasets/macs.py
+++ b/src/aac_datasets/datasets/macs.py
@@ -4,16 +4,8 @@
 import copy
 import logging
 import os.path as osp
-
 from pathlib import Path
-from typing import (
-    Callable,
-    ClassVar,
-    Dict,
-    List,
-    Optional,
-    Union,
-)
+from typing import Any, Callable, ClassVar, Dict, List, Optional, Union
 
 from torch import Tensor
 from typing_extensions import TypedDict
@@ -21,13 +13,12 @@
 from aac_datasets.datasets.base import AACDataset
 from aac_datasets.datasets.functional.macs import (
     MACSCard,
-    load_macs_dataset,
-    download_macs_dataset,
     _get_audio_dpath,
+    download_macs_dataset,
+    load_macs_dataset,
 )
 from aac_datasets.utils.globals import _get_root
 
-
 pylog = logging.getLogger(__name__)
 
 
@@ -80,7 +71,7 @@ def __init__(
         root: Union[str, Path, None] = None,
         subset: str = MACSCard.DEFAULT_SUBSET,
         download: bool = False,
-        transform: Optional[Callable] = None,
+        transform: Optional[Callable[[MACSItem], Any]] = None,
         verbose: int = 0,
         force_download: bool = False,
         verify_files: bool = False,
@@ -199,8 +190,8 @@ def get_competence(self, annotator_id: int) -> float:
         """Get competence value for a specific annotator id."""
         return self._annotator_id_to_competence[annotator_id]
 
-    def _load_competences(self, idx: int) -> List[float]:
-        annotators_ids: List[int] = self.at(idx, "annotators_ids")
+    def _load_competences(self, index: int) -> List[float]:
+        annotators_ids: List[int] = self.at(index, "annotators_ids")
         competences = [self.get_competence(id_) for id_ in annotators_ids]
         return competences
 
diff --git a/src/aac_datasets/datasets/wavcaps.py b/src/aac_datasets/datasets/wavcaps.py
index 00e6fe9..ab0be7e 100644
--- a/src/aac_datasets/datasets/wavcaps.py
+++ b/src/aac_datasets/datasets/wavcaps.py
@@ -3,9 +3,8 @@
 
 import logging
 import os.path as osp
-
 from pathlib import Path
-from typing import Callable, ClassVar, List, Optional, Union
+from typing import Any, Callable, ClassVar, List, Optional, Union
 
 from torch import Tensor
 from typing_extensions import TypedDict
@@ -13,13 +12,12 @@
 from aac_datasets.datasets.base import AACDataset
 from aac_datasets.datasets.functional.wavcaps import (
     WavCapsCard,
-    load_wavcaps_dataset,
-    download_wavcaps_dataset,
     _get_audio_subset_dpath,
+    download_wavcaps_dataset,
+    load_wavcaps_dataset,
 )
 from aac_datasets.utils.globals import _get_root, _get_zip_path
 
-
 pylog = logging.getLogger(__name__)
 
 
@@ -50,12 +48,12 @@ class WavCaps(AACDataset[WavCapsItem]):
     HuggingFace source : https://huggingface.co/datasets/cvssp/WavCaps
 
     This dataset contains 4 training subsets, extracted from different sources:
-    - AudioSet strongly labeled (as)
-    - BBC Sound Effects (bbc)
-    - FreeSound (fsd)
-    - SoundBible (sb)
-    - AudioSet strongly labeled without AudioCaps (as_noac)
-    - FreeSound without Clotho (fsd_nocl)
+    - AudioSet strongly labeled ("audioset")
+    - BBC Sound Effects ("bbc")
+    - FreeSound ("freesound")
+    - SoundBible ("soundbible")
+    - AudioSet strongly labeled without AudioCaps ("audioset_no_audiocaps")
+    - FreeSound without Clotho ("freesound_no_clotho")
 
     .. warning::
         WavCaps download is experimental ; it requires a lot of disk space and can take very long time to download and extract, so you might expect errors.
@@ -110,7 +108,7 @@ def __init__(
         root: Union[str, Path, None] = None,
         subset: str = WavCapsCard.DEFAULT_SUBSET,
         download: bool = False,
-        transform: Optional[Callable] = None,
+        transform: Optional[Callable[[WavCapsItem], Any]] = None,
         verbose: int = 0,
         force_download: bool = False,
         verify_files: bool = False,
@@ -127,7 +125,7 @@ def __init__(
             The data will be stored in the 'MACS' subdirectory.
             defaults to ".".
         :param subset: The subset of the dataset. Can be one of :attr:`~WavCapsCard.SUBSETS`.
-            defaults to "as".
+            defaults to "audioset".
         :param download: Download the dataset if download=True and if the dataset is not already downloaded.
             defaults to False.
         :param transform: The transform to apply to the global dict item. This transform is applied only in getitem method when argument is an integer.
@@ -177,7 +175,6 @@ def __init__(
             subset=subset,
             verbose=verbose,
             hf_cache_dir=hf_cache_dir,
-            repo_id=repo_id,
             revision=revision,
         )
 
diff --git a/src/aac_datasets/utils/collections.py b/src/aac_datasets/utils/collections.py
index 9309454..fd8300b 100644
--- a/src/aac_datasets/utils/collections.py
+++ b/src/aac_datasets/utils/collections.py
@@ -1,14 +1,13 @@
 #!/usr/bin/env python
 # -*- coding: utf-8 -*-
 
-from typing import Any, Dict, Iterable, List, TypeVar
-
+from typing import Any, Dict, Iterable, List, Mapping, Sequence, TypeVar
 
 T = TypeVar("T")
 
 
 def list_dict_to_dict_list(
-    lst: List[Dict[str, T]],
+    lst: Sequence[Mapping[str, T]],
     key_mode: str = "intersect",
     default: Any = None,
 ) -> Dict[str, List[T]]:
@@ -41,7 +40,7 @@ def list_dict_to_dict_list(
     return {key: [item.get(key, default) for item in lst] for key in keys}
 
 
-def intersect_lists(lst_of_lst: List[Iterable[T]]) -> List[T]:
+def intersect_lists(lst_of_lst: Sequence[Iterable[T]]) -> List[T]:
     """Performs intersection of elements in lists (like set intersection), but keep their original order."""
     if len(lst_of_lst) <= 0:
         return []
diff --git a/src/aac_datasets/utils/download.py b/src/aac_datasets/utils/download.py
index a134c3c..df49e32 100644
--- a/src/aac_datasets/utils/download.py
+++ b/src/aac_datasets/utils/download.py
@@ -4,13 +4,12 @@
 import hashlib
 import os
 import os.path as osp
-
 from pathlib import Path
 from typing import List, Union
 
+import tqdm
 from torch.hub import download_url_to_file
 
-
 HASH_TYPES = ("sha256", "md5")
 DEFAULT_CHUNK_SIZE = 256 * 1024**2  # 256 MiB
 
@@ -32,24 +31,38 @@ def safe_rmdir(
     root: Union[str, Path],
     rm_root: bool = True,
     error_on_non_empty_dir: bool = True,
+    followlinks: bool = False,
+    verbose: int = 0,
 ) -> List[str]:
     """Remove all empty sub-directories.
 
     :param root: Root directory path.
     :param rm_root: If True, remove the root directory too. defaults to True.
     :param error_on_non_empty_dir: If True, raises a RuntimeError if a subdirectory contains at least 1 file. Otherwise it will leave non-empty directories. defaults to True.
+    :param followlinks: Indicates whether or not symbolic links shound be followed. defaults to False.
+    :param verbose: Verbose level. defaults to 0.
     :returns: The list of directories paths deleted.
     """
-    deleted = []
-    for dpath, dnames, fnames in os.walk(root, topdown=False):
+    to_delete = []
+
+    for dpath, dnames, fnames in tqdm.tqdm(
+        os.walk(root, topdown=False, followlinks=followlinks),
+        disable=verbose < 2,
+    ):
         if not rm_root and dpath == root:
             continue
-        elif len(dnames) == 0 and len(fnames) == 0:
-            os.rmdir(dpath)
-            deleted.append(dpath)
+        elif len(fnames) == 0 and (
+            len(dnames) == 0
+            or all(osp.join(dpath, dname) in to_delete for dname in dnames)
+        ):
+            to_delete.append(dpath)
         elif error_on_non_empty_dir:
-            raise RuntimeError(f"Cannot remove non-empty dir {dpath}.")
-    return deleted
+            raise RuntimeError(f"Cannot remove non-empty directory '{dpath}'.")
+
+    for dpath in to_delete:
+        os.rmdir(dpath)
+
+    return to_delete
 
 
 def hash_file(
diff --git a/tests/test_datasets_base.py b/tests/test_datasets_base.py
index eefc24c..75d8634 100644
--- a/tests/test_datasets_base.py
+++ b/tests/test_datasets_base.py
@@ -2,7 +2,6 @@
 # -*- coding: utf-8 -*-
 
 import unittest
-
 from unittest import TestCase
 
 from aac_datasets.datasets.base import _flat_raw_data, _unflat_raw_data
@@ -12,11 +11,11 @@ class TestDatasetBase(TestCase):
     def test_flat_raw_data(self) -> None:
         raw_data = {
             "captions": [["a1", "a2", "a3"], ["b1"], ["c1", "c2"], []],
-            "idx": list(range(1, 5)),
+            "index": list(range(1, 5)),
         }
         expected_flat = {
             "captions": [["a1"], ["a2"], ["a3"], ["b1"], ["c1"], ["c2"], []],
-            "idx": [1, 1, 1, 2, 3, 3, 4],
+            "index": [1, 1, 1, 2, 3, 3, 4],
         }
         expected_sizes = [3, 1, 2, 0]