Burst LoRAs trained on Gemini labels, beside the ones that ship

Same audio, same recipe (rank 16, α 32, 5 epochs, lr 1e-4, base sft3/export), same prompts, same seeds. Only the burst labels and the burst spans differ.

Read the detector column as evidence, not as a verdict. Every burst adapter that ships was trained on rows labelled by this project's own detector, and that detector is also the instrument that scores the clips below. Over 60 DramaBox clips it emitted Shriek zero times, used 8 of its 83 labels and gave 70 % of its detections to two labels; asked blind, Gemini 3.8 Flash put the requested burst in its top-3 on 53/60 against the detector's 3/60. An adapter trained on Gemini's labels may therefore produce a burst the detector cannot name. A null on strict hit rate is ambiguous, not negative — which is why this page exists and why the ear decides.

What was trained

Rows are re-bucketed by the label the row's own rendered script names, not by the bucket the detector originally filed them under: a row whose only Gemini event is an Exhausted Groan would otherwise train an adapter named affirmative_grunt on a cue that contradicts its own name. The floor is 100 rows; below that a rank-16 LoRA on 5 epochs memorises.

classGemini rowsdetector-control rows shipped adapter rowsGemini adaptercontrol adapter
affirmative_grunt132171600yesno
breathy_giggle1312131600yesyes
chuckle6752271600yesyes
deep_breath409201895yesyes
exasperated_sigh4731011600yesyes
exhausted_groan1502151600yesyes
frustrated_groan14432568yesno
heavy_breathing110131357yesno
humming19437348yesno
panting19417765yesno
relief_sigh123371582yesno
scream154200891yesyes
sharp_inhale3922071600yesyes
wistful_sigh1212231600yesyes
yawn1641991600yesyes

Excluded (below the 100-row floor, Gemini count in brackets): soft_hum (93), snicker (85), contented_sigh (82), displeased_grunt (57), surprised_gasp (40), effort_grunt (26), ahem (24), cackle (24), growl (22), mournful_wail (21), sniff (20), cough (19), clears_throat (18), deep_breathing (18), pain_moan (18), childlike_giggle (16), guffaw (16), shriek (14), resonant_hum (14), snort (13), fearful_gasp (13), low_mumble (13), pleasure_moan (11), soft_whistle (11), trembling_whimper (8), sharp_whistle (7), tsk (6), fast_breathing (6), lip_smack (4), coughing (3).

Every difference is paired at the prompt level: the runner seeds on the prompt index alone, so the arms drew identical sampling noise. d hit (strict) is the change in the fraction of prompted cues the detector named as the target class; d hit (family) counts a near miss inside the same burst family as a hit. d wrong and d WER are the guardrails: a hit-rate gain bought with bursts of the wrong kind, or with speech nobody can transcribe, is not a gain.