{ "status": "supplementary_audit_complete", "audit": { "same_session_unambiguous_matches": 291, "cross_reset_matches": [ { "reference_index": 131, "hypothesis_index": 68, "text": "on", "reference_start": 132.91000000000003, "reference_end": 133.42000000000002, "emitted_seconds": 138.75317345303483, "lag_from_word_end_seconds": 5.333173453034817, "lag_from_word_start_seconds": 5.843173453034808, "alignment_unambiguous": true, "possible_hypothesis_indices": [ 68 ], "may_be_unmatched_in_optimal_alignment": false } ], "maximum_seconds": 0.7862039440218496, "p95_seconds": 0.5573388150148162, "over_one_second": 0, "omitted_and_substituted_words_remain_errors": true }, "original_counts": { "hits": 301, "substitutions": 39, "deletions": 138, "insertions": 4, "reference_units": 478, "hypothesis_units": 344, "errors": 181, "error_rate": 0.3786610878661088 }, "original_alignment_timing": { "measured": true, "actual_arrival_paced": true, "timing_basis": "Host receipt relative to post-ready20ms source feeder; complete exposed meeting", "correctly_matched_units": 301, "ambiguous_matched_units": 9, "ambiguity_method": "all minimum unit-cost Levenshtein paths, without timing-based selection", "unambiguous_matched_units": 292, "unambiguous_maximum_seconds": 5.333173453034817, "unmatched_reference_units": 177, "matched_reference_fraction": 0.6297071129707112, "min_seconds": -2.344643686963238, "p50_seconds": 0.43729305997493384, "p95_seconds": 0.5664107729122065, "maximum_seconds": 5.333173453034817, "matched_units_above_one_second": 1, "all_matched_units_within_one_second": null, "whole_release_pass": false, "limitation": "Lag summaries use one deterministic text alignment and are descriptive when alternative optimal alignments exist. Only unambiguous matches can support per-word timing claims. Omissions/substitutions remain errors, never zero-latency successes. Published reference timing uncertainty and overlapping-speaker word order are not independently corrected." }, "production_ready": false, "whole_release_pass": false, "not_a_replacement_score": true, "method": "Textual matches wholly outside the emitting ASR session cannot identify the spoken word for latency. They remain unresolved recognition evidence, never zero-latency successes. Original global score and latency are retained.", "inputs": { "reports/published-q3-cuda-long-v1/run.json": "dfcd3b82a7f8da0601c34ad503927bcbd5635ae1e19053b355b7facf6d280f6d", "reports/published-q3-cuda-long-v1/run.jsonl": "b692e0201e7dfd3ed7b905adde71f717fe1d1d7b1b9f90c8f1060ba8b36b9f7c", "reports/published-q3-cuda-long-v1/run.observed.jsonl": "d33d1e90c7aecf1bd5e037729a499e2efbeeb9bf3574365b3ee9accbd6d6703b", "/Users/Ajab/conductor/workspaces/visualears/experiments/audio8-ternary/scripts/audit_gpu_caption_cross_reset_v1.py": "b70d33a0c21e7e654afe3fa3f10a494506d3006cfd0732280b0d41d59b4d4dc1", "/Users/Ajab/conductor/workspaces/visualears/experiments/audio8-ternary/scripts/verify_mlx_q3_memory_regression_v1.py": "368b63a07853603f4eb8dd7bcaba05f73764d2c35ab220a3eb3f8d6c6e3aca37" } }