How to Verify Speech Intelligibility (STI/STIPA)

What is Speech Intelligibility and Why Measure It?

Speech intelligibility is the degree to which speech can be understood by listeners. While this sounds subjective, the Speech Transmission Index (STI) defined by IEC 60268-16 provides an objective, repeatable measurement that correlates closely with listener comprehension tests. STI ranges from 0 (completely unintelligible) to 1 (perfect intelligibility), and is determined by measuring how well the modulation patterns of speech are preserved as sound travels from the talker to the listener through the acoustic environment and any reinforcement system.

Measuring STI is essential for:

• Public address and voice alarm systems: NFPA 72, BS 5839, and EN 54-16 require STI ≥0.50 for emergency voice communication
• Houses of worship: Sermon intelligibility verification
• Classrooms and lecture halls: Learning environment quality per ANSI S12.60
• Courtrooms: Legal requirement for proceedings to be heard by all parties
• Conference rooms: AV system commissioning and certification
• Theaters: Dialogue clarity verification for productions

Step 1: Understand the STI Scale

Before measuring, understand what the numbers mean:

STI Range	Rating	Practical Meaning
0.00 - 0.30	Bad	Almost nothing understood. Major system or room problem.
0.30 - 0.45	Poor	Main ideas grasped but significant loss of detail. Listener fatigue is high.
0.45 - 0.60	Fair	Most content understood with concentration. Minimum acceptable for PA/VA systems.
0.60 - 0.75	Good	Content understood with ease. Target for most reinforcement systems.
0.75 - 1.00	Excellent	Near-perfect understanding. Typically only achieved in well-treated, quiet rooms.

Step 2: Prepare Your Equipment

You need two things: a sound source to play the STIPA test signal through the system under test, and a microphone to capture the received signal. SonaVyx generates the STIPA test signal internally and processes the captured audio using Rust WASM — no external hardware beyond a microphone is required.

Sound Source Setup

Connect a device (laptop, phone, or tablet) running SonaVyx to the sound system's audio input. The STIPA test signal must play through the system being measured — not through a separate loudspeaker. If testing a PA system, connect to the mixer's line input or media player input. Set the system to its normal operating level for speech (the same gain structure used during actual announcements or presentations).

Measurement Microphone

Position a second device running SonaVyx at the measurement position with its microphone exposed. A calibrated measurement microphone provides the most accurate results, but any reasonable quality microphone (including laptop or phone built-in microphones) provides useful comparative data for identifying coverage problems and tuning improvements. If using a calibration file, load it via SonaVyx's mic calibration feature.

Step 3: Measure Background Noise

Before running the STI test, document the ambient noise level. Use the SPL Meter with A-weighting to measure the noise floor for at least 60 seconds with the sound system off but all building systems (HVAC, lighting, other equipment) running normally. Record the LAeq value. This establishes the signal-to-noise context for the STI measurement. STI is strongly affected by ambient noise — a room with NC-40 noise will always score lower than the same room at NC-25, regardless of system quality.

Step 4: Run the STIPA Measurement

Open the STI measurement tool in SonaVyx. Select the measurement mode:

• Quick Mode (15 seconds): Single measurement pass. Suitable for spot checks and coverage surveys where you need to measure many positions quickly.
• Pro Mode (3× averaged): Three consecutive measurements averaged for improved statistical confidence. Use for formal compliance testing and commissioning reports.

Start playback of the STIPA test signal through the sound system. The signal sounds like a continuous, somewhat harsh noise — it consists of seven octave bands (125 Hz to 8 kHz), each modulated at two specific frequencies per IEC 60268-16 Table F.1, for a total of 14 modulation frequencies. Wait for the signal to stabilize (2-3 seconds), then start the measurement capture on the receiving device.

Multi-Position Measurement

For compliance testing, measure at multiple positions. Place the measurement microphone at seated ear height (1.2m for seated audiences, 1.5m for standing) at representative positions:

1. Front row center — typically highest STI (closest to speakers)
2. Mid-audience center — representative of average experience
3. Rear row center — often the worst-case position
4. Rear row off-center — tests coverage at the edges
5. Under balcony (if applicable) — commonly the lowest STI position
6. Balcony front and rear (if applicable)

Record the STI value and position identifier for each measurement. SonaVyx displays the per-octave-band Modulation Transfer Function (MTF), which reveals which frequency bands are limiting the overall STI score.

Step 5: Interpret the Results

Overall STI Score

Compare each position's STI against the applicable standard:

• PAVA/emergency: STI ≥0.50 at every position (NFPA 72, BS 5839, EN 54-16)
• Worship (speech): STI ≥0.50 minimum, ≥0.60 target
• Classroom/lecture: STI ≥0.60 per ANSI S12.60 intent
• Conference room: STI ≥0.60 at all seated positions
• Theater: STI ≥0.55 minimum, ≥0.65 target

Per-Band MTF Analysis

The per-octave-band MTF breakdown is more diagnostically useful than the overall STI score alone. Each band's MTF ranges from 0 to 1:

• Low MTF at 125-250 Hz: Indicates room modes and low-frequency reverberation are masking speech modulations. Treatment: low-frequency absorption, bass management EQ.
• Low MTF at 500-1000 Hz: The most critical bands for speech. Low MTF here usually indicates excessive mid-frequency reverberation. Treatment: broadband absorption panels, reduced RT60.
• Low MTF at 2000-4000 Hz: Often caused by insufficient HF level from the speakers at distance, or HF absorption from seating and audience. Treatment: adjust HF EQ, verify speaker aiming.
• Low MTF at 8000 Hz: Air absorption and speaker distance limitations. Less critical for overall STI but indicates potential HF coverage issues.

Step 6: Improve STI if Below Target

If the measured STI falls below the target, systematic improvement follows this priority order:

Priority 1: Signal Level

The simplest improvement is increasing the direct sound level relative to ambient noise. Check that the PA system level at the worst-case position is at least 15 dB above the ambient noise level. Use the SPL Meter to verify. If the signal-to-noise ratio is below 15 dB, the room noise is the primary limitation — address HVAC noise or increase PA system output.

Priority 2: Coverage

If certain positions show significantly lower STI than others, the issue is coverage uniformity. Use the Transfer Function to compare frequency response at high-STI and low-STI positions. Adjust speaker aiming, add delay fills for distant positions, or add under-balcony speakers for shaded areas.

Priority 3: Reverberation

If STI is uniformly low across all positions and the signal-to-noise ratio is adequate (>15 dB), excessive reverberation is likely the cause. Measure RT60 using the RT60 tool. For speech-critical spaces, RT60 above 1.5s will significantly limit achievable STI. Use the Treatment Calculator to model absorption requirements for reducing RT60 to acceptable levels.

Priority 4: System EQ

After addressing the above factors, fine-tune the system EQ. Reducing energy in frequency bands with the lowest MTF values (typically 250-500 Hz in reverberant spaces) while maintaining 2-4 kHz presence can improve STI by 0.03-0.08 points. Re-measure after EQ changes using the Before/After comparison to verify improvement.

Step 7: Document and Report

Generate a compliance report documenting: measurement positions, ambient noise levels, STI at each position, per-band MTF for the worst-case position, pass/fail verdict per the applicable standard, and any corrective actions taken. For PAVA systems, this report forms part of the mandatory system commissioning documentation required by fire safety authorities.

Common Pitfalls

• Measuring with an empty room: Audience absorption significantly affects RT60 and therefore STI. If possible, measure with the expected audience present, or note that results represent worst-case (empty room) conditions.
• Incorrect signal level: The STIPA signal must play at the system's normal speech operating level. Playing it too loud or too quiet changes the signal-to-noise ratio and produces unrepresentative results.
• Insufficient measurement duration: The STIPA signal requires at least 15 seconds of capture to produce statistically valid modulation indices. Use the Pro mode (3× averaged) for formal compliance testing.
• Ignoring per-band MTF: Two rooms can have the same overall STI score but very different per-band profiles. The MTF breakdown guides targeted improvement — always document it.