Release v1.1: span labels, cohort column, accented and short-turn English rows

Adds a `label` to every silence span and a `cohort` column to every row. Adds 800 `accented` and 1,088 `short_turns` English rows as a second English shard. Rows from the first release are otherwise unchanged.

README.md

@@ -89,26 +89,47 @@ without replacement using seed `20260603`. Languages with fewer than the request
 
 All configs use the same schema. English rows include `language = "en"`.
 
+## Version 1.1
+
+Version 1.1 adds 1,888 English rows in two cohorts and two schema fields.
+
+- `accented`: 800 turns of accented English speech.
+- `short_turns`: 1,088 short replies. 544 rows end at the end of the turn. The
+  other 544 rows stop at a mid-turn pause, so they have no end-of-turn span.
+
+Each silence span now has a `label`, and each row has a `cohort`. The rows from
+the first release are unchanged apart from these two fields. Their cohort is
+`general`.
+
 ## Schema
 
 - `id`: globally unique turn id.
 - `audio`: 16 kHz mono WAV audio.
 - `language`: normalized language code.
 - `duration`: audio duration in seconds.
-- `silence_spans`: candidate silence spans with `start` and `end`.
+- `silence_spans`: candidate silence spans with `start`, `end`, and `label`.
+  The label is `eot` for the true end of the turn and `hold` for a mid-turn
+  pause. A row has at most one `eot` span, and it is always the last span. A
+  `short_turns` row can end in a `hold` span.
 - `words`: word-level transcript timings.
-- `messages`: prior conversation messages.
+- `messages`: prior conversation messages. Rows in the `accented` and
+  `short_turns` cohorts contain only the preceding assistant message.
+- `cohort`: `general`, `accented`, or `short_turns`.
 
-Rows whose final end-of-turn silence span is shorter than 0.2s are excluded
-from this release split, so every turn can be evaluated at a 0.2s score point.
+In the `general` cohort, rows whose final end-of-turn silence span is shorter
+than 0.2s are excluded, so every turn can be evaluated at a 0.2s score point.
 Rows with any silence span longer than 5.0s are also excluded so the benchmark
-focuses on ordinary endpointing pauses.
+focuses on ordinary endpointing pauses. The `accented` cohort keeps longer
+mid-turn pauses; the benchmark harness ignores `hold` spans longer than 5.0s.
 
 ## Languages
 
+The `en` count is 400 `general` rows, 800 `accented` rows, and 1,088
+`short_turns` rows.
+
 | Language | Rows |
 | --- | ---: |
-| `en` | 400 |
+| `en` | 2288 |
 | `ar` | 381 |
 | `de` | 400 |
 | `es` | 400 |
@@ -122,4 +143,4 @@ focuses on ordinary endpointing pauses.
 | `pt` | 400 |
 | `tr` | 400 |
 | `zh` | 344 |
-| `all` | 5454 |
+| `all` | 7342 |

data/ar/validation-00000-of-00001.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ab08f587b30b19fc8f6846572ba131fffb7d481534728edcd3f4654d634eb30c
-size 145145023
+oid sha256:02b573fb86f22810c7a903f9359acbaaab5796b7c41dfacc3fbcdc7b895b5848
+size 122019965

data/de/validation-00000-of-00001.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e02dd1afa056b594810689eac364b632c712c18f1e3d7468fd19d07ec6da5f3d
-size 138163610
+oid sha256:82c6d47d218107023fecc175c92535b39c8f6d69f5d77715e9e6b27c1b2364e9
+size 109307809

data/en/{validation-00000-of-00001.parquet → validation-00000-of-00002.parquet}

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e475d435c8e693912243e8a810bf7e34a59e01e208551ce5b362c9c778b0fdc4
-size 162406142
+oid sha256:52bf7ff71e83d0214d744693839f8a6b14e78e0f26db20dfdb7f589453a50103
+size 109936364

data/en/validation-00001-of-00002.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9737b395ee1ee3479cbbc9ee58490e9526b82412d8e1827b5084c4d876f04a8b
+size 481037805

data/es/validation-00000-of-00001.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:54da6201ad3aa303d4fb2e8416ded3fbfcdaaf090df66e48ee432c4bb75b782f
-size 166059372
+oid sha256:68570a61460318463e144db1ff271cb50ea3976aa2cd77e5850b56c6f26fd146
+size 135728857

data/fr/validation-00000-of-00001.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:daeb94120bbaa7879d8921effb25b434c9f62c687e007ff2708507f131b66157
-size 137615422
+oid sha256:01514d120d29ddb6b44829c48c5e4bb3e89c77345436ad41b1960f5e96001e36
+size 111057041

data/hi/validation-00000-of-00001.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d02181dce0b75d0b203d61d1c519c8c20324c91297039122acfddcc6c235ff1f
-size 132810327
+oid sha256:1011c399d42583ee8e830545ecf351a289d9c3a96bf1720bf40f9fb08c60a386
+size 101431117

data/id/validation-00000-of-00001.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cdc792a99e58c9718c9080c9606b33a1522aa293a0a56a506616fef43c0b4bb3
-size 139854090
+oid sha256:ee06b26835c7b04a4e09a93a2024966b939d2cc25fd460cfcff1e6557f5cf0ae
+size 112275386

data/it/validation-00000-of-00001.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d1bf75b11b7bd265f61db2c7bb28ca018fe3e9519af3467a1588131f8777edd4
-size 133680197
+oid sha256:b61ca0b5f92a809b705d27a9ea2343333411a71adac34caf54d74baebcc8ff0d
+size 108593243

data/ja/validation-00000-of-00001.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:347cc9d555a495099c0d061ef200e18924b3c6d1c019205325eba6a7b418a8f5
-size 138452115
+oid sha256:518a378abeaaab5736c30750f95a87b8d8d453612639baa0f26373aa11d81eae
+size 109809994

data/ko/validation-00000-of-00001.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6526150783f8be2ecaaac5c487db56e806df05250c956f8006b095419ae74127
-size 114700081
+oid sha256:2a083d3b2bca4ccc2c84707f0e1c1078126918c8df54f7d5c2a21e8b1890f002
+size 91532738

data/nl/validation-00000-of-00001.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:86f78d7b7c2a915b87488d3e0c1851638aa3e606a94a7d7198bf098eb24db092
-size 124181863
+oid sha256:b463b06e4437331b65858fce508aa68888e3b4e8168ee0af14e624a7a781c474
+size 100392476

data/pt/validation-00000-of-00001.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ab0446412c366903cecb7f1d095cff7e7e02fe45a73be08136b4f8049472695d
-size 134064610
+oid sha256:17c2eb6b38b0bb245bb641e20d15c5c710d00b9993b0798d14d611eea67fc12b
+size 107761278

data/tr/validation-00000-of-00001.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a09c8587a99fe222dc593ed809ebc1f8e415aa62f832a1fbe3202c04d4609944
-size 96458483
+oid sha256:89c613d553b15a665dc8f4d2f882be59bc87e7c3f03af45549004bdad8d8e0ce
+size 76716195

data/zh/validation-00000-of-00001.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5fd7a30919721832642bea7201ea3d9b09fa6ebce13ae7902ec21ad2dbc9f0f8
-size 129303772
+oid sha256:eeb9193e3ac5516ff250f7353a06babef6feefa31a81a1920f85289c4aea073e
+size 102795747