Add VoxCelebSA dataset for MTEB audio classification evaluation

README.md

@@ -0,0 +1,23 @@
+---
+dataset_info:
+  features:
+  - name: audio
+    dtype: audio
+  - name: file
+    dtype: string
+  - name: instruction
+    dtype: string
+  - name: label
+    dtype: string
+  splits:
+  - name: test
+    num_bytes: 544602881.0
+    num_examples: 3553
+  download_size: 520664380
+  dataset_size: 544602881.0
+configs:
+- config_name: default
+  data_files:
+  - split: test
+    path: data/test-*
+---

data/test-00000-of-00002.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:58bf3497be18f96347227aef67d9f7fa883b7851178ec27434e0985b910f1aba
+size 269631512

data/test-00001-of-00002.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4b772d179113bf7c61e4a9ca471934e661e37b4d35285f38167a6224ab9360e7
+size 251032868