Upload README.md with huggingface_hub

README.md

@@ -0,0 +1,42 @@
+---
+dataset_info:
+  features:
+    - name: split
+      dtype: string
+    - name: dataset
+      dtype: string
+    - name: source
+      dtype: string
+    - name: clip_id
+      dtype: string
+    - name: sample_id
+      dtype: string
+    - name: caption
+      dtype: string
+    - name: clap_score
+      dtype: float32
+    - name: sampling_rate
+      dtype: int32
+    - name: channels
+      dtype: int32
+    - name: duration
+      dtype: float32
+    - name: path
+      dtype: string
+    - name: audio
+      dtype: binary
+    - name: active_duration
+      dtype: float32
+    - name: static_params
+      dtype: string
+    - name: moving_params
+      dtype: string
+---
+
+# PhysWave Processed Audio
+
+Processed PhysWave training data exported as Parquet shards. Each row contains
+one 10-second 16 kHz mono WAV clip in `audio`, the original text metadata, CLAP
+score, and spatial parameters.
+
+`static_params` and `moving_params` are JSON strings.