Upload folder using huggingface_hub

README.md

@@ -1,6 +1,5 @@
 # Discrete Speech Units Dataset
 
-        This dataset contains discrete speech unit representations with various augmentations.
+This dataset contains discrete speech unit representations with various augmentations.
 
-        Columns: id, tokens, text, augmentation
-        
+Columns: id, tokens, text, augmentation

dataset_card.yaml

@@ -1,16 +1,15 @@
 ---
-                    language:
-                    - en
-                    license: cc-by-4.0
-                    task_categories:
-                    - speech-processing
-                    task_ids:
-                    - speech-representation
-                    size_categories:
-                    - 100K<n<1M
-                    tags:
-                    - speech
-                    - discrete-units
-                    - augmentation
-                    ---
-                    
+language:
+- en
+license: cc-by-4.0
+task_categories:
+- speech-processing
+task_ids:
+- speech-representation
+size_categories:
+- 100K<n<1M
+tags:
+- speech
+- discrete-units
+- augmentation
+---

dsu_dataset.py

@@ -0,0 +1,38 @@
+
+import json
+import datasets
+
+_DESCRIPTION = "Discrete Speech Units Dataset"
+
+class DsuDataset(datasets.GeneratorBasedBuilder):
+    DEFAULT_CONFIG_NAME = "default"
+    
+    def _info(self):
+        return datasets.DatasetInfo(
+            description=_DESCRIPTION,
+            features=datasets.Features({
+                "id": datasets.Value("string"),
+                "tokens": datasets.Sequence(datasets.Value("int32")),
+                "text": datasets.Value("string"),
+                "augmentation": datasets.Value("string")
+            })
+        )
+
+    def _split_generators(self, dl_manager):
+        return [
+            datasets.SplitGenerator(
+                name=datasets.Split.TRAIN,
+                gen_kwargs={"filepath": "train.json"},
+            )
+        ]
+
+    def _generate_examples(self, filepath):
+        with open(filepath, "r", encoding="utf-8") as f:
+            data = json.load(f)
+            for i, example in enumerate(data):
+                yield i, {
+                    "id": example["id"],
+                    "tokens": example["tokens"],
+                    "text": example["text"],
+                    "augmentation": example["augmentation"],
+                }