Validate audio stream timelines and add genuine Google transcript diagnostics

tools/speech-eval/score.py

@@ -6,6 +6,13 @@ from pathlib import Path
 from run import FIXTURE_SHA
 
 
+def require_eligible(report):
+    if report.get("complete") is not True or report.get("quality_score_eligible") is not True:
+        raise ValueError("Cannot score incomplete runs or reduced/invalid competitor pools")
+    if report.get("excluded_windows") or report.get("excluded_window_count", 0):
+        raise ValueError("Cannot score rankings with excluded ASR windows")
+
+
 def score(fixture, rankings):
     queries = [query for query in fixture["queries"] if query["category"] == "video-speech"]
     if len(queries) != 8:
@@ -41,6 +48,7 @@ def main():
     args = parser.parse_args()
     data = args.fixture.read_bytes()
     rankings = json.loads(args.rankings.read_text(encoding="utf-8"))
+    require_eligible(rankings)
     digest = hashlib.sha256(data).hexdigest()
     if digest != FIXTURE_SHA or rankings["fixture_sha256"] != digest or not rankings.get("asr_report_sha256") or not rankings.get("google_worker_sha256"):
         raise ValueError("Missing/mismatched actual ASR/Google provenance")