Add: Audio duration is now returned by default for each item.

Labbeti · Mar 21, 2024 · 10367cb · 10367cb
1 parent 40f6318
commit 10367cb
Show file tree

Hide file tree

Showing 4 changed files with 6 additions and 0 deletions.
diff --git a/src/aac_datasets/datasets/audiocaps.py b/src/aac_datasets/datasets/audiocaps.py
@@ -40,6 +40,7 @@ class AudioCapsItem(TypedDict):
     index: int
     subset: str
     sr: int
+    duration: float
     # AudioCaps-specific attributes
     audiocaps_ids: List[int]
     start_time: int

diff --git a/src/aac_datasets/datasets/clotho.py b/src/aac_datasets/datasets/clotho.py
@@ -32,6 +32,7 @@ class ClothoItem(TypedDict):
     index: int
     subset: str
     sr: NotRequired[int]
+    duration: NotRequired[float]
     # Clotho-specific attributes
     keywords: NotRequired[List[str]]
     sound_id: NotRequired[str]  # warning: some files contains "Not found"
@@ -185,6 +186,7 @@ def __init__(
             removed_columns = (
                 "audio",
                 "sr",
+                "duration",
                 "fname",
                 "sound_id",
                 "keywords",
@@ -195,6 +197,7 @@ def __init__(
             )
         else:
             removed_columns = ()
+
         for name in removed_columns:
             column_names.remove(name)
 

diff --git a/src/aac_datasets/datasets/macs.py b/src/aac_datasets/datasets/macs.py
@@ -33,6 +33,7 @@ class MACSItem(TypedDict):
     index: int
     subset: str
     sr: int
+    duration: float
     # MACS-specific attributes
     annotators_ids: List[str]
     competences: List[float]

diff --git a/src/aac_datasets/datasets/wavcaps.py b/src/aac_datasets/datasets/wavcaps.py
@@ -30,6 +30,7 @@ class WavCapsItem(TypedDict):
     index: int
     subset: str
     sr: int
+    duration: float
     # WavCaps-specific attributes
     author: Optional[str]  # FSD and SB
     description: Optional[str]  # BBC, FSD and SB only