Drop en_za config + sw_tz legacy 42 shards (empty transcripts, legacy schema). sw_tz: 1002.9->21.7h. Total: 55008.3->53884.2h.

README.md

@@ -306,12 +306,6 @@ configs:
     path: data/en_sl/*train-*.parquet
   - split: test
     path: data/en_sl/*test-*.parquet
-- config_name: en_za
-  data_files:
-  - split: train
-    path: data/en_za/*train-*.parquet
-  - split: test
-    path: data/en_za/*test-*.parquet
 - config_name: en_zm
   data_files:
   - split: train
@@ -763,13 +757,9 @@ configs:
 - config_name: sw_tz
   data_files:
   - split: train
-    path:
-    - data/sw_tz/train-*.parquet
-    - data/sw_tz/tz_sw_znz-train-*.parquet
+    path: data/sw_tz/tz_sw_znz-train-*.parquet
   - split: test
-    path:
-    - data/sw_tz/test-*.parquet
-    - data/sw_tz/tz_sw_znz-test-*.parquet
+    path: data/sw_tz/tz_sw_znz-test-*.parquet
 - config_name: ta_in
   data_files:
   - split: train
@@ -893,7 +883,7 @@ A multilingual ASR dataset containing 53k hours of transcribed speech across 113
 
 | Metric | Value |
 | :--- | :--- |
-| Total hours | 55,008 |
+| Total hours | 53,884 |
 | Distinct languages | 88 |
 | Language variations | 127 |
 | Languages ≥ 1,000 h | 23 |
@@ -995,7 +985,6 @@ for row in ds:
 | `en_ke` | Kenya | English | 170.0 | 2.97 |
 | `en_nz` | New Zealand | English | 435.7 | 2.68 |
 | `en_sl` | Sierra Leone | English | 102.4 | 2.70 |
-| `en_za` | South Africa | English | 142.9 | 2.95 |
 | `en_zm` | Zambia | English | 291.6 | 2.70 |
 | `eo` | (international) | Esperanto | 15.0 | 3.27 |
 | `es_ar` | Argentina | Spanish | 251.7 | 2.78 |
@@ -1068,7 +1057,7 @@ for row in ds:
 | `st_za` | South Africa | Sesotho | 0.4 | 2.67 |
 | `sv_ax` | Åland Islands | Swedish | 66.0 | 2.73 |
 | `sw_ke` | Kenya | Swahili | 257.4 | 2.93 |
-| `sw_tz` | Tanzania | Swahili | 1,002.9 | 2.56 |
+| `sw_tz` | Tanzania | Swahili | 21.7 | 2.56 |
 | `ta_in` | India | Tamil | 36.1 | 3.06 |
 | `ta_lk` | Sri Lanka | Tamil | 204.0 | 2.80 |
 | `te_in` | India | Telugu | 77.0 | 3.13 |
@@ -1087,7 +1076,7 @@ for row in ds:
 | `yue_hk` | Hong Kong | Cantonese | 1,943.5 | 3.02 |
 | `zh_tw` | Taiwan | Mandarin | 1,482.1 | 2.47 |
 | `zu_za` | South Africa | Zulu | 19.0 | 2.71 |
-| **Total** | — | — | **55,008.3** | **2.83** |
+| **Total** | — | — | **53,884.2** | **2.83** |
 
 ⚠️ Five configs (`kh_km` Khmer, `la_lo` Lao, `mm_my` Burmese, `vn_vi` Vietnamese, `cn_ug` Uyghur, sourced from Radio Free Asia include only metadata, transcripts, and source URLs. Audio must be downloaded separately from `source_url`.