Audiometry

The Audiometry element simulates a pure-tone hearing test. You design a patient with a particular hearing loss — by ear, by frequency, by air and bone conduction — and the learner sits at a realistic two-channel audiometer to find that patient's thresholds. They present tones, decide when masking is needed, run the masking procedure, and plot the audiogram, exactly as they would in a real test booth. When they're done, their plotted audiogram is compared against the one you designed.

Like every element, audiometry follows the same lifecycle — Design → Settings → Testing → Scoring → Report. The audiogram is also the foundation other tests build on: the speech elements (SRT, WRS, and Sound Sensitivity) link to an audiogram and derive their patient behaviour from it.

Design

When you add an Audiometry element to a lesson, you create the patient's true hearing — the thresholds the learner is supposed to discover.

Template vs. fixed

Every audiometry element is either a template or fixed, toggled from the element list in the designer:

• Template — you describe each ear at a high level (a degree range, a type, and a configuration) and Theta generates a fresh, plausible audiogram every time a learner starts the lesson. Two learners get two different patients that both match your description. You can't hand-edit individual numbers in template mode.
• Fixed — the exact thresholds you set are the patient, every time. Use this when a specific case matters (a textbook example, a particular masking scenario).

Describing each ear (the profile)

For both the Right and Left ear you set:

Use Shuffle to draw new thresholds from these profiles. It also draws a new speech patient for every linked speech test. On a fixed case with hand-edited thresholds, or speech values typed on a linked test, it asks first, since it replaces them. In template mode the profile is the design; in fixed mode it's a fast way to lay down a starting pattern you can then refine by hand.

Changing any profile field regenerates the thresholds from it, and draws a new speech patient the same way. On a fixed case where you have already edited thresholds or typed speech values by hand, Theta asks first — cancel to keep your numbers and change only the profile label.

Editing exact values (fixed only)

In fixed mode, open Manually Adjust Values to edit the underlying numbers directly:

• Thresholds — the patient's true (masked) thresholds per frequency, for R/L × AC/BC. You edit these; the unmasked shadow is derived automatically from the true thresholds and the interaural-attenuation rules, so there's no unmasked row to hand-enter. Below them, the Required points checkboxes control which points the learner is expected to obtain. They are the same grid as Required tests & frequencies on the Settings tab: untick a point here and it is unticked there too, straight away, and the other way round. Points a Settings rule decides show greyed out with the reason — a half-octave added between two thresholds the interoctave threshold apart, or the ear that doesn't need bone when bone is allowed in one ear only — and change through that setting. With hearing screening on, the screened frequencies are the points, so the whole table is locked.
• Comfort levels — MCL and UCL per air-conduction frequency. Every patient has them, drawn from the air thresholds whenever new thresholds are drawn (so a template's vary with each attempt), and they follow a threshold you edit: one that now sits at or above its comfort levels gets fresh ones. Adjust any of them by hand. They are plotted and scored only when MCL/UCL testing is on in Settings; otherwise they are what the patient answers if a learner tries the test anyway. A linked Hearing Aid takes the patient's UCL from here only when UCL testing is on, and estimates it from the loss otherwise. A fixed case's comfort levels are exactly what you saved, on every attempt and under any element preset. An older case with gaps shows a Fill in comfort levels button; until you use it, a gap stays a gap in the learner's attempt rather than being drawn afresh for each one.
• Masking variables — the reference masking levels (ISL, IML, plateau start/end, saved masking level) and a read-only Masking required grid: whether each point needs masking, from the thresholds and interaural attenuation.

The plot at the top of the editor previews exactly what you've built. The Appearance menu lets you combine the two ears onto one plot, show the PTA, show the audiometer's output limits, overlay the degree-of-loss bands, and overlay either count the dots or the familiar speech sounds (one at a time). None of these is saved with the lesson; they only change your preview.

Functional (non-organic) cases and the Stenger test

To build a patient who exaggerates or feigns a loss, tick Non-organic (functional) assessment on the Settings tab, under Required Tests & Frequencies. The audiogram on the Design tab (profile, plot, and thresholds) is then the patient's true hearing, and a Functional / Stenger Assessment section appears on the Design tab (it is hidden until the option is on), open on a Feigned thresholds grid (R/L × AC/BC) for the thresholds the patient pretends to have. Leave a frequency blank wherever they are honest.

• The patient hears with their true hearing, but only responds once a tone reaches the feigned level in the ear they hear it in. The feigned audiogram is what the learner should record, and the one they're scored against. A preview labelled What the patient will admit appears once you enter a feigned value.
• Nobody can feign hearing better than they have, so a feigned value at or below the true threshold has no effect.
• No shadow curve. A tone sent to a feigned ear with normal hearing underneath always sounds louder in that ear than the copy that crosses over to the other ear, so the patient keeps quiet until the feigned level. An unmasked threshold in the poorer ear that is worse than cross-hearing allows is a classic sign of a non-organic loss. Bone conduction shows it too: the patient holds back on the side the oscillator is on, so the unmasked bone line can come out far worse than the other ear's.
• If the true loss is itself past the cross-hearing point (a real loss the patient exaggerates), the shadow appears as it would for anyone. Masking then takes the threshold straight up to the feigned level. The answer key's masking plateau follows the same rules.
• When the learner links the two channels and presents a tone to both ears at once (the Stenger test), the patient hears it in the ear where it's louder relative to their true hearing. They stay silent if that tone is below the level they're admitting in that ear, which gives a positive Stenger. If the tone is equally loud in both ears, the patient hears it in the middle of the head, including their honest ear, so they respond.
• An ear counts as a positive Stenger when at least one of its air-conduction values is feigned 15 dB or more above the true threshold. The grid highlights smaller values in amber.
• On a template, the category generates the true hearing for each learner and the feigned values stay as you typed them. Keep them well above anything the category can generate, or some learners get an ear that is honest or too weakly feigned to score as a positive Stenger.
• A functional case never scores or reports the degree / type / configuration classification. The profile you set describes the true hearing, while the learner would classify the loss the patient admits, so the profile row on the Scoring tab is locked off and the learner is not asked for one.
• Linked elements. An ABR linked to this audiogram follows the true hearing, since a patient can't fake an ABR. Speech follows the feigned loss. For speech that's better than the pure tones predict, choose PTA higher for the feigned ear (or type the SRT on a fixed case).

Auto-fill (giving the learner a head start)

By default the learner starts with a blank audiogram. In Settings you can instead auto-fill part of it:

• Off — blank; the learner finds everything.
• Unmasked only — the unmasked thresholds are pre-plotted; the learner still has to perform masking to establish the final thresholds.
• Thresholds, classify yourself — every threshold is provided but the classification is left blank. Only the Profile (degree, type, configuration) is scored — this is an interpretation exercise, and what the Audiogram classification batch uses.
• Complete audiogram — the entire audiogram, classification included, is provided. Nothing on the audiogram is scored — useful when the audiogram is just context for a linked task like speech testing.

With Unmasked only or Complete audiogram, a banner above the audiometer tells the learner what was provided. Thresholds, classify yourself shows no banner: the learner classifies each ear through the Audiogram Profile badge in the top-left corner.

A Skill Assessment exam or practice case can't auto-fill: it marks a threshold with no testing behind it wrong, and a provided threshold has none. On such a case the options other than Off are locked, with the reason, and a save that auto-fills is refused.

Settings

The Settings tab (the gear on the element) controls how the test behaves. These are pre-filled from the course's element preset and then owned by this element, so each case is self-contained.

The Classification & Symbols section of the Settings tab lets you tune the degree-of-loss bands and the audiogram symbol set for this case. These determine the degree names in the profile dropdowns and the symbols drawn on the plot and the report. (Configuration names are fixed and not editable.)

Symbols are set as two grids — one for thresholds a response was obtained at and one for no-response thresholds — so an NR point plots the symbol you chose for that exact ear, transducer, and masking state rather than a marked-up version of the response symbol. See element presets for how the two grids stay in step.

SimuHear and CARL mannequin lessons

Tick SimuHear under Console Options to take away the response light. The learner hears a tone only when the simulated patient hears it, so the tone itself is the response and no tone means no response. The patient still responds exactly as they would with the light (same thresholds, cross-hearing, masking and variability), and scoring is unaffected.

What the learner notices:

• The response light above the console is replaced by a short note saying the lesson uses SimuHear.
• Channel 1's Power button is off, because a continuous tone has no response behind it. Tones are presented with Present or the spacebar.
• Play the tone in the console's settings menu is locked on, and the tone volume can be turned down but not off.
• The masker still plays. The patient hears it too.
• The Masking Plateau panel is not offered: it plots every tone the patient heard, which would put the response back on screen.
• In MCL / UCL mode the patient gives no spoken reply, as in Theta 3. The learner judges loudness without it, so avoid scoring MCL/UCL in a SimuHear or CARL lesson.

Tick CARL mannequin for a lesson run on a CARL training mannequin. It turns SimuHear on (and locks it), then:

• Silences the masker: all of Channel 2, plus narrow-band noise on Channel 1. CARL reacts to any sound in its test ear, so an audible masker would look like a response. The masker still masks in the simulation; only its sound is muted.
• Fixes the tone volume at the level CARL listens for (the same output Theta 3 used), and locks the monitor sliders.

The learner watches CARL rather than the screen for responses. The lesson fixes the audiometer's output, not the computer's, so set the computer's volume once when you set up CARL and leave it there.

An element preset can make either setting the default for new audiograms. The attempt replay and the report both note when a lesson used SimuHear or CARL.

Hearing screening mode

Tick Screen instead of finding thresholds to run the case as a hearing screening. Choose which frequencies to screen and the screening level (dB HL) for each; the default protocol is 500 Hz at 25 dB HL and 1000, 2000 and 4000 Hz at 20 dB HL. (An organization's element preset can seed a different default protocol, but the preset editor does not expose it yet — ask us to set one.)

What changes when screening is on:

• The screened frequencies replace the required frequencies — air conduction only, no bone conduction — so the Required Tests grid greys out (it returns intact when you turn screening off), and the console and audiogram cover just the screened range.
• The learner gets a Hearing Screening panel under the audiogram with a Pass / Refer button pair for every screened frequency, per ear.
• The audiometer is not restricted. Learners have to dial to the screening level themselves — the point of the exercise — and the attempt replay and the Screened at the screening level analytic show whether they did.
• Turning screening on switches the element's scoring to the Pass/Refer call by default and turns final-threshold scoring off (turning it off again hands scoring back to the thresholds). Threshold scoring still works at the screened frequencies, so re-enable it on the Scoring & Report tab if a case should ask for the call, the thresholds, or both.

The expected call is derived from your designed thresholds at scoring time: a frequency is Pass when the final air-conduction threshold is at or below the screening level, and Refer when it is above it or is a no-response. Templates work as usual — pick a profile range that spans normal hearing and a loss, and each submission gets a fresh mix of passes and refers.

Batches — many patients, dealt together

Click Add a batch… under the element palette to drop a whole set of generated cases into a lesson at once. Two batch recipes ship for audiometry:

• Hearing screening — N patients in screening mode. You set three things: how many patients, the refer rate (the dialog shows what that deals — exactly 2 refer · 8 pass at 20% of 10), and the protocol: tap each frequency to screen and set its level underneath. What a refer looks like is fixed: one or both ears, mild to moderately severe, flat, sloping or notched. Passes are normal-hearing ears. The generator checks every patient it draws against the protocol and re-rolls a miss (nudging the loss worse each time) — so with the standard protocol a refer always refers at ≥1 screened frequency and a pass always passes at all of them. The check is bounded (eight tries), so an unusual protocol that a shape can never fail — screening only 500 Hz against a notched loss, say — ships the case as it is.
• Audiogram classification — N completed audiograms drawn from a realistic caseload (symmetric sensorineural, asymmetric, conductive, mixed). The mix is fixed — about 60% symmetric sensorineural, the rest spread across the other shapes — so ten audiograms are six symmetric and one of each of the others. The learner classifies the degree, type and configuration of each ear.

Batches aren't only audiograms — the same button builds a batch of tympanograms, OAE screenings or otoscopy images.

The batch is just N ordinary templated elements plus one rule stored on the lesson: deal the mix across these elements every attempt. So the patients appear in the lesson list like any other element — remove one, reorder them — and a Batch panel above the list shows the recipe, the current patient count and the mix, with Add … more for extra patients; a screening batch's refer rate and protocol stay editable there. Things to know:

• A batch holds up to 20. Twenty patients, audiograms or ears — a tympanogram batch of 20 patients is forty tympanograms. The dialog and Add … more stop there, and a save that would take a batch past 20 is refused. A batch saved with more before the limit existed still saves and plays as it is; it just can't grow. Every learner gets every patient you built; only the Practice decks let the learner pick how many.

• A batch lesson holds only its batch and Information elements. A batch is a self-contained exercise, so nothing else can sit among its patients: with a batch in the lesson, the palette offers Information alone, and Add a batch… stays locked while the lesson holds any other case element. Saving a lesson that mixes them is refused, naming what to remove.

• The mix is dealt exactly. Every attempt deals the stated share in a fresh shuffled order — never an independent draw per patient, which would let ten audiograms come out eight or nine alike. (A learner who works out that exactly two of ten refer could, in principle, stop looking after finding them.) When the shares don't divide evenly — seven tympanogram types over twenty ears — which ones get the extra seat is drawn fresh too, never the first ones in the list.

• Every patient is generated. Batch patients have no Fixed switch: each attempt draws all of them fresh.

In the submission the tabs read Patient 3 of 10, and the results screen and PDF open with a batch summary: overall accuracy, sensitivity (refers caught), specificity (passes left alone), the number of unanswered frequencies, and one line per patient showing the expected and recorded call per ear. Unanswered frequencies still count against the score, but are left out of sensitivity and specificity rather than being booked as a miss or a false alarm.

Testing

During the submission the learner works at a simulated audiometer. The audiogram is on top; the console with two channels sits below it. (For the full lesson context, see the submission flow.) On a shorter desktop screen — a TV, or a monitor with large display scaling — the console's controls get smaller, so the console keeps to about a quarter of the height and the audiogram keeps the rest.

The two channels

• Channel 1 is the test channel — the tone presented to the patient.
• Channel 2 is the masker — narrow-band noise sent to the non-test ear.

For each channel the learner sets the ear (R/L), the transducer, the stimulus type (pure, pulsed, warble, pulsed-warble, or narrow-band noise), and the level in dB HL with the +/- buttons (5 dB steps). The transducer and stimulus buttons show the current choice in short (Inserts, Pulsed); clicking one opens the full list. Levels are clamped to the audiometer's output ceiling for that frequency and transducer. A Power toggle holds the tone on continuously; the Present button delivers a tone while held (and momentarily interrupts a powered-on tone). In a SimuHear lesson, Channel 1's Power is off.

If the case uses the simplified console, Channel 2 shrinks to just a masking level and an On/Off button, and automatically tracks the non-test ear.

On the full console Channel 2 can be routed to either ear, and the patient responds to where the noise actually is. Noise in the non-test ear masks the cross-heard shadow, occludes that ear during bone testing, and can cross back onto the test ear (overmasking). Noise routed into the test ear — a real clinical error — does none of that: it masks the test cochlea directly (a tone has to reach the masker's level to be heard), occludes the test ear during bone testing, and leaves the opposite ear's shadow standing until the noise's own crossover reaches it. The Masking Tool panel redraws for this and labels "Masker is in the test ear". Because that noise masks nothing in the non-test ear, the console records such a save as an unmasked measurement — the effective-masking row cannot credit it, and where the design required masking it grades as "did not mask".

Presenting tones and reading responses

The learner picks a frequency from the selector (limited to the tested range) and presents tones up and down to bracket the threshold. When the simulated patient hears a tone, a response indicator flashes above the console. (In a SimuHear lesson there is no indicator: the tone sounds only when the patient hears it.) The response reflects the designed thresholds, cross-hearing (interaural attenuation and occlusion effect), and any masking in play — so under-masking, over-masking, and masking dilemmas all behave as they should.

You can drive everything from the keyboard (arrow keys to change frequency and level, spacebar to present), or with the mouse — see Keyboard shortcuts for the full list.

Recording a threshold

With a value bracketed, the learner presses Save to plot the point at the current frequency, ear, and conduction. If Channel 2 was on, the point is saved as masked and Theta records the masking levels the learner used (initial stimulus and masking levels, plateau start/end, and the saved masker level) for scoring. The save mode can be switched between Threshold, MCL, and UCL — either from the badge on the audiogram or in the console's settings — and the settings menu also offers No response, Vibrotactile, and Masking dilemma flags. Clear point removes the current point — Channel 1's ear and transducer at the selected frequency, never Channel 2's. In MCL or UCL mode it clears that comfort level and leaves the threshold; otherwise it clears the threshold together with everything saved with it (whether it was masked, the masking levels, and the No response / Vibrotactile / Masking dilemma notes), so a cleared point is graded like one never tested; Clear all asks for confirmation, then returns the audiogram to how it started — which on an auto-filled case means the provided thresholds come back rather than being wiped.

Comfort-level testing

In MCL or UCL mode, presenting a tone produces a verbal patient response ("That's comfortable", "Too loud", etc.) shown as a speech bubble (not in a SimuHear or CARL lesson, which has no spoken replies). The learner adjusts level until the patient's reaction matches the level they're after, then saves it.

The audiogram view

• A gear menu lets the learner combine the ears, show the true thresholds (a study aid, hidden in exam mode), overlay the degree bands, count the dots or the familiar speech sounds (one or the other — they cover the same part of the chart), and open the masking tool or the masking plateau panel. Only combined or split is remembered in that browser — across elements, sittings and lessons — and once a learner picks it themselves, it wins over the lesson's default view. Everything else starts off each time the element opens, so a peek at the true thresholds while practicing can never carry into an exam.
• If a profile is scored, an Audiogram Profile badge sits in the top-left corner and shows how many profile fields are set out of the total (e.g. 3/8), with a green check once all are set. Tapping it pops the degree, type and configuration for each ear open over the audiogram; a tap outside, Escape or × puts it away.
• If a PTA is scored, a PTA badge sits in the audiogram's bottom-left corner (the test-mode badge is bottom-right). Tapping it opens a box for each ear's air (AC) and bone (BC) PTA — only the ones you score — and nothing else: no PTA type, no frequency list, no thresholds. The learner chooses which frequencies to average. The badge shows a check once every box is filled, and the count while part-way. On phones and tablets both badges sit in a strip just under the audiogram instead of over its corners.

Scoring

Scoring compares the learner's plotted audiogram against your designed one. Open the Scoring & Report tab to choose what counts. (For how scores roll up across the lesson, see Scoring.)

Each field can be set to one of three states:

• Off — not scored, not shown on the report.
• Scored — compared against the design and counted.
• Report-only — shown on the report for context, but not counted.

You set these per ear (R/L) and per conduction (AC/BC). A Sync ears toggle (on by default) mirrors your choice to both ears at once.

Fields you can score, grouped as in the UI:

Two tolerances let you control strictness:

• Threshold tolerance — how many dB a plotted threshold may differ from the design and still count as correct.
• PTA tolerance — the allowed difference on the Pure Tone Average.

When the profile is scored, Configuration: also accept (below the table) lets more than one configuration count as correct. For each configuration, tick the others that should also score when it is the answer — for example, accept Steeply Sloping and Sloping when the case is Notched. It works one way only (that rule does not accept Notched on a Sloping case). Because it is a rule rather than an answer, it also covers batch and daily/weekly cases, whatever configuration each patient is dealt (an ordinary template keeps the configuration you chose; only its thresholds change). The report marks a configuration scored this way OK (alt.).

By default, only the final (masked) thresholds are scored — a sensible starting point you can build on.

Report

The generated PDF renders each ear's audiogram as a clean plot, with the learner's results drawn solidly on top of your designed thresholds shown faded behind them — so it's immediately clear where they matched and where they diverged. Symbols follow the case's symbol set — including its dedicated no-response symbols — AC points are connected, and vibrotactile (VT) and masking-dilemma (MD) situations are marked with a small annotation and a legend.

Alongside the plot, a data table lists every scored and report-only field with three columns — the value you designed, the value the learner recorded, and whether it was scored correct — so an instructor can see at a glance what was right, what was off, and by how much. Fields set to Off don't appear. On a screening case the Screening row lists the expected and recorded pass/refer call at each screened frequency.

If the lesson used SimuHear or a CARL mannequin, a line under the Audiometry heading says so, on the report and on the results screen alike.

Technique flags

Technique flags look at how each threshold was found, read from what the learner actually did on the audiometer. They are set where you choose Practice or Exam — the lesson's settings in the course (the gear on the lesson's row in the course's Build tab) — so the same lesson can be Report only in a practice course and Scored in an exam course. The section appears for lessons with an audiogram.

How a search is read

Theta reads the testing as searches: one continuous attempt at one threshold — the same frequency, ear, earphone and masker state, tone after tone. A search ends when the learner saves a threshold there, tests anything else in between, turns the masker on or off, changes earphones, or comes back in another sitting. The first tone of a search is never judged as a step from the last one, so going back to re-check 1000 Hz, returning to a frequency to mask it, or switching to inserts is never a step error. Only tones actually presented count — turning the dial several clicks between tones is fine.

The flags

A step goes wrong in one way only — going the wrong way is one flag, going the right way by the wrong amount another — and repeating a level is not a mistake. Masked searches follow the plateau procedure, so the Hughson-Westlake step flags don't apply to them; the masking flags do. Screening lessons and audiograms given to the learner complete raise no flags; an audiogram given with its unmasked thresholds asks nothing of the unmasked search (no starting level, ascending or coverage flags), only of the masking, the steps, the tone length and the earphones.

The flags' settings are fixed onto each attempt as it starts: changing them later never changes an attempt already made unless staff Re-score it, which applies the current settings (so flags switched on later can be shown on earlier attempts). A lesson imported from Theta 3 brings its program-flag settings with it (Theta 3's starting-level and +20 dB checks didn't exist, so those two arrive off).

Advanced Analytics

Besides the numbers a learner recorded, Advanced Analytics reads each audiogram attempt's action trail — every tone presented, at what level, with what masking — so it can judge how the testing was done. The Technique & integrity starter analysis, the set for signing off that someone tests correctly, reports these columns:

Threshold confirmed by repeat response — which this starter reported before the ascending criterion — counts a response at the saved level however it was reached, a descending run included. It stays available to add from the column picker.

An organization with the Skill Assessment also has a Skill Assessment starter: these technique and integrity columns, the masking levels and plateau, a transducer switch and coverage of the required frequencies, beside the three skills the exam scores — see Your organization's Skill Assessment.

The instructor's threshold tracing panel shows the same criterion under each ear's search, one line per saved threshold (each re-check is its own attempt at it): Ascending: 2 of 3 climbs heard ✓, or Ascending: 1 of 2 climbs heard ✗. Only the climbs the rule reads are counted. The panel also marks every tone a technique flag was raised at, lists the flags under each ear's search, and the attempt replay shows each flag on the step that raised it.