-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathguide.html
More file actions
128 lines (122 loc) · 6.45 KB
/
Copy pathguide.html
File metadata and controls
128 lines (122 loc) · 6.45 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>qCLS Operator Guide</title>
<style>
body { font-family: 'Segoe UI', Tahoma, Geneva, Verdana, sans-serif; background: #f4f4f9;
margin: 0; padding: 24px; color: #333; line-height: 1.55; }
main { background: #fff; max-width: 780px; margin: 0 auto; padding: 32px 40px;
border-radius: 12px; box-shadow: 0 4px 15px rgba(0,0,0,0.1); }
h1 { margin-top: 0; font-size: 1.6em; }
h2 { margin-top: 1.7em; font-size: 1.15em; border-bottom: 1px solid #e2e2e2; padding-bottom: 5px; }
table { border-collapse: collapse; width: 100%; margin: 10px 0; font-size: 0.93em; }
th, td { border: 1px solid #ddd; padding: 6px 9px; text-align: left; vertical-align: top; }
th { background: #f0f0f5; }
code { background: #f3f3f7; padding: 1px 5px; border-radius: 3px; font-size: 0.92em; }
.warn { background: #fff6f6; border-left: 4px solid #cc3333; padding: 10px 14px; margin: 12px 0; }
.note { background: #f5f8ff; border-left: 4px solid #4a6fbf; padding: 10px 14px; margin: 12px 0; }
ol, ul { padding-left: 22px; }
a.back { display: inline-block; margin-bottom: 16px; color: #007bff; }
</style>
</head>
<body>
<main>
<a class="back" href="index.html">← back to the test</a>
<h1>qCLS Operator Guide</h1>
<p>Categorical loudness scaling in the browser. Everything runs locally; no
participant data leaves the machine.</p>
<h2>1. Before each session</h2>
<ol>
<li><strong>Calibrate.</strong> Click <em>Calibrate</em> for a continuous 1 kHz
tone at 60 dB SPL and set the hardware to match. Click again to stop.
Every level, every contour and the audiogram prior inherit this step.</li>
<li><strong>Set Max Output.</strong> The loudest level that will ever be
presented. Default 100 dB SPL. Lower it for a listener with reduced
tolerance.</li>
<li><strong>Enter the participant details</strong> and stimulus settings.</li>
<li><strong>Optionally enter the audiogram</strong> (see below).</li>
</ol>
<h2>2. Safety</h2>
<p>Nothing is presented above <strong>Max Output</strong>. Whatever is typed is
clamped to the 110 dB SPL equipment limit, and the field locks once a run
starts.</p>
<p>The working ceiling also drops <strong>5 dB automatically every time the
listener responds "Too Loud"</strong>, in both phases, and never rises again
during a run. A final check sits immediately before the level becomes sound.</p>
<div class="note">Stimulus selection targets the boundary it is least certain
about, and the top category sits near the top of the range, so an unbounded run
would drift toward loud levels rather than away from them. The ceiling and the
backoff are what prevent that.</div>
<h2>3. The audiogram (optional)</h2>
<p>Left blank, the procedure behaves exactly as it would without this feature.
Filled in, it starts the listener from a prior matched to their hearing rather
than from a population average, worth about 0.4 dB on average.</p>
<ul>
<li>Nine fields, 250 Hz to 6 kHz. <strong>Leave any frequency blank if it was
not measured</strong>; blanks fall back to the population distribution for
that band rather than to a guess.</li>
<li><strong>Units</strong> must match how the audiogram was recorded, dB HL or
dB SPL.</li>
<li><strong>Transducer</strong> selects the conversion used for dB HL.</li>
</ul>
<div class="warn"><strong>Units and transducer both matter, and neither fails
visibly.</strong> On the same audiogram the prior at 250 Hz differs by about
10 dB between the default table and ER-3A insert earphones. If you are unsure which
applies to your setup, leave the transducer on <em>unspecified (legacy)</em>,
or leave the audiogram blank, rather than guessing.<br><br>
Only tables measured at BTNRH are offered. Published figures for other
transducers were removed after the ER-3A row from that source proved up to
15.5 dB low against BTNRH's own measurements. Measure a transducer before
adding it.</div>
<h2>4. Running the test</h2>
<p>Click <em>Run Test</em>. The listener sees the response scale and the
instructions available from <em>View Instructions</em>.</p>
<table>
<tr><th>Phase 1</th><td>Brackets the listener's range. Descends in 10 dB steps
until "Can't Hear", then ascends from the start level until "Too Loud" or
the ceiling.</td></tr>
<tr><th>Phase 2</th><td>About 100 trials. Each one is chosen where the current
estimate is least certain, so the listener will sometimes hear very similar
sounds in succession. This is expected, and the instructions tell them it
is fine to repeat an answer.</td></tr>
</table>
<h2>5. Reading the result</h2>
<p>The plot shows <strong>equal-loudness contours</strong>: one curve per
category boundary, giving the level that elicits CU 5, 10, … 50 against
frequency.</p>
<ul>
<li>The <strong>shaded band</strong> around each curve is ±1 standard
deviation. A wide band means that boundary was not pinned down; treat it
with caution rather than reading the curve alone.</li>
<li>If an audiogram was entered it is drawn as a <strong>dashed line</strong>,
converted to dB SPL. It should sit near the lowest contour.</li>
</ul>
<table>
<tr><th>Loudness Growth</th><td>CU per dB between the CU 5 and CU 50 contours,
averaged over bands. Higher means a narrower dynamic range.</td></tr>
<tr><th>False-Alarm Rate</th><td>Responses above "Can't Hear" to sounds well
below the listener's threshold, with the number of such trials in
brackets. A screening indicator of attention, not a precise rate. Shown as
<code>--</code> when too few trials qualify.</td></tr>
<tr><th>Posterior SD</th><td>Median uncertainty across all boundaries. The
single best summary of how much to trust the run.</td></tr>
</table>
<h2>6. Saving the results</h2>
<p>Two downloads appear when the test completes.</p>
<table>
<tr><th>Download trials (TBT)</th><td>The raw record: phase, trial, frequency,
level, response.</td></tr>
<tr><th>Download contours (ELC)</th><td>The result: band, frequency, CU, level
and SD, 90 rows.</td></tr>
</table>
<p>Completed sessions also appear under <em>Previous Sessions</em> and can be
reloaded.</p>
<div class="warn"><strong>Session storage is not a record system.</strong> It
lives in this browser on this machine, holds the 50 most recent sessions, and is
lost if the browser data is cleared. Download the CSV files for anything you
need to keep.</div>
</main>
</body>
</html>