9.4 Speech Mapping Verification & Fitting Tolerances

9.4 Speech Mapping Verification & Fitting Tolerances

Quick Answer: Live Speech Mapping is the modern clinical realization of probe-microphone verification, displaying real-ear output in dB SPL on an SPLogram using calibrated digital speech stimuli like the International Speech Test Signal (ISTS). The verification protocol requires testing at three speech input levels: Soft Speech (50–55 dB SPL) to verify audibility above threshold, Medium Conversational Speech (65 dB SPL) to match prescriptive targets (NAL-NL2 or DSL v5) within $\pm 5\text{ dB}$ from 500 to 4000 Hz, and Loud Speech (75–80 dB SPL) to confirm comfort. Finally, Maximum Power Output (MPO) is verified with 85–90 dB pure tones to guarantee output never exceeds the patient's Loudness Discomfort Level (LDL).

Traditional hearing aid verification relied on static pure-tone sweeps or insertion gain curves (REIG). However, modern digital hearing aids utilize non-linear, multi-channel wide dynamic range compression (WDRC), digital noise reduction, transient suppression, and feedback cancellation algorithms. Static artificial pure tones trigger hearing aid noise reduction circuits, driving the amplifiers into attenuation and invalidating measurements. Live Speech Mapping overcomes this limitation by using calibrated real speech signals, displaying real-time acoustic output directly within the patient's individual dynamic range.

+-----------------------------------------------------------------------------+
|                   The Multi-Level Speech Mapping Protocol                   |
+-----------------------------------------------------------------------------+
|  LEVEL 1: Soft Speech (50-55 dB SPL Input)                                  |
|  - Clinical Objective: AUDIBILITY                                           |
|  - Criteria: LTASS and speech peaks must exceed hearing threshold (dB SPL)  |
|                                                                             |
|  LEVEL 2: Conversational Speech (65 dB SPL Input)                           |
|  - Clinical Objective: INTELLIGIBILITY & TARGET MATCHING                    |
|  - Criteria: LTASS must match prescriptive target within ±5 dB (0.5-4 kHz)  |
|                                                                             |
|  LEVEL 3: Loud Speech (75-80 dB SPL Input)                                  |
|  - Clinical Objective: COMFORT & COMPRESSION OPERATION                      |
|  - Criteria: Output must remain comfortably below LDL without distortion    |
|                                                                             |
|  LEVEL 4: MPO / RESR Sweep (85-90 dB SPL Pure Tone Burst)                   |
|  - Clinical Objective: SAFETY & OVER-AMPLIFICATION PREVENTION               |
|  - Criteria: Output curve MUST remain strictly below patient LDL curve      |
+-----------------------------------------------------------------------------+

---

Live Speech Mapping Methodology & Calibrated Speech Stimuli

Speech mapping evaluates how a digital hearing aid processes the complex spectral and temporal modulations of human speech:

+-----------------------------------------------------------------------------+
|             The International Speech Test Signal (ISTS) Profile             |
+-----------------------------------------------------------------------------+
|  - Origin: European Hearing Instrument Manufacturers Association (EHIMA)    |
|  - Source: 6 female speakers speaking 6 languages (EN, AR, ZH, FR, DE, ES)  |
|  - Structure: Phonemic segment remixing; semantically meaningless           |
|  - Acoustic Features: F0 ≈ 200 Hz; Crest factor 12-15 dB; natural syllables |
|  - Function: Prevents DSP noise reduction from attenuating speech stimulus   |
+-----------------------------------------------------------------------------+

1. The International Speech Test Signal (ISTS)

• Developed by an international research consortium under EHIMA, the ISTS is the recognized clinical gold standard test stimulus for real-ear verification.
• It is composed of natural speech recordings from six female speakers reading in six different languages (American English, Arabic, Mandarin Chinese, French, German, and Spanish). The recordings were segmented into short phonemic units ($100\text{ to }500\text{ ms}$) and re-combined into a continuous audio stream.
• Acoustic Fidelity: ISTS possesses all the physical properties of natural human speech—natural fundamental frequency ($F_0 \approx 200\text{ Hz}$), authentic speech envelope modulations, typical syllabic cadence ($~4\text{ syllables/sec}$), and a standardized crest factor of $12\text{ to }15\text{ dB}$ (the difference between peak and RMS amplitude).
• Non-Semantic Design: Because the segments are spliced across six languages, the signal carries zero semantic meaning. This prevents patient cognitive distraction while completely preventing digital hearing aid algorithms from misidentifying speech as continuous ambient noise.

2. The Long-Term Average Speech Spectrum (LTASS)

When calibrated speech (such as the ISTS or standardized speech passages like the "Rainbow Passage") is presented over a 15-to-30-second measurement window, the probe microphone measures the Long-Term Average Speech Spectrum (LTASS):

• Spectral Roll-Off: The LTASS reflects the modal acoustic energy of speech across frequency. Human speech energy is heavily concentrated in the low frequencies (vowel fundamentals around $250\text{ to }500\text{ Hz}$) and rolls off at approximately $-6\text{ dB per octave}$ above $500\text{ Hz}$.
• The 30 dB Speech Dynamic Range: Human speech is not a flat line; it is a dynamic envelope fluctuating continuously over a $30\text{ dB}$ acoustic window:
  • Speech Peaks ($+12\text{ dB}$ above LTASS): High-energy vowel formants and consonant bursts.
  • Speech Valleys / Troughs ($-18\text{ dB}$ below LTASS): Weak acoustic cues, such as unvoiced fricatives (/th/, /s/, /f/) and acoustic transitions.

---

The SPLogram Display & Auditory Dynamic Range Mapping

Traditional audiograms plot hearing loss upside down in dB Hearing Level (dB HL) relative to average young adult audiometric zero. In contrast, Live Speech Mapping utilizes the SPLogram—a Cartesian coordinate graph plotting absolute acoustic energy in dB Sound Pressure Level (dB SPL) across frequency ($250\text{ to }8000\text{ Hz}$):

                             THE SPLOGRAM DISPLAY
   dB SPL
     ^
 130 | ============================================= [ LDL / UCL Curve ]
 120 | 
 110 | --------------------------------------------- [ RESR / MPO Max Output ]
 100 | 
  90 |               +12 dB Speech Peaks
  80 |              /~~~~~~~~~~~~~~~~~~~\
  70 |             (===== LTASS 65 =====)            [ Prescriptive Targets +++ ]
  60 |              \___________________/
  50 |               -18 dB Speech Valleys
  40 | 
  30 | ............................................. [ Patient Hearing Threshold ]
  20 |                                                (Converted from dB HL to SPL)
   0 +------------------------------------------------------------------------>
      250       500       1000       2000       4000       8000 Hz

Converting dB HL to dB SPL: The Role of the RECD

To display an individual patient's hearing loss on an SPLogram, their behavioral audiometric thresholds (dB HL) must be converted to ear canal sound pressure (dB SPL). This conversion requires adding the Reference Equivalent Threshold Sound Pressure Level (RETSPL) and accounting for individual canal acoustic impedance:

$\text{Threshold (dB SPL)} = \text{Threshold (dB HL)} + \text{Transducer RETSPL} + \text{RECD}$

• Real-Ear-to-Coupler Difference (RECD): The mathematical difference in dB across frequency between the SPL generated in an individual's real ear canal versus the SPL generated in a standardized 2cc coupler for the identical transducer signal. Measuring an individual RECD provides exact personalized threshold mapping on the SPLogram.

Visualizing the Patient's Dynamic Range

The SPLogram displays four vital acoustic boundaries simultaneously:

1. Bottom Boundary (Auditory Threshold): The minimum sound pressure required for the patient to detect sound. Any acoustic energy falling below this line is completely inaudible.
2. Top Boundary (Loudness Discomfort Level / LDL): The upper boundary of acoustic tolerance. Any sound exceeding this curve causes pain, discomfort, or acoustic injury.
3. Auditory Dynamic Range: The acoustic space between threshold and LDL. In patients with sensorineural hearing loss, outer hair cell damage elevates thresholds while leaving LDLs unchanged (or reduced by recruitment), severely constricting the dynamic range (e.g., from $100\text{ dB}$ in normal ears down to $30\text{ to }40\text{ dB}$).
4. Amplified Speech Envelope: The real-time LTASS and speech envelope ($+12\text{ dB}$ peaks to $-18\text{ dB}$ valleys) displayed within the dynamic range.

---

Multi-Level Verification Protocol

To ensure that wide dynamic range compression (WDRC) amplifiers are functioning correctly, speech mapping must be verified at three discrete input levels, followed by an MPO saturation sweep:

1. Soft Speech Verification ($50\text{ to }55\text{ dB SPL}$ Input)

• Stimulus: Calibrated speech delivered at $50\text{ to }55\text{ dB SPL}$ (representing soft speech, distant talkers, or quiet conversation).
• Clinical Objective: AUDIBILITY.
• Verification Criteria: The clinician must verify that the LTASS, and particularly the $+12\text{ dB}$ speech peaks, fall consistently above the patient's hearing threshold curve across all frequencies from $500\text{ to }4000\text{ Hz}$. If soft speech cues fall below threshold, weak consonant sounds will be inaudible, devastating speech recognition.

2. Medium Conversational Speech Verification ($65\text{ dB SPL}$ Input)

• Stimulus: Calibrated speech presented at $65\text{ dB SPL}$ (representing normal, average conversational speech at 1 meter).
• Clinical Objective: SPEECH INTELLIGIBILITY & TARGET MATCHING.
• Verification Criteria: The amplified LTASS curve must match the prescriptive target curve (NAL-NL2 or DSL v5) within $\pm 5\text{ dB}$ across the key speech bandwidth ($500\text{ Hz}$ to $4000\text{ Hz}$). This ensures that the acoustic spectrum of conversation is delivered at the exact sensation levels required for optimal phonemic identification.

3. Loud Speech Verification ($75\text{ to }80\text{ dB SPL}$ Input)

• Stimulus: Calibrated speech presented at $75\text{ to }80\text{ dB SPL}$ (representing loud conversation, shouting, or noisy restaurant environments).
• Clinical Objective: COMFORT & COMPRESSION RATIO VERIFICATION.
• Verification Criteria: The amplified response must remain comfortably loud without approaching the LDL curve. This verifies that the hearing instrument's compression circuits are actively compressing high-level inputs, preventing loudness discomfort.

4. Maximum Power Output (MPO / RESR) Verification ($85\text{ to }90\text{ dB SPL}$ Sweep)

• Stimulus: High-intensity swept pure tones or tone bursts delivered at $85\text{ to }90\text{ dB SPL}$.
• Clinical Objective: ABSOLUTE SAFETY & OVER-AMPLIFICATION PREVENTION.
• Verification Criteria: The measured RESR curve must not exceed the patient's measured LDL curve at any frequency. A safety cushion of at least $3\text{ to }5\text{ dB}$ below LDL is recommended.

---

Target Matching Tolerances & Prescriptive Rationales

Prescriptive fitting formulas calculate the precise frequency-specific gain required based on the patient's audiometric configuration. Two major rationales dominate modern clinical practice:

+-----------------------------------------------------------------------------+
|                     NAL-NL2 vs. DSL v5 Prescriptions                        |
+-----------------------------------------------------------------------------+
|  NAL-NL2 (National Acoustic Laboratories - Non-Linear 2):                   |
|  - Population: Primary choice for ADULTS                                    |
|  - Philosophy: Maximize speech intelligibility while maintaining normal      |
|    overall perceived loudness; applies loudness equalization                |
|                                                                             |
|  DSL v5 (Desired Sensation Level version 5):                                |
|  - Population: Primary choice for PEDIATRICS / CONGENITAL LOSS              |
|  - Philosophy: Maximize audibility across the entire speech bandwidth;      |
|    applies loudness normalization; prescribes higher gain than NAL-NL2      |
+-----------------------------------------------------------------------------+

Clinical Gold Standard Fitting Tolerances

Professional guidelines established by the American Academy of Audiology (AAA), the International Hearing Society (IHS), and the Academy of Doctors of Audiology (ADA) define exact numerical tolerances for matching prescriptive targets during speech mapping:

Exam Alert: The board exam frequently questions the permissible tolerance for matching prescriptive targets. The universal standard of care is matching target within $\pm 5\text{ dB}$ at key octave and inter-octave frequencies ($500$, $1000$, $2000$, $3000$, and $4000\text{ Hz}$). Deviations greater than $\pm 5\text{ dB}$ in this core speech zone result in clinically demonstrable reductions in speech intelligibility (if under-amplified) or acoustic rejection and fatigue (if over-amplified).

---

The Speech Intelligibility Index (SII, ANSI S3.5)

The Speech Intelligibility Index (SII), standardized under ANSI S3.5, is an objective numerical metric that calculates the proportion of acoustically usable speech cues available to the listener:

+-----------------------------------------------------------------------------+
|                 Speech Intelligibility Index (SII) Mechanics                |
+-----------------------------------------------------------------------------+
|  - Scale: 0.0 (Zero audibility = 0%) to 1.0 (Complete audibility = 100%)    |
|  - Calculation: Divides speech into frequency bands; calculates audibility  |
|    of 30 dB speech dynamic range in each band; multiplies by weight         |
|  - Band Importance: 1500 to 3000 Hz carries highest weighting for speech    |
|  - Clinical Role: Quantifies aided improvement (e.g., Unaided 0.25 -> Aided 0.72) |
+-----------------------------------------------------------------------------+

1. Calculation Principles

1. The speech spectrum is divided into discrete frequency bands (typically 1/3-octave bands from $160\text{ to }8000\text{ Hz}$).
2. For each band, the system calculates the proportion of the $30\text{ dB}$ speech dynamic envelope that falls above the patient's hearing threshold (expressed as an audibility factor from $0.0$ to $1.0$).
3. Each band is multiplied by an "importance weight" reflecting that frequency's relative contribution to human speech understanding:
   • Low frequencies ($250\text{–}500\text{ Hz}$) carry lower importance weights (~$0.05$ to $0.10$).
   • High-mid frequencies ($1500\text{ to }3000\text{ Hz}$) carry the highest importance weights (~$0.30$ to $0.40$), as they contain the primary spectral cues for distinguishing consonants.
4. The weighted products of all bands are summed, yielding a single decimal score from $0.0$ to $1.0$ (or $0%$ to $100%$).

2. Clinical Application of the SII

• Objective Outcome Measurement: Comparing the unaided SII to the aided SII provides immediate, objective proof of benefit. For example, a patient with a moderate ski-slope loss may exhibit an unaided SII of $0.28$ (only 28% of speech cues audible); with verified target matching, the aided SII jumps to $0.74$ (74% audibility).
• Patient Counseling Tool: Patients and family members often struggle to interpret audiograms or decibel scales. Explaining that the hearing aids increase accessible speech information from 28% to 74% provides an intuitive, empowering visualization of why amplification is effective.

---

Clinical Troubleshooting of Real-Ear Disparities

When verified real-ear curves fail to match prescriptive targets or cause patient complaints, the specialist must apply algorithmic adjustments in the fitting software:

+-----------------------------------------------------------------------------+
|                    Troubleshooting Real-Ear Disparities                     |
+-----------------------------------------------------------------------------+
|  PROBLEM: Under-amplification of Soft Speech (<55 dB) in High Frequencies   |
|  -> Action: Lower Compression Threshold (TK) / Increase Soft (G50) Gain    |
|                                                                             |
|  PROBLEM: High-Frequency Disparity (>5 dB below target at 3-4 kHz)         |
|  -> Action: Rule out standing wave artifact; increase channel gain;         |
|             consider Libby horn or acoustic tubing change                   |
|                                                                             |
|  PROBLEM: Upward Spread of Masking / Boomy Sound Quality                   |
|  -> Action: Excessive LF gain masks HF consonants; reduce 250-500 Hz gain;  |
|             enlarge vent diameter to allow low frequencies to escape        |
|                                                                             |
|  PROBLEM: RESR / MPO Curve Exceeds LDL Curve at 2000-3000 Hz                |
|  -> Action: Decrease MPO / Output Limiting in specific channel;             |
|             DO NOT lower soft/medium gain; prevents acoustic shock          |
+-----------------------------------------------------------------------------+

1. Correcting Soft Speech Under-Amplification

If soft speech ($50\text{–}55\text{ dB SPL}$) is below target while conversational speech is on target, do not increase overall volume. Increasing overall gain will make medium and loud sounds excessively loud. Instead, access the software's non-linear compression controls:

• Increase Soft Gain (G50 / Low-Level Gain) in the affected frequency channels.
• Lower the Compression Knee-Point / Threshold (TK): Lowering TK causes the compressor to engage at softer acoustic levels, amplifying whisper-level sounds without increasing maximum output.

2. Resolving Upward Spread of Masking

If low-frequency gain ($250\text{ to }500\text{ Hz}$) is over-amplified, high-energy low-frequency vowel energy physically drives the basilar membrane toward the cochlear base, acoustically drowning out and masking softer, higher-frequency consonant cues. The patient reports that voices sound "loud, muffled, boomy, but completely unclear."

• Correction: Reduce low-frequency gain ($250\text{–}500\text{ Hz}$) by $3\text{ to }6\text{ dB}$ or enlarge the physical vent diameter to vent excess low-frequency acoustic energy.

3. Managing Loud Sound Discomfort

If the patient reports that clattering dishes or running water sound painfully sharp, check the RESR / MPO curve:

• If the RESR exceeds the patient's LDL curve at $2000\text{ to }4000\text{ Hz}$, decrease the MPO (Output Compression Limiting) in those specific channels.
• Do not reduce overall gain or soft gain, as this destroys conversational speech intelligibility.