Upload dataset

README.md

@@ -1,14 +1,24 @@
 ---
-annotations_creators: ['no-annotation']
-language: ['en']
-license: ['mit']
-task_categories: ['audio-classification', 'anomaly-detection']
-tags: ['dcase', 'audio', 'anomaly-detection']
+annotations_creators:
+- no-annotation
+language:
+- en
+license:
+- mit
+task_categories:
+- audio-classification
+- anomaly-detection
+tags:
+- dcase
+- audio
+- anomaly-detection
 pretty_name: DCASE 2025 Task 2 Dev
 dataset_info:
   features:
-  - name: audio
-    dtype: audio
+  - name: file_id
+    dtype: string
+  - name: split
+    dtype: string
   - name: machine_type
     dtype:
       class_label:
@@ -34,6 +44,8 @@ dataset_info:
           '0': source
           '1': target
           '2': unknown
+  - name: section
+    dtype: string
   - name: d1p
     dtype:
       class_label:
@@ -46,9 +58,9 @@ dataset_info:
     dtype:
       class_label:
         names:
-          '0': 0
-          '1': 4
-          '2': 5
+          '0': '0'
+          '1': '4'
+          '2': '5'
           '3': A
           '4': B
           '5': C
@@ -70,8 +82,8 @@ dataset_info:
     dtype:
       class_label:
         names:
-          '0': 0
-          '1': 5
+          '0': '0'
+          '1': '5'
           '2': none
   - name: d3p
     dtype:
@@ -83,6 +95,33 @@ dataset_info:
       class_label:
         names:
           '0': none
+  - name: audio_path
+    dtype: string
+  - name: audio
+    dtype:
+      audio:
+        sampling_rate: 16000
+  splits:
+  - name: train
+    num_bytes: 1601921625.0
+    num_examples: 5000
+  - name: test
+    num_bytes: 320384725.0
+    num_examples: 1000
+  - name: supplemental
+    num_bytes: 237068088.0
+    num_examples: 700
+  download_size: 2134670161
+  dataset_size: 2159374438.0
+configs:
+- config_name: default
+  data_files:
+  - split: train
+    path: data/train-*
+  - split: test
+    path: data/test-*
+  - split: supplemental
+    path: data/supplemental-*
 ---
 # DCASE 2025 Task 2 - Development Dataset
 Attributes `d1v`, `d2v`, `d3v` are encoded as **ClassLabels** for training convenience.