Per-language subsets: hindi, urdu, bengali, other

README.md

@@ -44,11 +44,27 @@ tags:
 - consented-data
 - human-transcription-on-demand
 configs:
-- config_name: default
+- config_name: all
   default: true
   data_files:
   - split: test
-    path: data/test-*
+    path: data/*/test-*
+- config_name: hindi
+  data_files:
+  - split: test
+    path: data/hindi/test-*
+- config_name: urdu
+  data_files:
+  - split: test
+    path: data/urdu/test-*
+- config_name: bengali
+  data_files:
+  - split: test
+    path: data/bengali/test-*
+- config_name: other
+  data_files:
+  - split: test
+    path: data/other/test-*
 ---
 
 # Indic Spontaneous Speech — Silencio
@@ -81,11 +97,12 @@ Contributors answer an open prompt in their own words, on their own devices, whe
 ```python
 from datasets import load_dataset
 
+# everything (default)
 ds = load_dataset("SilencioNetwork/indic-languages-speech", split="test")
 print(ds[0]["language"], ds[0]["dialect"], ds[0]["country"], ds[0]["proficiency"])
 
-# filter to one language
-hindi = ds.filter(lambda x: x["language"] == "Hindi")
+# or one language at a time: "hindi", "urdu", "bengali", "other"
+hindi = load_dataset("SilencioNetwork/indic-languages-speech", "hindi", split="test")
 
 # datasets v4 returns a torchcodec AudioDecoder:
 sample = ds[0]["audio"].get_all_samples()
@@ -160,9 +177,22 @@ Requires `pip install "datasets>=4.0"` and FFmpeg ≥ 4.
 | Mobile | 136 | 84.5% |
 | Desktop | 25 | 15.5% |
 
+## Language subsets
+
+The release ships as named subsets, so you can pull one language without downloading the rest.
+The preview at the top of this page has a dropdown for them.
+
+| Subset | Contents | Clips |
+|---|---|---:|
+| `all` (default) | every clip | 161 |
+| `hindi` | Hindi | 61 |
+| `urdu` | Urdu | 55 |
+| `bengali` | Bengali | 39 |
+| `other` | Marathi, Nepali, Sindhi, Gujarati | 6 |
+
 ## Splits
 
-Single split, `test`, 161 rows. No train/dev/test partition is provided: at this scale a
+Single split, `test`, in every subset; 161 rows in `all`. No train/dev/test partition is provided: at this scale a
 partition would leave each part too small to be meaningful. **Every clip is from a different
 speaker**, so any split you construct is speaker-disjoint by construction, with no speaker
 leakage to control for.

data/{test-00000-of-00006.parquet → bengali/test-00000-of-00002.parquet}

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:348b1303a20b281a9b8409624c89a1d327d4d106ca83fe43f32dd9089c6dd84d
-size 198091991
+oid sha256:28ad7380f761857331947ab3f173575196e13d5a5fa71f57ae1032f02a9ac16c
+size 207991685

data/{test-00001-of-00006.parquet → bengali/test-00001-of-00002.parquet}

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7f6428a1a91900dbf667b74c63fe840dcd464c693366fa3d2fe37962d9605cee
-size 199312896
+oid sha256:a9e118b40735e6ac459b877b4ac153ca0d4d0cb7e3daeac71f05d622f1f3ade3
+size 99509287

data/{test-00002-of-00006.parquet → hindi/test-00000-of-00002.parquet}

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a16324a7360c16a94de49929bb47d4dde79809b68c564cecc36101eb7e4becbe
-size 192245246
+oid sha256:2c4d74b75d5b5b2f55e9eaad2b00ba18b6ded36fe013303c62c2b872b67687d8
+size 198091987

data/{test-00003-of-00006.parquet → hindi/test-00001-of-00002.parquet}

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ce716cb01b284168f4276c23aedd276679a808bf37aa83ca27e106923db4e679
-size 197664619
+oid sha256:df13998dd8f6c284b72d18f7fd07bf267007adba5342bb3f48ec24af8fa6e0c2
+size 199312892

data/other/test-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b8f5da9415fd8bc86bc07a6e0428a5a991d880621598eb43f04977d44d323ffc
+size 42152067

data/test-00004-of-00006.parquet

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:c3f095510cc7d043705b4ea99410852d00ca99a71001243afcf03cbde64bb6a3
-size 207505883

data/test-00005-of-00006.parquet

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:764580332dec46acd57809b3922d4eca2887c8ce299b286049c7b309c02475de
-size 112459865

data/urdu/test-00000-of-00002.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c693976f5cd9da78e9faf8fe6654a07dbdb87b5d17f178a950552b8d5dfb9e58
+size 192245242

data/urdu/test-00001-of-00002.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d8c30e4e5a6e9c00c26c81ab22155065067063393227ea9593ae264aed83b466
+size 168116940