Mod: AC do not output 'tags' value when with_tags=False. Path managem…

…ent now be used with aac_datasets module.

src/aac_datasets/__init__.py

@@ -9,6 +9,14 @@
 from .datasets.clotho import Clotho
 from .datasets.macs import MACS
 from .datasets.wavcaps import WavCaps
+from .utils.paths import (
+    get_default_ffmpeg_path,
+    get_default_root,
+    get_default_ytdl_path,
+    set_default_ffmpeg_path,
+    set_default_root,
+    set_default_ytdl_path,
+)
 
 
 __author__ = "Etienne Labbé (Labbeti)"

src/aac_datasets/datasets/audiocaps.py

@@ -26,7 +26,7 @@
 from torch import Tensor
 from torch.hub import download_url_to_file
 from torchaudio.backend.common import AudioMetaData
-from typing_extensions import TypedDict
+from typing_extensions import TypedDict, NotRequired
 
 from aac_datasets.datasets.base import AACDataset, DatasetCard
 from aac_datasets.utils.paths import _get_root, _get_ffmpeg_path, _get_ytdl_path
@@ -49,7 +49,7 @@ class AudioCapsItem(TypedDict, total=True):
     # AudioCaps-specific attributes
     audiocaps_ids: List[int]
     start_time: int
-    tags: List[int]
+    tags: NotRequired[List[int]]
     youtube_id: str
 
 
@@ -211,10 +211,16 @@ def __init__(
         ]
         raw_data["index"] = list(range(size))
 
+        column_names = list(AudioCapsItem.__required_keys__) + list(
+            AudioCapsItem.__optional_keys__
+        )
+        if not with_tags:
+            column_names.remove("tags")
+
         super().__init__(
             raw_data=raw_data,
             transform=transform,
-            column_names=AudioCapsItem.__required_keys__,
+            column_names=column_names,
             flat_captions=flat_captions,
             sr=sr,
             verbose=verbose,